102
Consultant Report Project Number: 45206-001 September 2020 Nepal: Water Resources Project Preparatory Facility Flood Forecasting and Early Warning System: Mohana – Khutiya Basin This document is being disclosed to the public in accordance with ADB's Access to Information Policy.

190404 FFWES Mohana Khutiya

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Consultant Report

Project Number: 45206-001 September 2020

Nepal: Water Resources Project Preparatory Facility Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

This document is being disclosed to the public in accordance with ADB's Access to Information Policy.

WRPriRisNeFlooMoh

4 Apr

RPPFioritysk Maepal od Forechana – K

ril 2019

Mini

F: Prey Riveanage

casting aKhutiya B

istry of Energ

Department

eparaer Basemen

and EarBasin

GOVE

gy, Water Re

t of Water Re

ation osins

nt Pro

ly Warn

ERNMENT O

esources and

esources and

of Flood

oject,

ing Syst

OF NEPAL

d Irrigation

d Irrigation

d ,

tem:

1243 124 124 C:\Users\Erik Klaassen\Documents\Work\01 Project\WRPPF - 383877 MM - Nepal\04

Deliverables\11 FFEWS\Mohana-Khutiya\1\190404 FFWES Mohana Khutiya.docx Mott MacDonald

Mott MacDonald 22 Station Road Cambridge CB1 2JD United Kingdom T +44 (0)1223 463500 F +44 (0)1223 461007 mottmac.com

WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

4 April 2019

Mott MacDonald Limited. Registered in England and Wales no. 1243967. Registered office: Mott MacDonald House, 8-10 Sydenham Road, Croydon CR0 2EE, United Kingdom

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Issue and Revision Record

Revision Date Originator Checker Approver Description 0 24/11/18 Iqbal

Hassan Christian Hetmank D. Ocio

Christian Hetmank

1st submission

1 04/04/19 Iqbal Hassan

Peter Ede A Akindiji

Christian Hetmank

Final submission

Document reference: 383877 | REP | 0040 Information class: Standard

This document is issued for the party which commissioned it and for specific purposes connected with the above-captioned project only. It should not be relied upon by any other party or used for any other purpose.

We accept no responsibility for the consequences of this document being relied upon by any other party, or being used for any other purpose, or containing any error or omission which is due to an error or omission in data supplied to us by other parties.

This document contains confidential information and proprietary intellectual property. It should not be shown to other parties without consent from us and from the party which commissioned it.

This Report has been prepared solely for use by the party which commissioned it (the 'Client') in connection with the captioned project. It should not be used for any other purpose. No person other than the Client or any party who has expressly agreed terms of reliance with us (the 'Recipient(s)') may rely on the content, information or any views expressed in the Report. This Report is confidential and contains proprietary intellectual property and we accept no duty of care, responsibility or liability to any other recipient of this Report. No representation, warranty or undertaking, express or implied, is made and no responsibility or liability is accepted by us to any party other than the Client or any Recipient(s), as to the accuracy or completeness of the information contained in this Report. For the avoidance of doubt this Report does not in any way purport to include any legal, insurance or financial advice or opinion.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Contents

Executive summary 3

1 Introduction 81.1 Project background 81.2 Problem statement 81.3 Understanding the need for an FFEWS 91.4 Study area 91.5 River system in the basin 111.6 CDMA and GSM coverage in Nepal 14

2 Hydro meteorological data 162.1 Introduction 162.2 Hydro-meteorological gauge densities in Nepal and other countries 172.3 Existing hydro-meteorological network Nepal 17

2.3.1 Rainfall 172.3.2 Evaporation 182.3.3 Temperature 18

2.4 Water level stations 182.5 Discharge stations 182.6 Gridded Meteorological data 19

2.6.1 APHRODITE precipitation data 192.6.2 TRMM3B42 Precipitation 192.6.3 MODIS Snow Cover Data 20

2.7 Forecasted Meteorological data 202.8 Summary of availability of data 20

3 DHM and existing flood forecasting models 223.1 DHM’s mandate 223.2 Existing flood forecasting models in Nepal – an overview 233.3 Examples of operational flood forecasting models from other countries 233.4 Dissemination of forecast 24

4 Flood forecasting modelling 254.1 Flood forecasting modelling frame work 254.2 Objectives of flood forecasting modelling 264.3 Gauge-to-gauge correlation 274.4 Hydrological modelling 284.5 Routing modelling 284.6 Hydrodynamic modelling 28

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

4.7 Modelling software 294.8 Modelling software comparative list 304.9 FFEWS cost consideration 30

5 Rain gauge network design 315.1 Introduction 315.2 Auto telemetry rain-gauge 31

5.2.1 Description 315.2.2 Time of observation 325.2.3 Operation and measurement 325.2.4 Data transmission, storage and archive 33

5.3 Radar rain gauge 335.3.1 Description 335.3.2 Specification 33

5.4 Rain gauge network recommended for installation 345.5 Budget for proposed rain gauge network installation 37

6 Hydrometric network design 386.1 Water level gauge network 38

6.1.1 Description 386.1.2 Time of observation 396.1.3 Operation, measurement and maintenance 396.1.4 Data transmission, storage and archive 39

6.2 Discharge measurement stations 396.2.1 Description 396.2.2 Discharge measurement equipment 396.2.3 Cableway flow measurement 406.2.4 Equipment budget for discharge measurement 41

6.3 Hydrometric gauge recommended for installation 426.4 Hydrometric gauging network budget 45

7 Topographic and asset survey 467.1 Topographic survey 467.2 Survey budget 477.3 Satellite imagery 47

8 Flood forecasting model development 488.1 Mathematical modelling 488.2 Rationale for different forecasting approaches 508.3 Gauge-to-gauge correlation 518.4 Hydrological modelling 53

8.4.1 Review of existing data and models 548.4.2 Catchment delineation 54

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

8.4.3 Hydrological input: Rainfall, temperature and PET 558.4.4 Bias correction 558.4.5 Calibration 558.4.6 Validation 56

8.5 Combined rainfall-runoff and gauge-to-gauge correlation 568.6 Pilot pure 2-d modelling 578.7 1-d modelling 59

8.7.1 River network 608.7.2 Calibration and validation 62

8.8 1-d/2-d linked modelling 628.9 Operation of forecasting model 64

8.9.1 Key tasks 648.9.2 Real-time data transmission and maintenance 658.9.3 Existing forecast model operating system within DHM 668.9.4 Delft-FEWS 678.9.5 Dissemination of forecast 688.9.6 Data assimilation 68

8.10 Evaluation of forecast 698.11 Model development schedule 698.12 Model development budget 708.13 Person-months for experts 71

References 73Appendices 75

A. Modelling software comparison 76

B. Comments and responses 79

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 1Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

List of abbreviations

ADB - Asian Development Bank ASCE - American Society of Civil Engineers CBA - Cost Benefit Analysis CBDRM - Community Based Disaster Risk Management CDMC - Community Disaster Management Committee DDC - District Development Committee DDRC - District Disaster Relief Committee DEM - Digital Elevation Model DEOC - District Emergency Operation Centre DHM - Department of Hydrology and Meteorology DMF - Design and Monitoring Framework DoWRI - Department of Water Resources and Irrigation DPR - Detailed Project Report DWIDM - Department of Water Induced Disaster Management EARF - Environmental Assessment Review Framework EIA - Environmental Impact Assessment EIRR - Economic Internal Rate of Return EMP - Environmental Management Plan EPR - Environmental Protection Rule EWS - Early warning system FFEW - Flood forecasting and early warning FHRMP - Flood Hazard Mapping and Risk Management Project FIRR - Financial Internal Rate of Return FMA - Financial Management Assessment GDP - Gross Domestic Product GESI - Gender and social inclusion GFS - Global forecast system GIS - Geographic information system GLOF - Glacier Lake Outburst Flood GoN - Government of Nepal GPS - Global Positioning System ICIMOD - International Centre for Integrated Mountain Development IEE - Initial Environmental Examination IP - Indigenous People IPP - Indigenous People Plan IPPF - Indigenous People Plan Framework IRP - Involuntary Resettlement Plan IRPF - Involuntary Resettlement Plan Framework LDC - Least Developed Countries MoHA - Ministry of Home Affairs MoEWRI - Ministry of Energy, Water Resources and Irrigation MOUD - Ministry of Urban Development NAPA - National Adaptation Programme of Action NEOC - National Emergency Operation Centre NPR - Nepalese Rupiah NPV - Net Present Value OPEC - Organization of the Petroleum Exporting Countries

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 2Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

PAM - Project Administration Manual PCP - ADBs Public Communication Policy PEOC - Provincial Emergency Operation Centre PEP PET

- -

People’s embankment program Potential evapotranspiration

PMU - Project management unit PRA - Project Risk Assessment PSA - Poverty and Social Analysis RAH - Resettlement Affected Household RRP - Recommendation Report to the President RUDP - Regional Urban Development Project SDAP - Social Development Action Plan SDG - Sustainable Development Goals SMS - Short Message Service SPRSS - Summary poverty reduction and social strategy SPS - ADB Safeguard Policy Statement TOR - Terms of Reference UK - United Kingdom USD - Unites States Dollar VDC - Village Development Committee WC - Working Committee WECS - Water and Energy Commission Secretariat WRF - Weather research and forecasting WRPPF - Water Resources Project Preparatory Facility

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 3Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Executive summary

It has been proposed to develop a flood forecasting and early warning system (FFEWS) for Mohana-Khutiya basin. This includes a simple tool for use with the most advanced hydraulic models being employed in countries like Australia and UK. The simple tool could be operational within eight months or as soon as some hydrometric data become available from the new proposed gauge network. Over a period of three years, advanced hydraulic models will be developed, calibrated, validated and will be made operational as more and more data become available from the new gauging network and new measurements. The following forecasting tools have been proposed:

● Gauge-to-gauge correlation: the simplest and cheapest method, fast to develop, and thus could be operational soon; however, it has a very short lead time and is not appropriate in upper steep slope river reaches. There are also other limitations.

● Combined rainfall-runoff and gauge-to-gauge correlation: due to the addition of a runoff model, the forecast lead time could be up to 72 hours; however, this requires a stage-discharge rating curve at each gauging station; such rating curve is difficult to develop for out-of-bank flow conditions without a hydraulic model.

● 1-d model: this tool will be developed for the entire river system in the Terai and is appropriate for flood forecasting; the same model type is used in Bangladesh.

● 1-d/2-d linked model: this will be the final delivery around month 24; the pure 2-d model and 1-d model will be transformed into a 1-d/2-d linked model; this is the advanced forecast model used in some areas of Australia, UK and Malaysia.

Rationale for different forecasting approaches

The four approaches described above are inter-linked and essential and/or complementing components to the final deliverable/ flood forecasting and early warning system (FFEWS) model, i.e., the 1-d/2-d linked FFEWS model. The rationale, advantages and disadvantages of each approach are described below:

● Gauge-to-gauge correlation: the simplest and cheapest method. This could be an option to use as a quick forecasting tool. It can generate new knowledge, to be translated into the final deliverables (1-d model and 1-d/2-d linked model). Advantages will be that CBDRM could be operational earlier and potential areas of uncertainty in flood level forecast could be identified. DHM is using this method in many of their river basins, e.g., in Karnali. This tool and expertise from DHM could readily be used in this basin with some nominal input from international consultant; as the tool has to be customised for new basin, need for minor changes in code and parameters may be required and thus international consultant’s input is considered. Thus, a minimum budget has been proposed for developing this tool. There will be a deployment time in all of these six basins, for new hydro-meteorological data to become available, and that this work is a good utilisation of the deployment time, as it generates the opportunity for transferring early knowledge to the final product.

● Rainfall runoff model is the main input to all other components: a) gauge-to-gauge correlation, b) 1-d river model, c) pure 2-d model and d) 1-d/2-d linked model. Combining the rainfall model with gauge-to-gauge correlation will increase the lead time (as in the rainfall forecast) up to 24, 48 and 72 hours. However, at the forecasting points, the discharge vs water level rating curve shall be required so that forecasted runoff can be converted to the water level using the rating curve. The rainfall runoff model provides inflows from the upper

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 4Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

catchment and distributed inflows from intermediate catchments to the 1-d, 2-d and 1-d/2-d linked model. The accuracy of the flow forecast depends on the accuracy of the rainfall forecast, which decreases with increasing lead-time

● The 1-d hydraulic model, as a standalone tool, can be applied as a forecasting tool once it is ready; without the 1-d model, a linked 1-d/2-d model (which is proposed as final deliverable) cannot be developed. Therefore, we have proposed to employ a 1-d model as forecasting tool as soon as it is ready. In any case, for certain reaches of the river, there will only be a 1-d model, as a 1-d/2-d linked model is not feasible to be developed for the entire reach of the river. This tool will also give useful feedback on forecasting performance, which then could be translated into the final deliverable. In summary, 1-d model development is not a duplicating tool; it is an essential pre-requisite. Should DoI and ADB decide not to take forward 1-d/2-d linked modelling, then a 1-d model will be the final product. This is the tool which DHM operate in the Bagmati, Koshi and West Rapti basins. The advantage of a 1-d model is that it runs efficiently, which is a key requirement for real time forecasting. However, a 1-d model does not have direct map output for flood risk or hazard; these require separate and customised GIS development, e.g., as practiced by forecast model in Bangladesh (http://ffwc.gov.bd/). Such a GIS tool is under development within DHM. It will need to be developed in this project in the 1-d only model reaches of the river.

● A 1-d/2-d linked model is the final deliverable; such FFEWS models are already in operation in countries like Australia, New Zealand, Malaysia and UK (Syme, 2007; Huxley, 2016). Therefore, we recommend developing this next generation FFEWS tool, otherwise by the time this project is complete (2-3 years from now), it might seem that Nepal uses less advanced tools than other countries. The 1-d/2-d linked model can forecast flood levels with better accuracy (as it is linked to 2-d floodplain model); flood risk and hazard maps are direct outputs from such modelling. However, run-time is longer than for the 1-d model. As such it is not feasible to develop it for all reaches of the river. For selected river reaches, where such modelling will be useful, like in the Lower Terai, this tool shall be developed using dense cross-sectional data (proposed for this study) in combination with DEM. To overcome run-time issues for real time forecasting, GPU (graphical processing unit) or HPC (heavily parallelised computing) versions of modelling software shall be used.

In several meetings with DHM, the consultant has proposed the development of a similar FFEWS model, with regard to modelling tools and types of models. We have proposed the same type of advanced 1-d model development for FFEWS, which DHM is presently operating in three different basins (West Rapti, Bagmati and Koshi). The same (or similar) modelling software (e.g. MIKE11 and HEC-RAS), for both hydrological and hydrodynamic modelling, has been recommended (in parallel with other software), thus giving DHM wider options to choose from.

Rain gauge network installation

Eleven new auto telemetry rain gauge stations have been proposed for installation. Data will be recorded at 15-minute intervals. There are six existing rain gauge stations within this basin; none of them are telemetry stations (source: http://www.hydrology.gov.np, real time data). This will deliver one rain gauge in every 64km2 over the basin, similar to England where rain gauge density is highest in Europe. Flood prediction in rural and urban areas requires dense spatial gauge networking: one gauge between 10 to 100km2.

Hydrometric gauge network installation

Nine new telemetry hydrometric stations have been proposed: five in Mohana-Godawari and four in Khutiya-Shivaganaga catchments. Data will be recorded at 15-minute intervals. There are two existing hydrometric stations operated by DHM within this basin; neither of them are

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 5Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

telemetry stations ((source: http://www.hydrology.gov.np, real time data). The locations have been chosen carefully to allow good calibration of runoff from the hydrological model and river levels in the hydrodynamic model. At five stations, both water level and discharge will be recorded, and at four stations, only water level will be recorded. Discharge will be measured using ADCP or propeller type current meter depending on flow condition (at very low flow, ADCP measurement is not suitable). One of the above discharge stations will be a cableway station. Stage-discharge rating curve shall be developed at all five discharge stations. The proposed locations will be finalised through discussion with DHM. DHM’s site selection criteria and criteria in other international manual shall be followed.

Hydrometric equipment

ADCP, DGPS and echo-sounder will be purchased for discharge measurement. This set of equipment will be used for discharge measurement; this basin will have one set of equipment as the discharge measurement frequency are fortnightly, and there are five discharge stations within this basin.

Total three set of equipment has been proposed for six basins (Mohana-khutiya one, Mawa-Ratuwa one, Lakhandei one, and East Rapti, West Rapti and Bakraha none. Bakraha will share the Mawa-Ratuwa one, and East and West Rapti will share from DHM’s existing set of equipment).

Topographic and asset survey

Topographic survey will include river sections sufficiently extended across the adjacent floodplain, any existing structures and a flood embankment profile. Survey will have to be done in the Mohana, Godawari, Monohara, Khutiya and in Shivaganga river. Along 140km, 369 cross-sections will have to be surveyed. In steep river sections for accuracy in hydraulic model, cross-sections between 200 and 500m intervals are generally essential (HEC-RAS, Users’ Manual, Version 4.1, Figure 8-34). We have proposed cross-sections at, on average, 379m intervals.

For topographic survey, no survey equipment has been proposed for purchase; survey will be done through outsourcing.

FFEWS modelling budget

The FF model includes development of models, development of a tool for automation of FF operation and development of a tool for automation of forecast dissemination. This budget (Table 1) will be required over a three-year period; forecasting will start with the simplest tool from month 8 or 9 of the project using the gauge-to-gauge correlation method. Over the three-year period, advanced sophisticated 1-d, 2-d and 1-d/2-d linked models will be delivered and will remain operational for the years to come.

Table 1: Mohana-Khutiya basin FFEWS budget

Categories Parameter Unit Quantity Capital/Development cost

($)

Operation cost ($)

Dissemi-nation

cost ($)

Unit cost

($) Data: collection, processing, analysis

Per basin No. 1 48,500 - - 48,500

Hydrological modelling

Catchment area

km2 702 137,500 - - 196

Gauge-to-gauge correlation

River length km 88 80,500 32,500 27,500 1,597

Pure 2-d modelling River length km 31 93,000 37,500 27,500 5,097

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 6Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Categories Parameter Unit Quantity Capital/Development cost

($)

Operation cost ($)

Dissemi-nation

cost ($)

Unit cost

($) 1-d modelling River length km 131 144,500 73,500 35,000 1,931 1-d/2-d linked modelling

River length km 52 169,500 64,500 26,250 4,967

Modelling software Suite No. 1 13,000 - - 13,000 Total 686,500 208,000 116,250 Note: Modelling software licence cost is distributed over five basins; software will have multi user network licence, and cost shown here is per basin. West Rapti is excluded from software cost Source: Mott MacDonald

The hydro-meteorological data network budget includes establishing auto-rain gauges (Table 2), and auto and manual water level gauges and discharge measurements (Table 3); discharge measurement is to be carried over a period of three years, while rainfall and water levels are to be collected for three years for this project and also to be maintained beyond the period of this project. Measurement of discharge beyond three years (this project period) will be left to DHM’s choice whether further occasional discharge measurement to be carried out (or not) by their trained technical staff (who will be trained in this project).

Table 2: Budget for proposed rain gauge network in Mohana-Khutiya basin Hydrometeorological data network Mohana-Khutiya budget (US$) No. of

stations No. of

measurements Capital cost/

measurement cost

Unit cost Maintenance cost: 3 years

Total cost

Ground-based tipping bucket auto telemetry

11 - 55,000 5,000 6,000 61,000

Total 11 55,000 5,000 6,000 61,000 Source: Mott MacDonald

Table 3: Water level and discharge gauge network budget in Mohana-Khutiya basin Hydrometric data network Mohana-Khutiya budget (US$) No. of

stations No. of

measurements Capital cost/

measurement cost Unit cost

Operation & Maintenance cost: 3 years

Total cost

Discharge 4 120 520,000 4,333 6,000 526,000 Water level 5 - 35,000 7,000 6,000 41,000 Note: a) Discharge measurement to be carried out fortnightly from mid-May to mid-October; this will be 10 measurements per year, 30 in 3 years at one station and total 120 measurements in 4 stations; b) Operation and maintenance cost is $2,000 per basin for all stations per year; this involves routine site visits, repair and maintenance of the gauge, sediment removal etc. c) Discharge measurement cost is a continuous expenditure, like model development cost (and should be considered similar to capital cost); it includes cost for all skilled human resources and the logistics required

Source: Mott MacDonald

Topographic survey will be outsourced, and thus no procurement of survey equipment is proposed. The budget (Table 4) included here is for surveying cross-sections in Mohana, Godawari, Monohara, Khutiya and Shivaganga rivers. Budget for purchasing high resolution satellite imageries has been included (Table 5) for Mohana and Khutiya basin for lower catchment only in Terai to provide DEM to 1-d and 2-d model development and flood inundation map preparation.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 7Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Table 4: Topographic survey budget for Mohana-Khutiya basin Topographic cross-section survey

Mohana - Khutiya survey budget (US$) Length of survey

(km) No. of XS Total cost Unit

cost Mohana 56 147 29,400 200 Godawari 12 30 6,000 200 Monohara 10 26 5,200 200 Khutiya 42 109 21,800 200 Shivaganga 21 55 11,000 200 Total 141 367 73,400 -

Source: Mott MacDonald

Table 5: Satellite imagery purchase budget for Mohana-Khutiya basin High resolution (50cm) satellite imagery

area (km2) Total cost Unit cost (USD for 1 sq.km)

Mohana basin in Terai 232 11,600 50 Khutiya basin in Terai 214 10,700 50 Total 446 22,300 -

Source: Mott MacDonald

Hydrometric equipment (DGPS, Echo-sounder and ADCP) is a capital expenditure. This equipment set will be used for fortnightly discharge measurement as mentioned earlier and will be used and remain available for discharge measurement over a period of three years and beyond through maintenance of the equipment set. The equipment set also includes cost for construction of one Cableway discharge measurement station on Mohana river (Table 6).

Table 6: Discharge measurement equipment / station budget in Mohana-Khutiya basin Hydrometric equipment and installation

Mohana-Khutiya budget (US$) Capital cost Operation and

maintenance (total for 3 years)

Total cost

DGPS 25,000 1,250 26,250 Echo-sounder 25,000 1,000 26,000 ADCP 35,000 1,000 36,000 Cable way discharge station: construction cost

95,000 - 95,000

Total 180,000 3,250 183,250

Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 8Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

1 Introduction

1.1 Project background Acknowledging the importance of the Terai region to Nepal, the Government of Nepal (GoN), through the Ministry of Energy, Water Resources and Irrigation (MoEWRI), is implementing the ‘Priority River Basins Flood Risk Management Project’ in the Southern Nepal Terai region. The project is the continuation of the pre-feasibility study: Package 3: Flood Hazard Mapping and Risk Management Project (FHMRMP, 2016).

During the pre-feasibility study from the 25 basins, six priority basins were selected and included in the cost-benefit analysis: i) West Rapti, ii) Mawa–Ratuwa, iii) Lakhandei, iv) Mohana -Khutiya, v) East Rapt, vi) Bakraha. Bakraha replaced the Biring basin; the Khutiya basin was added to the Mohana basin, and Mawa was added to the Ratuwa basin.

In this study, feasibility level design for developing an FFEWS in the above basins (excluding West Rapti) has been prepared. Note here that FFEWS for the West Rapti basin is currently being developed by the Department of Hydrology and Meteorology in Nepal, funded by the World Bank in the project ‘Building Resilience to Climate Related Hazards (BRCH)’

This feasibility report for development of FFEWS is for Mohana-Khutiya basin.

1.2 Problem statement Nepal is considered to be one of the most disaster-prone countries in the world. Alongside other natural hazards, such as earthquakes and landslides, flooding poses risk to large sections of the population. Heavy damage to infrastructure, loss of agricultural production, disruption of livelihoods and loss of lives in Nepal due to floods are frequent occurrences during summer monsoons. It is also expected that economic losses associated with floods is likely to rise with increasing economic and development activities in the flood plains.

Holistic management of flood risk requires actions to reduce impact before, during and after extreme events and includes preventive technical measures as well as socioeconomic aspects to reduce vulnerability to hazards. Although flood disaster risk assessment and management processes have been used by the Government agencies in Nepal to help estimate and manage risks associated with floods, these tools are in general not available (other than isolated flood and erosion control structures) in the five basins under this study and as a result may not serve these basins in an optimal way.

One of the first steps in flood disaster risk reduction is to identify risks. Knowledge of risks raises awareness and allows pre-event planning in contrast to post-event response and recovery. In this context, flood risk management must be coordinated with other development activities in the flood plains, and particularly water resources development in a river basin. To do this, it is necessary to understand better the extent to which the current level of information related to flood disaster risk is adequate for development planning, and societal risk management practice and whether or not this level can be improved. Besides, it requires assessing the degree to which flood risk management has been integrated in other development activities so far and whether or not this integration can be improved by a thorough understanding of flood hazards in river basins, especially in Terai with large flat flood plains.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 9Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

1.3 Understanding the need for an FFEWS One of the key non-structural measures of reducing flood disaster risks is the provision of a reliable and accurate flood forecasting system including a thorough understanding of flood propagation in rivers and inundation of large flood plains.

Provision of a reliable and accurate flood forecasting system with adequate lead time has been recognised by the Government of Nepal as a key non-structural measure to reduce flood disaster risks. However, due to lack of an integrated hydro-met monitoring network in these six basins and due to lack of real time forecasting technology and tools, the capability to meet the demand of a modern real-time flood forecasting and warning system is limited. An effective flood forecasting and warning system has to be based on hydrological and hydrodynamic models to simulate rainfall-runoff from precipitation and to simulate propagation of floods along the tributaries, main streams and the flood plains. Using real time rainfall data from upper and lower catchments, meteorological forecasts and river gauge data, flood forecasts for up to three days in advance can be developed using the modelling tools. The generated forecasts on flood level and discharge shall be translated easily into understandable warnings including flood inundation maps/risk maps for community-based disaster risk management activities. The forecasts, warnings and risk information shall be disseminated as widely as possible via Internet, mobile phones, public and private media, social media and other means of communication.

Flood risk consists of three key components: i) problem of repeated occurrence, ii) exposure of people ad assets to flood, and iii) vulnerability. FFEWS will reduce exposure and vulnerability of those exposed.

1.4 Study area The catchment of the Mohana Khutiya river basin lies between Northing 3168000m to 3208000m (latitude 28°38′1.76″N and 28°58′59.15″N), and between Easting 448000m to 478000m (longitude 80°31′36.73″E and 80°45′24.74″E) in WGS 84, UTM Zone 44 N (see Figure 1). The basin extends from the Chure Hills (Siwalik Hills, also known as sub-Himalayan hills, at low altitude) in the north and in Terai (meaning low flat land) up to the Indo-Nepal border in the south. The catchment covers an area of 702km2 in the far west of Nepal (Figure 1). The Mohana-Khutiya river system lies in the district of Kailali in Province no 7. This river system has 359 settlements distributed over rural and urban municipalities with a population of 190,063 and 37,681 households (CBS, 2011). Dhangadi and Attrariya are the two major towns located in this catchment.

The basin has two hydrometric stations, which collect water level and occasional discharge, and also seven meteorological stations which collect rainfall (with other meteorological parameters) with another two located close to the catchment boundary.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 1: M

Source: Mott

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Mohana-Khu

MacDonald

ration of Priority Rning System: Moh

9 ning System: Moh

utiya basin lo

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

ocation map

d Risk Managemesin

sin

p

ent Project, Nepal 100

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 11Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

1.5 River system in the basin Brief summary of the river system in this basin is presented below in Table 7 and in Figure 2. Reach-wise detailed classifications of these morphologically active rivers have been presented in a separate report in this study (Mott MacDonald, 2018). A brief description, mainly useful for developing hydrological, hydraulic and flood forecasting models, is presented below.

This catchment originating from Chure Hills is relatively small (702km2), about 1/5th of the size of West Rapti at Bagasoti gauging station. The West Rapti at Bagasoti has a time of concentration of 10hrs (DHM, 2018). Thus, this basin is expected to have much smaller time of concentration. And this will affect both forecast lead time and response time. In steep slope reaches (upper reaches), the flood travel time, in general, is fast. Therefore, gauge to gauge correlation forecasting, which is practised in many basins by DHM in Nepal, will not be generally suitable, particularly in upper reach, as both forecast lead time and response time will become small in this basin. Rivers in Nepal are flashy; as such the lag time (time required to attain peak flow after a rainfall event) is very short. This necessitates that rainfall-runoff and hydraulic models are developed for the well-defined reaches of river and get the benefit of 1 to 3 days of lead time on rainfall forecast from the weather forecast model. The hydraulic model, irrespective of flood lag time, response time and slope of the rivers, will be able to forecast water level and their propagation with same lead time (1 to 3 days) as in weather forecast model. However, depending on the river characteristics, appropriate type of hydraulic model should be developed. In steep slope reaches, flood propagates fast and flooding spreads less in the limited floodplain; so developing 1-d model will be more appropriate in those reaches. In gentle slope reaches, the flood propagation is slow and flood inundates more areas in meandering and braided floodplain; thus, 1d/2d linked model will be more accurate and beneficial to warn people. Considering the characteristic features of the river systems, e.g., hills, river braiding, meandering etc., the types and domains of models, have been identified (see Section 8). However, development of appropriate type of model (1d, 2d or 1d/2d linked) should require several iterations during development stage of these models.

Table 7: Summary of river system in Mohana-Khutiya basin River Reach

ID Reach Characteristics Channel

length (km) Slope (%)

Mohana River

1 Hill 4.97 14.1 2 Fan 8.37 1.1 3 Peripheral fan (Godawari Confluence) 13.87 0.1 4 Flood plain, meander (India border) 21.29 0.1 5 Flood plain, meander (India border-Khutiya) 21.34 0.1

Godawari River

1 Hill 9.26 13.4 2 Braided 6.97 1.2 3 Flood plain, Meander (Mohana Confluence) 6.52 0.2

Monohara 1 Hill 6.29 16.2 2 Braided 4.80 0.9 3 Flood plain, meander (Mohana confluence) 8.59 0.1

Khutiya 1 Hill 19.91 8.3 2 Braided 6.51 1.1 3 Flood plain, Meander (Shivaganga confluence) 15.46 0.2 4 Meander (Mohana confluence (Nepal-India

border) 9.77 0.1

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 12Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

Shiva Ganga

1 Hill 10.92 14.7 2 Braided 3.78 1.4 3 Flood, plain, meander Shiva Ganga (Khutiya) 19.55 0.2

Source: Mott MacDonald

Mohana river has been divided into five reaches (Table 7). Total length is 70km. Reaches are mainly straight and meandering; there may be transitional lengths between them. The meandering reach is much higher, 57km; the straight/transitional reach is 13km. Channel slope is very steep, particularly in the first hilly reach. In the first two straight reaches, sediments are mainly boulders, gravels and sands, while in the meandering reaches, sediments are sand and silt (and might have a minor fraction of fine gravel).

Godawari is a tributary to the Mohana and located in the Hills. The river has been divided into three reaches (Table 7). Total length is 23km. All reaches tend to be straight, though the third downstream reach tends towards meandering. The first two reaches have steep slopes, and the last reach has relatively mild slopes. Sediments are mainly boulders, gravels and sands in the two steep reaches and sand and silt in the last reach.

Monohara river is a tributary to the Mohana and located in the Hills. Total length is 23km. The river has been divided into three reaches (Table 7). The first two reaches are straight, and the third reach tends towards meandering. The first reach has steep slopes, the other two reaches have relatively mild slopes. Sediments are mainly boulders, gravels and sands in the first two steep reaches and sand and silt in the last reach.

The Khutiya river is the major tributary to the Mohana. The river has been divided into four reaches (Table 7). Total length is 52km. The first reach is straight and the other three in the downstream tend towards meandering. The first and second reaches have steep slopes, and the third and fourth have very mild slopes. Sediments are mainly boulders, gravels and sands in the first two steep reaches, and sand and silt in the last two downstream reaches.

The Shivaganga river is a tributary to the Khutiya; the river has been divided into three reaches (Table 7). Total length is 34km. The first two reaches are straight and the third (downstream) reach is meandering. Straight reaches have steep slopes, while the meandering reach has mild slope. Sediments are mainly boulders, gravels and sands in the first two steep reaches, and sand and silt in the meandering reach.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 2: R

Source: Mott

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

River system

MacDonald

ration of Priority Rning System: Moh

9 ning System: Moh

m in Mohana

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

a Khutiya ba

d Risk Managemesin

sin

asin

ent Project, Nepal 133

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 14Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

1.6 CDMA and GSM coverage in Nepal Real time data acquisition and the issuing and dissemination of flood alerts and warnings from the FFEWS model are the prime objectives. The proposed FFEWS will primarily use the code-division multiple access (CDMA) and Global System for Mobile communication (GSM) technologies of Nepal for dissemination of the flood alerts and warnings. CDMA technology has been used by NTC while the GSM technology is supported by other mobile providers in Nepal. Ncell is one of the companies with the largest GSM networks in Nepal. Both companies are providing services to DHM for real-time data acquisition. However, these service providers have several gaps in their network coverage. In these gap areas, the hydro-meteorological stations cannot transmit data on a real-time basis. Since these technologies are based on line-of-sight communication, some of the hydrometric stations located in deep gorges do not have connection even within the area of their coverage.

The existing CDMA and GSM in Nepal is shown in Figure 3. This basin, Mohana and Khutiya, seems to have good GSM network, and thus will be very useful for establishing the FFEWS in this basin. However, for installation of any new proposed hydro-meteorological monitoring station, GSM must be checked prior to installation, and if needed, station location could be shifted through discussion with DHM.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 3: G

Source: DHM

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

GSM coverag

M, 2018

ration of Priority Rning System: Moh

9 ning System: Moh

ge in Nepal

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

d Risk Managemesin

sin

ent Project, Nepal

155

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 16Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

2 Hydro meteorological data

2.1 Introduction Certain types of hydro-meteorological data are essential for different types of flood forecast modelling. Meteorological data types are: rainfall/precipitation (P), temperature (T) and potential evapotranspiration (PET); they are input data for rainfall-runoff (RR) modelling in any hydrological modelling tool. Hydrological data types are: discharge (Q) and water level (H); they are required at boundary conditions as well as at calibration and validation locations of rainfall runoff (Q required) and 1-d and 2-d hydrodynamic models (both Q and H required).

Development of the base model, whether for flood forecasting or for design of flood protection works, would not essentially be very different. In order to be more run efficient, the forecasting model could be simplified in some places, though not at the expense of accuracy. The model has to be calibrated against several past events and the calibration process would be enhanced by testing a greater number of events. The model should be able to replicate any event, for a wide range of return periods (from very low to high exceedance probability). Both long term data and short-term past storm event records can be used for calibration. Long-term data are more appropriate and do not need to be continuous. During the operational phase of the flood forecasting model, continuous data during monsoon will be essential to collect as the model will operate at real time daily.

Following data will be required at daily or sub-daily temporal resolution in each phase of FFEWS model development, namely, calibration, validation and operational phase.

● Cumulative rainfall; runoff is the response of total rainfall, rather than a rate of rainfall. As such rainfall is required as input to the runoff model

● Mean temperature ● Cumulative PET; for same reason as rainfall, cumulative PET data is used as input in the

runoff model ● Water level ● Discharge

Snow cover data may not be required as the altitude of the basin, is below 3000m, and thus, understood that catchment runoff is not from snow-fed (Putkonen, 2004).

For example, in the calibration phase, parameters’ values within the RR models for each of the sub-catchment will be tuned so that differences (errors) between modelled Q and observed Q are minimum and down to acceptable levels. The level of acceptability needs to be agreed with the client (DHM) and with due reference to best practice RR modelling (DHI, 2014; HEC-HMS). During the validation phase, the performance of the model is evaluated, without changing any parameters established during calibration phase, and similar standard of matching between observed and modelled Q and WL should be obtained; otherwise, a recalibration would be needed followed by validation.

In the event of observed meteorological data are insufficient or not available in all the sub-basins, data from other sources should be explored. This necessitates the consideration satellite-based rainfall, temperature, snow cover etc. Such data are available as gridded data, generally available in good spatial resolution, and are mainly derived from long records of observed gauge-interpolated data (see Section 2.6 for details).

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 17Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

In this chapter, available data from DHM, and gridded data from a number of sources have been discussed.

2.2 Hydro-meteorological gauge densities in Nepal and other countries Density of hydro-meteorological stations in Nepal and in some other countries in the world is presented in Table 8. The density is mainly dependent on purpose, e.g. the rainfall data will be used for irrigation, flood risk assessment or for flood forecasting purposes. Flood prediction in rural and urban areas requires a dense spatial gauge network, one gauge between 10 to 100km2 and higher temporal measurement frequency between minutes to hour (Berndtsson and Niemczynowicz, 1988).

Table 8: Rainfall gauge density in Nepal and in some selected countries in the world Country Number

of gauges Average area per gauge

(km2) Nepal - 550

Nepal: Siwalik region - 430

Nepal: Terai region - 370 UK 3,214 76 England 2,169 60 France - 116 Netherlands - 130 Germany - 88 USA - 1,040 India - 790

Source: DoWRI (2016) for Nepal and Allot (2010), Met Office, England for other countries

2.3 Existing hydro-meteorological network Nepal DHM is the designated government agency for predicting and disseminating weather-based forecasts and warnings. In June 2018, DHM, maintained a total of 175 hydrometric stations, 337 precipitation gauging stations, 68 climatological stations and 15 synoptic stations. These stations include both real-time telemetry stations and non-telemetric stations.

Among the above stations, DHM, presently maintains a network of 28 hydrometric stations and 88 meteorological stations, as real-time telemetry stations. DHM is further upgrading 59 hydrometric stations to real-time telemetry stations. An additional seven stations are also under consideration for upgrading to real-time telemetry. In total, 182 hydro-meteorological stations are scheduled to become operational as real-time telemetry data acquisition systems in the near future.

2.3.1 Rainfall

Existing metrological stations within Mohana-Khutiya basin are shown in Figure 6. There are seven stations within the basin, and there are two more stations within the 4km buffer zone of the boundary of this basin; none of these stations are rea-time telemetric stations. These rainfall stations are understood to generate daily data. Period of records vary from station to station. However, in general, data are available from the 1970s; for example, Godawari station is operational since June 1976 to present time. While they will be useful for calibration and

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 18Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

validation of the hydrological model, real time telemetry rainfall at sub-daily frequency, usually 15minute frequency, will be required for hindcasting and forecasting and early warning system. As such new telemetry rain gauges have been proposed (see Section 5).

DHM has planned installation of three radar rain gauges across Nepal. One is under installation at Surkhet near Mohana and one at West Rapti; installation will be completed in a few months. This C-band long range radar has a 400km diameter range. Temporal resolution from bid document is understood to be of a minimum of 1-hour; however, the temporal resolution is to be agreed with DHM. The spatial resolution, though not mentioned in the bid document, could be of 1-km as obtained by long range C-band radar in UK and Germany (Lengfeld et al., undated).

Therefore, the forecasting model development will initially use existing rainfall radar data from DHM (if available) supplemented with gridded rainfall data available from satellite-based sources (see Section 2.6).

2.3.2 Evaporation

The same meteorological stations, which monitor rainfall, also collect daily pan evaporation data. However, hydrological modelling software, like NAM, uses monthly PET. There are seven stations within the basin, and there are two more stations within the 4km buffer zone of the boundary of this basin. Period of records vary from station to station, but in general data are available from the 1970s. Pan evaporation data from these stations will be used in the development of the hydrological model within the FFEWS.

2.3.3 Temperature

The same meteorological stations, which monitor rainfall, also collect temperature, daily minimum and daily maximum data are available from DHM. There are seven stations within the basin, and there are two more stations within the 4km buffer zone of the boundary of this basin. Period of records vary from station to station, but in general data are available from the 1970s. Temperature data from these stations will be used for development of the hydrological model within the FFEWS, in case there is any snow fed runoff in the basin,.

2.4 Water level stations The existing hydrometric stations (water level and discharge) within the Mohana-Khutiya basin are shown in Figure 8. There are two hydrometric stations within this basin – one in Mohana basin and the other in the upper catchment of Khutiya basin. None of these two stations are real-time telemetric station (source: http://www.hydrology.gov.np, real time data), Data from these two stations have not been published by DHM; during visit and discussion with DHM, it seemed that use of the data is also not recommended. Therefore, for development of the FFEWS models in this basin, the water level must have to be obtained from new proposed monitoring stations, which will be real time telemetric data (please see Section 6).

2.5 Discharge stations The existing discharge stations within the Mohana-Khutiya basin are shown in Figure 8. There are two hydrological stations within this basin – one in Mohana basin and the other in the upper catchment of Khutiya basin. None of these two stations are real-time telemetric station (source: http://www.hydrology.gov.np, real time data), Data from these two stations have not been published by DHM; during visit and discussion with DHM, it seemed that use of the data is also not recommended. Therefore, for development of FFEWS models in this basin, discharge must be obtained from new proposed monitoring stations, which will be real time telemetric. Through conversion of water level to discharge, continuous discharge time series will be generated by

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 19Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

using a discharge vs water level rating curve. The discharge vs water level rating curve will be developed in this project (see Section 6).

2.6 Gridded Meteorological data Gridded time series of meteorological data (rainfall/precipitation, surface temperature, evaporation and snow cover), spreading over the Nepal and border basins in China and India are available from a number of sources. Data are satellite-based, re-analysis based or gauge-

interpolated estimates. Gridded time series data shall be needed due to non-availability or scarcity of hydro-meteorological observations within in these six priority basins. The gridded products available are:

● TRMMv7 precipitation estimates ● APHRODITE precipitation and temperature products ● MODIS snow cover products Availability and quality of some of the gridded data are briefly discussed below.

During model development phase, time series of gridded data could be used from the above sources. However, before use, availability and quality for long records shall be examined.

2.6.1 APHRODITE precipitation data

Asian Precipitation - Highly-Resolved Observational Data Integration Towards Evaluation (APHRODITE) collects and analyses rain gauge observations from thousands of Asian stations; it has about 57 years of daily precipitation (P) datasets available between 1951 and 2007. APHRODITE (APH) precipitation data is gauge-interpolated and takes account of the orographic effect. Temporal resolution of the data is daily; spatial resolution is 0.25° lat/long (approximately 9km cell). APH data also has air temperature data with the same resolution as precipitation data.

APH daily precipitation data is freely available for non-commercial purposes (for Academic Institutions and Researches) provided proper acknowledgement is given. No commercial use is allowed. Use of APH data by Government of Nepal may be considered non-commercial, however, for use of the APH data, permission has to be granted. The spatial coverage of APH monsoon dataset extends from 60°E longitude in the west to 150°E longitude in the east and from 15°S latitude in the south to 55°N in the north. This means APH data cover almost all of Asia, including Nepal. It is available for the period 1951-2007, i.e. 57 years of daily precipitation data for each grid. This implies availability of daily data for calibration and validation of runoff models with long historical records. APH data are available in two spatial resolutions –viz. 0.5° lat/long and 0.25° lat/long. Compared to observed data from existing DHM stations, APH-P provides better spatial resolution of precipitation distribution (as well as for longer periods) over Nepal in this basin and also in the other basins of this study. It is expected that the mean areal precipitation series for each sub-catchment could be obtained from APH-P and will be used for hydrological modelling.

2.6.2 TRMM3B42 Precipitation

Tropical Rainfall Measuring Mission (TRMM) is a joint venture between National Aeronautics and Space Administration (NASA) and Japan Aerospace Exploration Agency (JAXA). TRMM3B42v7 is gauge-adjusted version of satellite-based precipitation estimates of TRMM satellites. The products are continuously released with 2 months of latency period. Spatial resolution is 0.25° lat/long and temporal resolution is three hours. Data are available from January 1998 to the present time. Data are freely available for all purposes. These data will

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 20Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

complement APHRODITE data which end in 2007. Thus, any calibration and validation of hydrological models beyond 2007 will be carried out using these precipitation data.

2.6.3 MODIS Snow Cover Data

The data on snow cover is required to decide upon the portion of a sub-catchment where snow melt generated runoff is the dominant process. None of the six basins within this study is considered to be under snow cover for significant duration. NASA’s MODIS snow cover product can be obtained freely from NASA’s MODIS link: https://modis.gsfc.nasa.gov/data/dataprod/ mod10.php.

2.7 Forecasted Meteorological data For operation of FFEWS, Quantitative Precipitation Forecast (QPF) estimates are essential, and Quantitative Temperature Forecast (QTF) estimates are desirable. For PET, flood forecasting model often use average of past monthly records.

The Numerical Weather Prediction (NWP) system of Indian Meteorological Department (IMD) provides QPF and QTF estimates which could be obtained through FTP access. IMD forecasts have 9km spatial resolution (0.081° lat/long), has 3-hours temporal resolution and 72-hours lead time. IMD-QPF is available at a finer spatial resolution than GFS (details below on GFS) for catchments in Nepal. Generally, the accuracy of the forecasts from any NWP model deteriorates as lead time increases.

IMD-QTF products have the same spatial and temporal resolutions as that of IMD-QPF products i.e. 0.081° lat/long, three hours temporal resolution and 72-hours lead time.

The Global Forecast System (GFS) is a weather forecast model produced by the National Centers for Environmental Prediction (NCEP) of NOAA. The GFS-QPF with 0.25° lat/long spatial resolution, 3-hours temporal resolution and lead time of 10 days is freely available. It is understood that DHM already uses GFS and The Weather Research and Forecasting (WRF) model to obtain QPF for forecasting in their existing FFEWS models.

As IMD-QPF has finer spatial and temporal resolution, and GFS-QPF is already in use by DHM, IMD and GFS are considered as the best source of QPF available to Nepal catchment at present. The best approach may be to treat IMD-QPF as the primary QPF should DHM be able to sign a treaty with IMD to obtain the IMD-QPF; else GFS-QPF will be used as this is already being used by DHM.

2.8 Summary of availability of data Summary on the availability of existing hydro meteorological data and forecasted rainfall data and their sources are presented in Table 10.

DHM’s rainfall data from DHM are available for a considerable number of years in seven stations, and should be usable for calibration and validation of the rainfall run-off model. However, discharge and water level data are not available in this basin as the data from the two stations data have not been published. In the rainfall data, the published reports of DHM (downloaded from their web site) show missing rainfall data as well. Maximum years of data are missing at Godavari (6 years) and at Teghari, 1980 to present time. We should note here that data from manual stations have been published in Data Book upto 2016, which indicates that DHM needs about 2 to 3 years to organise, analyse and quality check data before officially publishing.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 21Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Satellite based rainfall also available for many years; both APHRODITE and TRMM data could be used for running hydrological model.

Table 9: Hydro meteorological data – summary of availability

Data type Source of Data Station Collection

method Collection frequency Availability Latency/

Publication Period

available

Rainfall & pan evaporation DHM

Baliya  Manual Daily Data Book 2 to 3 years 2000 to 2016 

Dhangadi  Manual Daily Data Book 2 to 3 years 1957 to 2016 

Bichawa  Manual Daily Data Book 2 to 3 years 2001 to 2016 

Atariya  Manual Daily Data Book 2 to 3 years Not published 

Malakheti  Manual Daily Data Book 2 to 3 years 2001 to 2016 

Teghari  Manual Daily Data Book 2 to 3 years 1976 to 1979 

Godavari  Manual Daily Data Book 2 to 3 years 19 to 2016 

Temperature DHM Dhangadi  Manual Daily Data Book 2 to 3 years 1977 to 2016 

Godavari  Manual Daily Data Book 2 to 3 years 1976 to 2016 

Water level and discharge

Malakheti  Manual Daily - - Not published 

Mudivabar  Manual Daily - - Not published 

Gridded rainfall

APHRODITE - satellite-based Daily Online - 1951-2007

TRMM satellite-based 3-hourly Online 2 months 1988-present

Forecasted rainfall

IMD satellite-based 3-hourly Online (Near) Real time -

NCEP, NOAA satellite-based 3-hourly Online 7 hours -

Forecasted Temperature IMD satellite-based 3-hourly (Near) Real

time -

Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 22Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

3 DHM and existing flood forecasting models

3.1 DHM’s mandate Department of Hydrology and Meteorology (DHM) is the sole organisation in Nepal responsible for flood forecasting. One of the mandates of DHM is to provide early flood warnings to vulnerable communities and to major stakeholders. DHM, so far, have been able to establish flood Early Warning Systems (EWSs) in major rivers, in a few flashy rivers and in areas downstream of two glacial lakes that are considered potentially dangerous (DHM, 2018). The major river basins covered under EWS are: Karnali, Babai, West Rapti, Narayani, Bagmati, Kamala, Koshi, Kankai and Biring. Details of some of the EWS is discussed in next Section (Section 3.2). DHM’s aim is to extend flood forecasting services throughout the country.

The re-organised structure of the Government of Nepal implemented on 23 February 2018 has created the Ministry of Energy, Water Resources and Irrigation (MoEWRI). DHM used to operate under the Ministry of Environment and Population but now has been brought under the wing of MoEWRI with a mandate to provide weather and flood forecasts. With a new regulatory setup in place, DHM has been mandated to develop EWS including information dissemination components.

The organisational structure of DHM, which combines hydrology and meteorology, puts DHM in an ideal position to generate flood forecasts and issue forecasts and warnings. DHM has also been working with the Ministry of Home Affairs (MoHA), an organisation involved in the entire disaster management cycle, on disseminating flood warnings through National Emergency Operation Centers (NEOCs) and District Emergency Operation Centers (DEOCs). DHM has been supported by WMO since its establishment through collaboration on different meteorological activities and activities related to operational hydrology. As a member of WMO, Nepal has access to global and regional meteorological data required for monitoring and forecasting floods. Furthermore, DHM has received funding for several projects from WMO in the past, including projects on upgrading meteorological observation systems, weather forecasting, agriculture meteorology and hydrological services. In collaboration with the International Center for Integrated Mountain Development (ICIMOD) and countries in the Hindu Kush-Himalayan (HKH) region, WMO has been promoting the World Hydrological Cycle Observation System (WHYCOS) under the name of HKH-HYCOS. DHM has been contributing to this program by sharing real-time data for effective flow forecasting for the rivers originating from the Hindu Kush-Himalayan region.

With the implementation of the project ‘Building Resilience to Climate Related Hazards’ (BRCH), the World Bank has supported DHM by upgrading the existing flood forecasting system in Koshi and Rapti. This five-year project started in 2013 and its goal is to upgrade the real-time data acquisition system and establish an end-to-end flood forecasting system. Upgrading of Koshi and Rapti FFEWS are under progress at the moment. Nepal was also able to receive small grants from the Danish and Finnish governments for promoting DHM’s flood forecasting capabilities. Besides collaborating with the international community, Nepal has also been working closely with its neighbouring countries on upgrading its hydrological and meteorological monitoring systems. Since all the rivers of Nepal merge into the Ganga-Brahmaputra river system, Nepal has bilateral arrangements with India and Bangladesh that support the sharing of hydro-meteorological data and flood information.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 23Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Several INGOs are involved in disaster mitigation in Nepal. Nepal Red Cross Society (NRCS) and UNDP are involved in most of the disaster mitigation activities in Nepal. UNDP has been working with DHM to strengthen Nepal’s hydrological services and promoting community-based flood warning systems. Similarly, small-scale community-based flood warning systems have been implemented by other INGOs either in collaboration with DHM or in collaboration with other government agencies and NGOs. Since flood forecasts and warnings are widely used by communities and with several organisations simultaneously involved in disaster management, there are innumerable stakeholders involved (DHM, 2018).

3.2 Existing flood forecasting models in Nepal – an overview Probabilistic Flood Forecasting Model The model was developed jointly through a research partnership with Lancaster University (UK) and the International NGO Practical Action. The model assimilates rainfall and water levels to generate localised hourly flood predictions, which are presented as probabilistic forecasts, increasing lead times from 2-3 hours to 7-8 hours. The model predicts future water level at a site where thresholds for warning and danger levels are known. In this approach each gauged site where warning thresholds are defined requires its own model. The model was piloted in 2002 for the East Rapti River. The pilot model was enhanced and extended, expanding over the next 10 years to cover eight river basins across Nepal (Karnali, West Rapti, Babai, East Rapti, Narayani, Bagmati, Kankai and Koshi basins (Gautam and Phaiju, 2013). This model was made operational by DHM. Among them, West Rapti, Koshi and Babai FFEWS have been (or being) upgraded to advanced hydrological and hydraulic models (see below).

Advanced hydrological and 1-d hydraulic forecast model Nepal currently has operational flood forecasting models for the Koshi, West Rapti, Bagmati, and Babai catchments based on NAM/MIKE11 or HEC-HMS/HEC-RAS software which use rainfall QPF estimates from GFS and WRF. These models are rainfall-runoff and hydrodynamic models which forecast flows and flood levels. In addition, Nepal has gauge-to-gauge correlation forecasting covering most of the country, except basins smaller than about 300 to 400km2. This was informed by DHM Forecast Specialist during meetings with them between July to October 2018.

The Koshi flood forecasting model uses NAM modelling software for hydrological/rainfall runoff simulation, and MIKE11 for advanced (fully dynamic) 1-d river flow modelling. As the model uses rainfall forecasts from the GFS and WRF model, the lead time is up to 72 hours. The West Rapti flood forecasting model is under development in the same NAM/MIKE11 modelling system and thus has a similar forecasting ability as the Koshi flood forecasting model.

The Bagmati forecasting model is also operational and is developed in the HEC-HMS and HEC-RAS modelling system.

3.3 Examples of operational flood forecasting models from other countries Bangladesh has a dedicated flood warning centre in existence for 30 years, which has a FFEWS for the entire country’s river system in one model; the model is referred to as a super model, developed in the advanced hydrological (rainfall-runoff) modelling tool NAM and 1-d hydrodynamic model developed in the modelling tool MIKE11. The Bangladesh FFEW model has a 72-hour lead time. Similarly, advanced FFEW models have been developed in India for the Bagmati river basin in Bihar, the Krishna and Bhima river basins in Maharashtra, and the Brahmaputra river basin in Assam. Unlike those of Bangladesh and India, the UK’s FFEW models are for smaller catchments/basins because of the hilly terrain of the country and require

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

separate inmodels for modelling s

3.4 DissDHM dissemwarning is cDHM disseProvincial EOperation Cdifferent com

DHM, as mautomatic rnational-levand river lev2014). Dataavailable on

Quality maassessmentto review Fwarning sysbeneficiary entities, the

Figure 4: F

Source: DHM

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

dividual modindividual poftware) and

seminationminates floocommunicateeminates theEmergency Centers (DEmmittees to t

mentioned inreal time telvel DHM-manvel, with reala is transmitn www.hydro

anagement ot as well as r

FEWS and rstem dependcommunity

n such FEW

lood foreca

M (Courtesy Mr B

ration of Priority Rning System: Moh

9 ning System: Moh

dels for eacpriority basinsd Flood Mode

n of forecasod forecast aed to the come forecast toOperation C

EOC). DEOCthe local com

Section 2.3emetric statnaged web-l-time data ptted to the Dology.gov.np

of forecast responses frreassess itsd on the resand releva

WS is not wort

asting dissem

Binod Parajuli, H

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

ch basin. Nes. The UK’seller Pro (hyd

st and early wammunity leveo DHM bas

Centers, whoC then dissmmunity.

3, has upgrations. The abased flood ublicly acces

DHM server throughout t

is evaluaterom stakehol

reliability, asponse prognt agencies th investing i

mination sy

Hydrologist/Fore

d Risk Managemesin

sin

epal will alsos FFEW moddrodynamic m

arning from nel. The flow sin offices ao transmitsseminates th

aded many automated te

early warnissible througevery 15 m

the monsoon

ed based olders and coaccuracy angrams. If wa

or fail to rin.

ystem within

ecaster, Flood F

ent Project, Nepal

o need to dedels are devmodelling so

national leveof informatio

and to NEOthe forecast

he forecast

rainfall and elemetric gaing system,

gh the DHM wminutes, with n period (Gau

on feedbackommunity mend usefulnesarnings are ereach the a

n DHM

Forecasting Sec

evelop similaveloped in Poftware).

el, from wheron is shown

OC. NEOC tt to District to media a

hydrometric uge networkwhich moni

website (Shrflood warni

utam and Ph

ks. This incembers. Feeds. Benefits either neglec

affected indiv

ction)

ar separate PDM (runoff

re the flood in Figure 4. then inform Emergency

and through

c stations to k links to a itors rainfall restha et al., ng bulletins

haiju, 2013).

cludes self-dback helps of an early cted by the viduals and

244

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 25Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

4 Flood forecasting modelling

4.1 Flood forecasting modelling frame work Flood forecasting models vary in complexity from simple gauge-to-gauge correlations to highly automated integrated catchment flood forecasting models. The degree of automation and sophistication should be considered based on the needs for a particular location and a particular community. While the type of forecast system will be typically based on existing hydrometric networks or the enhancement or development of new networks, the system will need to be achievable and affordable. The type of flood forecasting system will therefore depend on:

● Available data ● Basin characteristics/complexity ● Accuracy and reliability required ● Lead time requirements ● Needs of the flood risk communities ● Ability of the operating organisation to routinely operate, maintain and update the models

Catchment aspects that affect the magnitude and timing of floods are wide and varied. These might include:

● Degree of catchment urbanisation ● Presence of reservoirs and flood storage/attenuation ● Quality of the ratings at gauging stations ● Impact of tributary and ungauged catchments ● Impact of backwater effects, confluences and tidal locations ● Seasonality of rainfall ● Upland areas and snowmelt considerations ● Influence of groundwater

An operational flood warning and forecasting system typically uses some form of hydrological and hydraulic modelling to provide sufficient lead time to avoid loss of life and to allow flood defence measures to be operated. Forecast models are at the heart of reliable operational flood warning systems.

The Mohana-Khutiya basin’s runoff is primarily rain-fed. The upstream catchment in the Chure hills is below 3,000m amsl, and thus, runoff mainly originates from rainfall with a minor component from groundwater (base flow); snow-fed runoff normally should not need to be considered for catchments below 3,000m (Putkonen, 2004).

In the mathematical modelling system for flood forecasting in Mohana-Khutiya basin, three major approaches could be used:

1. Gauge-to-gauge correlation: this is one of the simplest forecasting tools, based on correlation between gauged water level at two stations: an upstream base station and a downstream target station. This method uses flood levels at the base station when the flood has actually arrived (i.e. it uses the real time observed flood level) to estimate the future water level at the target stations. The lead time for forecast in the target station is small, a maximum of probably five to six hours in this basin.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 26Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

2. Rainfall-runoff/hydrological modelling: Rainfall-runoff process modelling could be used to improve the forecast accuracy and lead time. The runoff in catchments with elevation below 3,000m (AMSL) can be modelled using a continuous rainfall-runoff model. Since the rivers are perennial in nature, the base flow component in the rainfall-runoff process is adequately represented. These continuous rainfall-runoff models typically require rainfall and PET as input to provide catchment flow as output hydrograph.

3. Channel routing/hydrodynamic modelling: Channel routing is the process which describes the propagation of flood waves along the river. This process can be modelled using hydrologic and/or hydrodynamic modelling. The employment of the model depends upon the need and complexity in the system. 1-d, 2-d and 1-d/2-d linked hydrodynamic models can be developed.

In terms of what types of modelling techniques should be used, the following key factors should be considered:

● Purpose of the study ● Level of complexity for both in-bank flows and out-of-bank flow paths ● Flow controls and structures in the river system ● Flood storages and their representation in the model ● Requirements on the level of model accuracy ● Computational resources available ● Data availability and accuracy ● Availability of time and budget

It is important to use the most appropriate modelling tool for the project rather than merely the tool that is available. Inappropriate tool selection (such as: i) the use of a steady flow model where unsteady flow conditions are prevalent or where storage is important, and ii) use of a 1-d unsteady model, where a 1-d and 2-d linked model is most appropriate, like in a dense urban area) can have significant technical and accuracy implications.

4.2 Objectives of flood forecasting modelling The FFEWS shall deliver the following in each basin:

● Implementation of a web-enabled Windows Application on server for routine operation of FFEWS models, dissemination of forecast and routine update of FFEWS models.

● Integration of the knowledge base from hydrological, hydrodynamic modelling and data analysis.

● Seamless connection to temporal and spatial database. ● Dynamic front end module for modifying model inputs, recalibration, data assimilation and

dissemination of model results. ● Processing of flood forecasting results in GIS environment into maps of flood inundation

extent, depth, arrival time, and duration, with other relevant themes in the background; this will particularly be necessary in the river reaches that use 1-d models;

● In 1-d/2-d linked model reaches; a flood inundation map will be a direct output on the forecasting website and ready for dissemination.

● Design and development of appropriate inundation mapping tools, using appropriate satellite /LiDAR derived from DEM for the critical floodplain to predict inundation.

● Development of a module to generate institutional and community targeted inundation forecasts and alert messages (via SMS) and web-enabled maps.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 27Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Implementation of a standalone user-friendly application, which integrates the flood forecast model with knowledge base of topographic and thematic GIS data and has the capability to run automatically.

● Module for embankment breaching scenarios or structural failure scenarios, in-built in the forecasting tool; magnitude of flooding through breaching can be more devastating than flooding from embankment over-topping

● Design of appropriate format, content and dissemination protocols to accommodate the current practice within DHM or improvement to the DHM practice for the Flood Alert to flood-affected residents, both designated and broad category to mobile phones in the likely-affected areas

● Post evaluation of performance of forecast accuracy with respect to level and timing each year.

● Obtain feedback from stakeholders and incorporate suggestions on dissemination of forecasts and warning.

4.3 Gauge-to-gauge correlation Flood forecasting in Nepal is mainly operational based on level-to-level (also called gauge-to-gauge) correlation across the country, except in a few basins where advanced hydrodynamic modelling is being developed and applied.

The level-to-level forecasting tools use observed water level data at upstream locations (base stations) to forecast water level at downstream target locations. This method is simple and most cost effective, as it only requires water level data in real time at base station and at target station.

Flood forecasting from gauge-to-gauge relationships has the limitation of needing to wait till the flood is observed at the base station upstream of the forecasting stations. Therefore, in the process, the possible lead time from the catchment lag (from rainfall to water level response), to the base station is lost. Such lead time can be easily added by introducing a hydrological model that can transform the observed precipitation into a simulated hydrograph at the base station. In case of combined use of runoff model and gauge-to-gauge correlation, the base station must have a stage-discharge rating curve, so that forecast runoff from the hydrological model could be transformed into forecast water level using the rating curve.

The gauge-to-gauge correlation procedure does not incorporate any addition of flow between the stretches from base station to the forecasting station nor does this approach consider any breach or over-topping of the embankments between the stretches. The method also does not provide any other information on the water surface profile, e.g. cumulative effect of many control structures in the stretch that is very crucial from the embankment safety and out-of-bank flow situation. Another major limitation to the correlation method is the availability of prediction only at selected sites (target stations) and not all along the main river, let alone the tributaries. Flood maps cannot be generated due to the coarseness of the forecast stations as the forecasts will be only at the gauging points along the river which are normally sparse. Thus, in this method, the inundated area and the time, depth and duration of inundation, which are essential for effective flood management, are not provided to relevant agencies and communities.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 28Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

4.4 Hydrological modelling Hydrological, i.e., rainfall-runoff modelling, can be carried out by various modelling software. One essential element is that the rainfall-runoff model should have a proven record of coupling with a river model (hydrodynamic), so that runoff can be applied as inflows to the river model. A coupled hydrological and hydrodynamic model is essential for designing flood forecasting and warning and will allow much higher lead time in forecasting. Some selected hydrological modelling software are discussed below:

● HEC-HMS (coupled to the 1-d river modelling tool HEC-RAS 1-d and 2-d) ● NAM (coupled to the 1-d river modelling tool MIKE11 1-d and MIKE21FM) ● PDM (coupled to the 1-d river modelling tool Flood Modeller Pro, previous name ISIS and

Infoworks ICM Live)

NAM (DHI, 2016) and HEC-HMS (US Army Corps of Engineers, 2017) are deterministic, lumped, conceptual hydrological models, comprised of a set of linked mathematical statements describing, in a simplified quantitative form, the land phase of the hydrological cycle. They mainly simulate surface and sub-surface runoff, and base flow components. The model parameters require calibration against observed runoff. These parameters remain fixed (constant) over time.

The distributed runoff model, e.g., Probability Distributed Model (PDM) from the Centre of Ecology and Hydrology (CEH Wallingford, 2016) accounts for seasonal variation and spatial and temporal effect on parameters (e.g., soil-moisture deficit).

There are many other hydrological modelling systems (e.g., URBS). However, the coupling of a hydrological model with the hydrodynamic model, the capability of automatic parameter adjustment (auto-calibration) and the long-term continuous simulation ability are important when considering selection of tools.

4.5 Routing modelling Normally the upper catchment is simulated without any need for inclusion in hydraulic model. If in the upper hilly catchments in Siwalik hills, the travel of flood waves and their volume and attenuation are found to be important, some of the tributaries in the hills could be considered for routing modelling, e.g. by Muskingum-Cunge flood routing units. The topography could be extracted from DEM for such modelling.

4.6 Hydrodynamic modelling Hydrological and hydrodynamic modelling for flood risk/flood inundation are the key components of a FFEWS. Logically, the hydrological and hydrodynamic flood inundation models are developed first, and then transformed into flood forecasting models.

Hydrological and hydrodynamic flood inundation models for flood forecasting purpose are calibrated and validated over a wide range of flood events, ranging from low flows to extreme high flows. During a monsoon, an extreme flood event can occur; as well a low magnitude flood can also occur. As such a FFEWS model should be ready and applicable for any flow condition. In a hydrodynamic model, accuracy of results, convergence of results (oscillation free results) and stability of model are three key controls. A model, which is suitable and developed for low flows, may generate instability at high flows due large depths in the channel and shallow depths in the floodplain. As the flood forecasting models will be run on real-time, they have to be suitable for all flow conditions. The forecast model run must not crush during monsoon period, e.g., due to model instability.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 29Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Flood inundation models, once calibrated and validated for wide range of flows, will then help to establish the locations for flood warning and to set-up flood alert and warning threshold trigger levels in rivers for out-of-bank flooding, agricultural land flooding and property flooding.

Flood forecasting models operate in real time, and thus, model run-time has to be efficient. To make the model run-time fast, the model will be simplified in reaches away from the area of flood forecasting points. This will be done by increasing the distance between computational nodes maximum permissible limit and by removing other model units, which are less important, e.g., structures which do not create any significant head loss. However, care should be taken as such that the simplified models maintain the same accuracy at the all flood forecasting and calibration points. This process will require several trials so that model accuracy is preserved.

4.7 Modelling software Some widely used flood forecasting modelling tools are presented in Table 10. The modelling software presented are mostly popular and widely used; thus, the list is not exhaustive. A more detailed list is presented in Appendix A.

Table 10: Examples of key modelling software for flood risk and flood forecasting modelling

Model Type Modelling Tools/ Technology Hydrological/rainfall runoff modelling Any lumped conceptual catchment runoff model, e.g,

NAM, HEC-HMS and PDM Hydrodynamic: flood inundation and flood risk modelling

MIKE11, MIKEFLOOD, MIKEURBAN, MIKE21, MIKE GPU MIKE21FM, HEC-RAS, FLOOD modeler Pro (former name ISIS), TUFLOW Classic, TUFLOW FV, TUFLOW GPU, Info-works ICM

Flood forecasting and warning NAM, and MIKE (11, 21FM, URBAN), and HEC-HMS and HEC-RAS, and PDM and Flood Modeller pro and TUFLOW HPC/GPU, and Infoworks ICM-Live etc.

Source: Mott MacDonald

Hydrodynamic modelling techniques are at the core of fluvial flood risk assessment and flood forecasting and warning (WMO, 2011). As a common practice which started in previous decades, hydrodynamic models are often developed and used to simulate the flood water in the river system as well as across the floodplain. They are used to predict the flood depth, water level, velocity, flood extent and even flood hazard level. Generally speaking, the river system is represented using 1-d models as the flow travels in the channel direction when it remains in the river channel, whilst the floodplain is represented using 1-d, quasi 2-d or 2-d models as the flood water spreads in different directions when the water exceeds the river banks. The 1-d river channel and the floodplain models are linked to represent the connection between the river and the floodplain.

It is important to use the most appropriate modelling tool for the project rather than merely the tool that is available. Inappropriate tool selection (such as use of a steady flow model where unsteady flow conditions are prevalent and where storage is important, and use of a 1-d unsteady model, where 1-d and 2-d linked models are most appropriate, such as in a dense urban areas) can have significant technical and accuracy implications for both current and future needs. Several modelling software tools have been tested through benchmarking studies (EA/Defra, 2004 and 2013) and are being employed for flood risk mapping studies around the world (Table 11). In general, software which has not been subject to benchmarking, is not recommended for developing models.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 30Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

A list of benchmarked hydrodynamic modelling software is presented in Table 11. Software on this list are widely used in flood risk assessment and flood forecasting in UK, Australia, USA, Bangladesh, India, Nepal and other countries. The benchmarking research of the software was conducted by Environment Agency and Defra of UK (EA/Defra, 2013). DoI and DHM will have the option to choose from the list. If required, it is recommended to consult recommendations and results of the benchmarking study and choose the suite of software best suited for Nepal.

Table 11: List of benchmarked modelling software Software 1-d 2-d 1-d/2-d Source MIKE11/21 http://www.dhigroup.com/Software/WaterResources.aspx HEC-RAS 1-d (see note)

http://www.hec.usace.army.mil/software

Flood Modeller (previously ISIS)

1 https://www.floodmodeller.com/about/

SOBEK https://www.deltares.nl/en/software/sobek/ InfoWorks ICM https://www.innovyze.com/ JFLOW 2 http://www.jbaconsulting.co.uk

TUFLOW 3 1,4 http://www.tuflow.com

1 Available through ISIS-TUFLOW link; 2 Not fully hydrodynamic (does not solve momentum); 3 Available as ESTRY (provided with TUFLOW); 4 Links to ESTRY; HEC-RAS 2-d and 1-d/2-d linked modelling version of software were released in 2016

Source: Environment Agency (http://evidence.environment-agency.gov.uk/FCERM/en/FluvialDesignGuide/Chapter7.aspx?pagenum=5

4.8 Modelling software comparative list In addition to the list of benchmarked software, a comparative tabular list of widely used other hydrological and hydraulic modelling software is presented in Appendix A. DoI and DHM will have the opportunity to choose a suite of modelling tools from this list. In the list, HEC-RAS is a free software, while most of the other widely used software is licensed software. For 1-d/2-d linked modelling and pure 2-d modelling, some software tools such as MIKE FLOOD, MIKE21 and TUFLOW have an advantage as they have been used for several decades. HEC-RAS 1-d/2-d linked version is a recent release from 2016.

4.9 FFEWS cost consideration A number of cost elements are required to operate a flood forecasting and warning system. The following components may need to be considered as part of a whole life cost appraisal:

● Setting up any new organisational structures, if they do not exist ● Installing, operating and maintaining telemetry hydro-meteorological gauge network and

hydrometric equipment ● Maintaining spatial and temporal databases ● Developing, configuring, running and maintaining (and troubleshooting) forecasting models ● Developing, running and maintaining (troubleshooting) systems for generating and

disseminating flood warnings and flood maps ● Buying computer software and hardware to support the above operations ● Obtaining meteorological forecasts from freely available sources, for example, from GFS,

WRF and IMD weather forecast models ● Staff training (continuous) and running flood exercises ● Raising public awareness of flooding and how to respond to flood warnings

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 31Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

5 Rain gauge network design

5.1 Introduction Rainfall is a main input for rainfall runoff and river flow and flooding models. These models all have different types of requirements of rainfall input data.

Two types of rainfall measurement methods have been proposed:

● Point rainfall measurement from ground based rain gauges ● Areal rainfall measurement from satellite or weather radar; this also forecasts rainfall which

the model needs to read.

5.2 Auto telemetry rain-gauge Automatic telemetered rain gauges are proposed for installation as part of this project. Modem (GSM) is proposed as the telemetry data transfer mechanism.

5.2.1 Description

The rain gauge should record rainfall and transmit the data through telemetry to the dedicated servers and hydrological and forecasting experts at DHM and consultants via e-mails at defined time intervals.

An automated telemetry rain gauge system (Figure 5) should consist of a rain gauge unit, e.g. tipping bucket rain gauge, an in-built data logger, a Modem (GSM) for connecting to internet and transferring data to the server. In addition, it should have the facility to access and download data remotely and preferably be solar powered. The tipping bucket rain gauge is to be mounted on a pipe within a stainless-steel enclosure that houses a data logger, modem and battery. Battery charging is to be done via solar panel. The Modem unit must be loaded with a data enabled SIM card purchased from a phone supplier. The user will need to define the time interval of data transfer.; At the programmed interval, the modem will initialise a communication with the logger and transmit the rainfall data to the server as well as via e-mails to the flood forecasting experts at DHM and to the flood forecasting consultants of this project.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 5: A

Source: www

5.2.2 T

With a tipptransferred

5.2.3 O

The bucket pulse from switch can a

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

A typical aut

w.phametechnolo

Time of obse

ping bucket automaticall

Operation an

tips when peach tip is salso transmit

ration of Priority Rning System: Moh

9 ning System: Moh

o telemetry

ogy.com

ervation

rain gauge,y at 15-minu

nd measure

precipitation, sensed by tht the pulse to

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

rain gauge

, rainfall deutes intervals

ment

e.g. of 0.5mhe reed swito a telemetry

d Risk Managemesin

sin

station

pth is recors, or at a freq

mm (resolutiotch and loggy system.

ent Project, Nepal

rded continuquency agree

on of bucket)ged to a data

uously. Rained with client

), has been ca logger. The

nfall data is t (DHM).

collected. A e dual reed

322

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 33Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

5.2.4 Data transmission, storage and archive

Data should typically be transferred once or twice per day from the gauge station to a central location (server) within DHM for analysis, storage/archive and for FFEWS modelling. Such transfer of data usually increases during times of heightened flood risk. API (application programming interface) shall be developed for access via internet to all gauging stations and for transferring data to the central server. Data could also be transferred to the regional office server where DHM has such facilities. During the model development phase, the modelling team should also have access to the API via internet to download data. In some countries like the UK, the public can download rainfall data using these API.

5.3 Radar rain gauge

5.3.1 Description

Radar rain gauge will not be implemented in this study; however, write-up here is considered for future reference by DHM.

Measuring rainfall by means of radar is not a new technique. The main advantages are that this provides a better spatially distributed measurement than that obtained from point rain gauge alone. Furthermore, the radar rainfall are grid-based outputs, which are becoming more widely used by rainfall runoff models. However, limitations include; measurement accuracy, range, attenuation of signal and calibration, which means that radar measurement does not provide great advantages over ground based rain gauges. Despite these limitations, radar rainfall is useful data which can supplement ground based data if missing from a rain gauge, help correcting suspicious ground based data, and still be used for rainfall runoff modelling where suitable. Ground based rain gauge data will be used to ground truth radar rain gauge. Capital expenditure as well as running costs are high, though there are low cost short range radars.

5.3.2 Specification

In the mountainous terrain in Nepal, precipitation is highly variable both in space and time because of orographic effects and interactions of mountains with wind fields. Moreover, narrow valleys surrounded by high reliefs cannot be effectively monitored by any of the common long-range weather radars because their beam cannot penetrate deep in the valleys due to the shadow effect.

In mountainous regions, gauge network needs to be very dense (Volkman et al., 2010). Thus, to supplement long range radars, X-band short range radar is a good alternative to the common long-range C-band radar for observing precipitation within a valley. It can be installed directly inside the valleys, at lower altitude. Rain gauge networks can be complemented by short range X-band radars. They can provide rainfall estimates with high spatial and temporal resolution and their installation cost is less expensive than C-band radar, allowing the placement of more sensors in order to gain optimised coverage. Examples of such radar are: CASA radar by the Remote Sensing Group (RSG) of Polytechnic of Turin and Local Area Weather Radar (LAWR) by FURUNO, Japan. X-band radar for rainfall estimates should comprise the following specifications:

● Range of 30km to 70km with radar maps produced at 75m to 150m resolution; ● Fitted with series of anti-clutter filters in order to recover, as far as possible, rain signature

even in the presence of clutter; ● Radar maps produced by the X-band mini radar unit are transmitted to the server via in-built

communication network through internet (telemetered);

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 34Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Provided with processing software for radar maps of rainfall intensity; ● Un-coherent, pulsed, one polarisation only, non-Doppler, with a fixed elevation of the

antenna; exclusively devoted to rain measurements and able to produce one rain map in a few seconds;

● Minimum or maintenance-free and possible to be remotely controlled with software adjustments; routine maintenance will be included in the procurement package during the warranty period;

● All the electronic equipment (antenna, radiofrequency unit, data processing unit, communication unit for data transmission and remote control, power unit) are placed inside a radome;

● All software is required to operate in dedicated applications in open source in order to allow greater reliability and flexibility in the configuration and full control of active processes and packages, as well as low costs. Data will be available in commonly used data format.

5.4 Rain gauge network recommended for installation Eleven new auto telemetry rain gauge stations have been proposed for installation in the Mohana-Khutiya basin under this project. The gauges will be tipping bucket auto telemetered using GSM. The Mohana-Khutiya catchment area is 702km2. This will deliver a rain gauge station density of one rain gauge per 64km2. This is similar to that in England where rain gauge density is the highest in Europe (one rain gauge per 60km2; Allot, 2010).

Gauge network density depends on many factors, particularly spatial and temporal resolution of rainfall over a basin and the purpose of the gauge network (e.g. irrigation management, flood forecast etc.). Flood prediction in rural and urban areas requires a dense spatial gauge network, i.e. one gauge between 10 and 100km2, and higher temporal measurement frequency, i.e. between minutes and hours (Berndtsson and Niemczynowicz, 1988). The gauge density may be even higher one gauge per 20 to 45km2 for mountainous areas of Nepal (Lopez et. al., 2015 and Volkman et al., 2010), considering that there is a greater variability of rainfall between the mountains, and the Siwalik and Terai region. However, in view of practicality, management, and with reference to other countries, the implementation of 11 new stations for this basin is considered to be a practical trade-off between cost and benefit. All proposed rain gauges should be stationed near settlements (for better accessibility and maintenance) and have been distributed considering the main channels and their tributaries. Their positions are shown in Table 12 and Figure 6.

Table 12: Proposed new auto telemetric rain gauge stations in Mohana-Khutiya basin Catchment/basin name Rain gauge ID Name of nearest

settlement Station coordinates

Longitude, E (deg) Latitude, N (deg) Mohana-Godawari-Monohara (MGM)

MGM_Rain_01 Tudela 80.517 28.886 MGM_Rain_02 Lalpur 80.605 28.835 MGM_Rain_03 Chunepani 80.577 28.875 MGM_Rain_04 Bela 80.622 28.758 MGM_Rain_05 Baluwaphanta 80.493 28.809 MGM_Rain_06 Dubki 80.602 28.933

Khutiya-Shivaganga (KS) KH_Rain_01 Koldanda 80.647 28.911 KH_Rain_02 Katauje 80.679 28.978 KH_Rain_03 Urmi 80.680 28.686

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 35Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Catchment/basin name Rain gauge ID Name of nearest ttl t

Station coordinates KH_Rain_04 Shiva Ganga 80.720 28.798 KH_Rain_05 Garbha Durbar 80.762 28.889

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 6: Estations in

Source: Mott

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Existing metMohana-Kh

MacDonald

ration of Priority Rning System: Moh

9 ning System: Moh

teorologicalhutiya basin

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

l stations ann

d Risk Managemesin

sin

nd propose

ent Project, Nepal

d new auto telemetric rain gauge

366

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 37Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

5.5 Budget for proposed rain gauge network installation The budget estimates for the proposed rain gauge network are considered for ground-based telemetric stations. The budget includes procurement, installation, testing, calibration, monitoring, and operation and maintenance for 3 years. Budgets are shown in Table 13.

Table 13: Budget for proposed rain gauge networking in Mohana-Khutiya basin Hydrometeorological data network

Mohana-Khutiya budget (US$)

No. of station

No. of measurement

Capital cost/ measurement

cost

Unit cost Maintenance cost: 3 years

Total cost

Ground-based tipping bucket auto telemetry

11 - 55,000 5,000 6,000 61,000

Total 11 55,000 5,000 6,000 61,000 Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 38Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

6 Hydrometric network design

6.1 Water level gauge network

6.1.1 Description

Flood risk assessment and flood forecasting rely heavily on hydrometric data. Key hydrometric data requirements include river levels, river flow and ground water level.

Any of the following water level monitoring methods are proposed for this project; each approach is telemetric monitoring.

● Measurement by float in a stilling well ● Water Level Radar Sensor ● Water Level Bubbler Sensor

Data will be automatically transferred to the dataset server by GSM telemetry.

Depending on site conditions of the gauging station, one of the monitoring approaches shall be selected. At each telemetric water level station (whether stilling well, radar or bubbler), there will be a manual water level staff gauge. This manual gauge shall be maintained by CBDRM Committee members and can be used by them in the event of a flood alert to communities. This staff gauge should have different distinct colour painting for water levels in flood alert zone, in flood warning zone and in danger level zone (DHM uses such colour level staff gauge).

The water level stations will be included as a forecast point in the FFEWS.

Specification of telemetry kit

The telemetry gauging station should allow:

● Remote monitoring of river levels; ● Transmission of alarms if level rises above user defined thresholds; ● Viewing of historical level data via simple web GUI; ● Transfer of data via API for use in applications and websites; ● Preference for solar powered; which will also include a back power system (battery)

The telemetry system should contain:

● Level Sensor having different options to suit depth of river from shallow depth (0.2 to 0.5m) to several metres of depth (>20m); accuracy 0.5% of range;

● Data logger ● Solar Powered Telemetry Unit with GSM module and antenna built in; ● Sim Card for the warranty period of 5-years, multi network; ● Readings every 15 minutes. ● Transmission of web-based data and alarms by email to designated professionals (DoI/DHM

to provide list of emails of designated persons and professionals, and thresholds for high water levels).

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 39Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

6.1.2 Time of observation

Stationarity of record, temporal resolution and the overall accuracy of water level data shall be considered. All water level records will be taken at 15-minute intervals in to the data logger, and are to be transmitted via telemetry to the central dataset server at DHM and to the hydrological and database experts at DHM.

6.1.3 Operation, measurement and maintenance

Backup and main recorders should be securely mounted and regularly visited and serviced, at least once every month (or more if required). It should be ensured that the pulleys are operating freely, and the float tape or wire sits properly on the drive pulley. Logbooks should be maintained, and calibration of the level sensor should be checked and reset if required.

Sites with known sediment problems shall be carefully checked at each visit, and if there are any indications of a siltation problem, the stilling well must be flushed as soon as possible or proper flow connection to the sensor must be maintained. The Mohana-Khutiya basin, particularly the Mohana and Khutiya rivers at downstream reaches, carries highly sediment laden flow, and this routine silt management will be required.

6.1.4 Data transmission, storage and archive

Data will typically be transferred once or twice per day to a central location dataset server within DHM for analysis and storage/archive. Such transfer of data usually increases during times of heightened flood risk. In periods of heightened flood risk, even hourly transfer could be required considering the flashy nature of a storm event in the basin. Data could also be transferred from the gauge stations and/or from central dataset server to DHM’s three basin Offices (Karnali Basin Office in Nepalgunj, Narayani Basin Office in Narayanghat and Kosi Basin Office in Biratnagar), and.to the regional offices where DHM has data storing facilities. During the model development phase, the modelling team should also have access to the real time data.

6.2 Discharge measurement stations

6.2.1 Description

Manual discharge measurement stations should be capable of measuring low to moderate flows while the water remains in the channel. For measuring flow beyond certain thresholds, especially when the water level is very high, exceeding the river banks and flowing across the floodplain, the river flows are normally derived from the relationship of stage (level) with discharge, called a stage-discharge relationship or rating curve.

6.2.2 Discharge measurement equipment

Discharge measurement at all stations shall be carried out for a wide range of flows, from low flow to high flows. Discharge measurement is to be carried out fortnightly from mid-May to mid-October at each station. A combination of the following discharge measurement methods can be used:

● Manual measurements using current meter (propeller current meter) during low flows except at the cableway station

● Velocity-depth measurements with ADCP (Acoustic Doppler Current profiler) during medium to high flows, except at the cableway station

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 40Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Cableway discharge measurements using propeller current meter (ADCP may also be mounted if required) at the uppermost hill station, where velocity measurements at medium to high flows are not advisable without cableway owing to safety reasons.

In manual measurement during low flows, cross-section/depth will be measured using a graduated pole, often referred as manual sounding. When depth is increasing and ADCP (with echo-sounder and DGPS) could be used, all discharge measurements shall be carried out using ADCP. While ADCP will provide velocity scatter, echo-sounder will provide depth and DGPS will provide horizontal positioning. The measurements will be carried out using local engine boat/inflatable boat The ADCP discharge measurement could be un-manned. The ARC-Boat (http://www.ceehydrosystems.com/products/unmanned-survey-vessels/arc-boat/) is designed to make safe unmanned discharge measurements in rivers and streams using acoustic ADCPs. The hull design minimises air entrainment for optimum ADCP data quality. With a maximum speed of 4.5m/s (15fps), even high velocity flood stage measurements may be completed. Effects of magnetic interference from the vehicle’s electrical systems are carefully managed to minimise induced compass deflection – critical to obtaining good discharge measurements.

ARC-Boat ADCP measurement at the cableway station could also be considered to replace Cableway depending on the magnitude of velocity and safe operation of measurement.

The budget (unit price) for each discharge measurement at fortnightly intervals is inclusive of all cost elements (2 days input from an equipment engineer including support staff, boats and accessories, travel cost to site, calculation of discharge from raw data, and preparation of report). Discharge measurement equipment (ADCP, DGPS and echosounder) will be provided for this project. Separate budget has been considered for this.

6.2.3 Cableway flow measurement

Slack-line cableways are commonly used for carrying out flow gauging on relatively small rivers and streams. This is particularly useful in rivers in mountainous regions with steep slopes and high velocities. Velocities are measured from a velocity traverse set at a series of fixed depths, and then multiplied with cross-section areas providing total discharge through the river section. Such discharges are useful for developing stage-discharge rating curves. Cableway stations are not suitable where flow goes out of bank and cross-section width is high, e.g., river section in the Terai region.

The components of a slack-line cableway comprise: a static ropeway, suspended between anchor ends; a traveller; a horizontal positioning mechanism; and a lifting mechanism. In operation, the traveller runs on the ropeway and functions as an unmanned ‘cable car’. All operations are carried out from the bank. The horizontal position of the traveller is controlled by means of a manual winch that feeds a line across the span and over a pulley mounted on the far side and back to the traveller. A separate line from a gauging reel feeds out to the traveller, runs over a pulley mounted on the traveller and connects to a current meter and counterweight, thereby suspending these from the traveller. The gauging reel controls the vertical position of the current meter. Depth will be taken by the sounding reel cable at pre-defined vertical positions by lowering the cable to the river bed; then depth data will be converted to obtain cross-section (river bed level with respect to masl) using the water level gauge reading. As ADCP, DGPS and echosounder will also be available, these set of equipment could also be occasionally mounted to the cableway for depth and velocity measurement, if the Equipment Engineer find it practical.

DHM operates cableway discharge measurement in West Rapti at Kusum. A similar cableway should be established at a station at Malakheti on the Mohana river Standard specifications for

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

establishingUSDA, 2000

Figure 7: Sand currencableway ctailhold on

Source: USDA

6.2.4 E

Acoustic Domeasuremeequipment smeasuremeone set of esix basins, tno equipmeMawa-Ratu

There will bbudget has will be used

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

g cableway c0. A schema

Schematic ont meter: (cable, (4) sfar side

A, 2000

Equipment b

oppler currenent) and DGset only for

ents at four sequipment fothree equipment set has wa, and Eas

be a warrant been consi

d extensively

ration of Priority Rning System: Moh

9 ning System: Moh

could be fouatic of cablew

of a cablew(1) near poounding ree

budget for d

nt profiler forPS (positionthis basin isstations, (foror this basin ment set havbeen consid

st Rapti and W

ty period of dered for op, such cost is

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

und in manuway system d

way dischargost with puel cable, (5

discharge m

r discharge mn measuremes presented itnightly at eaalone, which

ve been propdered. BakraWest Rapti w

3 years for peration, mas necessary.

d Risk Managemesin

sin

uals, e.g. NEdischarge me

ge measureulley drive 5) traveling

easurement

measuremenent), have bin Table 14ach station)h is located

posed; For Eaha basin wwill use equip

DGPS, Echaintenance a.

ent Project, Nepal

EMS Manuaeasurement i

ement systehousing, (2block, (6) c

t

t, supported been proposebelow. Therin this basinin the far we

East Rapti, Bwill use the epment availa

ho-sounder and servicing

l (NEMS, 20is shown in F

em with sou2) soundingcurrent met

by echosoued. The bude will be ext

n; thus there estern part ofakraha and W

equipment prble within DH

and ADCP. . As the equ

016) and in Figure 7.

unding reelg reel, (3)

ter, and (7)

under (depth dget for one tensive flow is need for f Nepal. For West Rapti, roposed for HM.

However, a uipment set

41

1

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 42Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Table 14: Discharge measurement equipment / station budget in Mohana-Khutiya basin Hydrometric equipment and installation

Mohana-Khutiya budget (US$) Capital cost Operation and

maintenance (total for 3 years)

Total cost

DGPS 25,000 1,250 26,250 Echo-sounder 25,000 1,000 26,000 ADCP 35,000 1,000 36,000 Cable way discharge station: construction cost

95,000 - 95,000

Total 180,000 3,250 183,250 Source: Mott MacDonald

6.3 Hydrometric gauge recommended for installation Five hydrometric gauging stations in the Mohana-Godawari and four in the Khutiya-Shivaganaga catchments have been proposed (see Table 15 and Figure 8). All stations are easily accessible and are located near a settlement. All stations have access to GSM (DHM, 2018).

The locations have been chosen carefully, which will allow calibration of runoff from the hydrological model and calibration of river levels in the hydrodynamic model. In the Mohana-Godawari catchment, there will be two water level (WL) and discharge (Q) stations (both parameters at the same station), as well as three WL only (no Q) stations. In the Khutiya-Shivaganga catchment, there will be two WL and Q stations and two WL only (no Q) stations.

Stage-discharge rating curves will be developed at all four discharge stations for generating continuous discharge data and for calibration and validation of hydrological and hydrodynamic model.

The four flow stations will allow calibration and validation of runoff from: i) Godawari river catchment alone, ii) combined runoff from Godawari, Monohara and Mohana catchment at their confluence point and, iii) total runoff from the Mohana catchment at immediately upstream of the confluence with the Khutiya.

Similarly, WL and Q stations in the Khutiya-Shivaganga catchment will allow calibration and validation of runoff from: i) Khutiya catchment at the east-west highway point and, ii) from combined runoff from Khutiya-Shivaganga catchment at downstream of their confluence point. The distribution of WL and Q stations will allow a very dense network for calibration of the hydrodynamic model; such a dense network should be considered essential due to flood forecasting and the relatively steep slope in the upper basin and the mild slope in the lower basin (Mott MacDonald, July 2018).

At all nine hydrometric stations, bed material samples shall be collected, only once; and at four discharge stations, suspended sediment sample shall be collected. Bed material shall be collected from mid-channel, and from bed near the banks. Concentration shall be collected from three positions as minimum over the cross-section. If river depth is high (>3m), we recommend collecting concentration over several vertical positions (e.g. 02d, 0.6d and 0.8d; d is total water depth) at each position. If the depth is shallow (<3m), only one sample at 0.5d is recommended.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 43Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Table 15: Proposed water level and discharge stations in Mohan-Khutiya basin River name

Water level and discharge gauge ID

Gauge type Name of nearest settlement

Station coordinates Longitude, E

(deg) Latitude, N

(deg) Mohana M_G_01 Water level Dhakhuwa 80.733 28.883

M_G_02 Water level Attariya 80.521 28.833 M_GD_03 Water level

and discharge Geta 80.503 28.846 M_GD_04 Water level

and discharge Jugeda Katan 80.543 28.813 M_G_05 Water level Chatakpur 80.540 28.765

Khutiya

K_G_01 Water level Bandagada 80.616 28.645 K_GD_02 Water level

and discharge Syauli Bazar 80.555 28.720 K_G_03 Water level Murkatti 80.649 28.849 K_GD_04 Water level

and discharge Beli 80.635 28.793

Note: M_G: stands for Gauge only (water level) station in Mohana River; M_GD stands for gauge and discharge; same applies to the Khutiya river

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 8: EKhutiya ba

Source: Mott

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Existing watsin

MacDonald

ration of Priority Rning System: Moh

9 ning System: Moh

ter level an

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

nd discharge

d Risk Managemesin

sin

e station an

ent Project, Nepal

nd proposed stations iin Mohan -

444

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 45Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

6.4 Hydrometric gauging network budget The budget for the hydrometric gauge network is presented in Table 16. The budget includes procurement, installation, testing, measurement, calibration, monitoring, and operation and maintenance for 3 years.

Discharge measurements will be carried out fortnightly from mid-May to mid-October for three years. The measurements shall also include bed material and sediment concentration collection; this will be 10 measurements per year, 30 in three years and a total of 120 measurements for the four discharge stations. Operation and maintenance cost is $2,000 per basin for all stations per year; this involves routine site visits, repair and maintenance if required. Sediment measurement will have multiple benefits including supporting working design of river training works and also in FFEWS, for example, improving the stage-discharge rating curve; changes in sediment load shall indicate the need for updating the rating curve.

Table 16: Water level and discharge gauge network budget in Mohana-Khutiya basin Hydro-meteorological data network

Mohana-Khutiya budget (US$)

No. of stations

No. of measurements

Capital cost/ measurement

cost

Unit cost

Operation & Maintenance cost: 3 years

Total cost

Discharge 4 120 520,000 4,333 6,000 526,000 Water level 5 - 35,000 7,000 6,000 41,000 Note: a) Discharge measurement to be carried out fortnightly from mid-May to mid-October; this will be 10 measurements per year, 30 in 3 years and total 120 measurements in 4 stations; b) Operation and maintenance cost is $2,000 per basin for all station per year; this involves routine site visits, repair and maintenance; c) Discharge measurement cost is a continuous expenditure, like model development cost (and should be considered similar to capital cost); it includes cost for all skilled human resources and the logistics required

Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 46Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

7 Topographic and asset survey

7.1 Topographic survey Current river cross-section and topographic data will be required to develop hydrodynamic (1d, 2d and 1d/2d linked model). Latest data will be surveyed during the FFEWS model development period. Existing cross-sectional data in this basin were collected in 2014 (during pre-feasibility, Package 3). Thus, this data is considered old for the dynamic rivers in Nepal; there were additional, but limited, cross-section survey during feasibility study (this study, Package 7). In UK, where rivers are very stable, the Environment Agency (responsible for flood forecasting), updates their models if topography is more than 5 years old. Thus, fresh cross-sectional survey, in higher spatial resolution (than in Package 3 and 7) should be surveyed. This will improve model calibration and validation, forecast accuracy, and inundation maps.

On steep slopes and in meandering/braided rivers, cross-sections between 200m and 500m intervals are essential (HEC-RAS, Users’ Manual, Version 4.1, Figure 8-34) for accurate model calibration, validation and forecast accuracy and for inundation mapping. Cross-sections, on average at 380m intervals, are proposed. Rivers in Nepal, typically, have very steep slopes. Although the Terai region is referred to as a flat region with mild slopes, the slopes are still many times steeper than those in low-lying plains. Moreover, rivers are braided and meandering in nature due to high sediment load. As a result, cross-sections in the dense intervals proposed will be useful.

Topographic survey will include river section, any existing structures and flood embankment profile. Survey will have to be done in Mohana, Godawari, Monohara, Khutiya and in Shivaganga river. In 141km, 367 cross-sections will have to be surveyed.

All cross-sections will cover the river, bank to bank and will be extended into the floodplain to sufficiently high ground of highest historic flood water mark. Horizontal projection for survey will be WGS 84 / UTM zone 45N. Ground elevation will be relative to metre above mean sea level (masl) for controlling vertical datum. All cross-sections shall be connected to Nepal National permanent bench mark for horizontal and vertical datum control. Temporary bench marks (TBM) shall also be established for cross-verification of data. All cross-sections, in a basin, shall be surveyed during dry season, prior to or after monsoon so that cross-sections are stable without much morphological change. All cross-section shall be surveyed in one season. This should not be done that half volume of total survey prior to monsoon, and the rest after the monsoon

Ground elevation (vertical position) in cross-section/topographic survey shall be accurate better than 20mm in case of level survey and be better than 50mm in case of echosounder depth survey. Horizontal position accuracy should be better than 1m. There should be enough vertical points to sufficiently represent a cross-section shape, dense points about 0.5 to 1m apart at scoured part of the cross-section, and less points over shallow sand bars, about 1 to 5m interval.

Bank/defences survey crest levels are to be provided at intervals that will adequately describe the river bank (typically every 10m).

The following deliverables are required:

Channel sections, longitudinal sections and structure elevation drawings in AutoCAD DWG and PDF format;

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 47Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Channel section data in the following formats: o Csv or txt or dat file o AutoCAD; and o Section chainage, X (easting), Y (northing), Z(elevation), distance to next

section Width tables for each bridge opening surveyed Cross section location plan in AutoCAD DWG, Shapefile and PDF format; Bank/defence survey location plan in AutoCAD DWG, ESRI shapefile, spreadsheet .csv

or Excel 2007 format and PDF format Site photographs: at least 3 photographs per cross-section, taken one looking

upstream, one looking downstream and one with another good angle

For topographic survey, no survey equipment has been proposed for purchase; survey will be done through outsourcing.

7.2 Survey budget Budget has been decided based on density (no. of cross-sections) of survey, and per cross-section (see Table 17). Topographic survey shall be outsourced and thus no equipment for such survey has been proposed.

Table 17: Topographic survey budget for Mohana-Khutiya basin Topographic cross-section survey

Mohana - Khutiya survey budget (US$) Length of survey

(km) No. of XS Total cost Unit

cost Mohana 56 147 29,400 200 Godawari 12 30 6,000 200 Monohara 10 26 5,200 200 Khutiya 42 109 21,800 200 Shivaganga 21 55 11,000 200 Total 141 367 73,400 -

Source: Mott MacDonald

7.3 Satellite imagery Budget for purchasing high resolution satellite imageries has been included (Table 18) for Mohana and Khutiya basin for lower catchment only in Terai to provide DEM to 1-d and 2-d model development and flood inundation map preparation. High resolution (50cm) Pleiades imageries have been proposed for purchasing. DHM informed that they have already used this imagery in their FFEWS modelling.

Table 18: Satellite imagery purchase budget for Mohana-Khutiya basin High resolution (50cm) satellite imagery

area (km2) Total cost Unit cost (USD for 1

sq.km) Mohana basin in Terai 232 11,600 50 Khutiya basin in Terai 214 10,700 50 Total 446 22,300 -

Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 48Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

8 Flood forecasting model development

8.1 Mathematical modelling It has been proposed to develop a suite of flood forecasting modelling tools. This includes a simple tool for use in combination with the most advanced hydraulic models being employed in countries like Australia and UK. Hydrological model is essential (input) component for each of the forecasting tools. The simple tool, gauge to gauge correlation, could be operational within eight months, or as soon as some hydrometric data becomes available from the new proposed hydrometric and rain gauge network. Over a period of three years, advanced hydraulic models will be developed, calibrated, validated and will be made operational as more data becomes available. A conceptual diagram of different components of the models and link between them are presented in Figure 9. The following forecasting tools have been proposed:

● Gauge-to-gauge correlation: the simplest and cheapest method, fast to develop, and thus could be operational within 7 to 8 months from the inception of the project. However, it has a very short lead time (2 to 5 hours) and is not appropriate in upper steep slope river reaches, as flood wave propagates fast in those reaches and correlation of two gauges is not strong due to presence of pools and riffles.

● Combined rainfall-runoff and gauge-to-gauge correlation: with the addition of a runoff model, the forecast lead time could be extended up to 72 hours. However, this requires a stage-discharge rating curve at each gauging station; such rating curve is difficult to develop for out-of-bank flow conditions without a hydraulic model; as soon the hydraulic model will be developed (within 9 to 10 month), such rating curve will be available from the model

● 1-d model: this tool will be developed for the entire river system in the Terai and is appropriate for flood forecasting which is the same model type used in Bangladesh.

● 1-d/2-d linked model: this will be the final delivery around month 24. A pure 2-d model will be developed and transformed into a 1-d/2-d linked model with the linkage to the 1d model.

Mott MacDonaldFlood Forecasti

383877 | REP | Flood Forecasti

Figure 9: Arequireme

Source: Mott

d | WRPPF: Prepaing and Early War

0040 | 4 April 201ing and Early War

A conceptuent of data

t MacDonald

aration of Priority rning System: Moh

19 rning System: Moh

al flow diag

River Basins Floohana – Khutiya Ba

hana – Khutiya Ba

gram of diff

od Risk Managemasin

asin

ferent comp

ent Project, Nepa

ponents of m

al

models, their links and

4

d

9

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 50Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

8.2 Rationale for different forecasting approaches The four approaches described above are inter-linked and essential and/or complementing components to the final deliverable/ flood forecasting and early warning system (FFEWS) model, i.e., the 1-d/2-d linked FFEWS model. The rationale, advantages and disadvantages of each approach are described below:

● Gauge-to-gauge correlation: the simplest and cheapest method. This could be an option to use as a quick forecasting tool. It can generate new knowledge, to be translated into the final deliverables (1-d model and 1-d/2-d linked model). Advantages will be that flood forecasting components of CBDRM could be operational earlier and potential areas of uncertainty in flood level forecast could be identified. DHM is using this method in many of their river basins, e.g. in Karnali. This tool and expertise from DHM could readily be used in this basin with some nominal input from international consultant; as the tool has to be customised for new basin, need for minor changes in code and parameters may be required and thus international consultant’s input is considered. There will be a deployment time in all five basins for the new hydro-meteorological data to become available, so this work is a good utilisation of the waiting time as it generates the opportunity for transferring early knowledge to the final product.

● Rainfall runoff model is an important input to all other components: a) gauge-to-gauge correlation, b) 1-d river model, c) pure 2-d model and d) 1-d/2-d linked model. Combining the rainfall model with gauge-to-gauge correlation will increase the lead time (as in the rainfall forecast) up to 24, 48 and 72 hours. However, at the forecasting points, the discharge vs water level rating curve shall be required so that forecasted runoff can be converted to the water level using the rating curve. The rainfall runoff model provides inflows from the upper catchment and distributed inflows from intermediate catchments to the 1-d, 2-d and 1-d/2-d linked model.

● The 1-d model, as a standalone tool, can be applied as a forecasting tool once it is ready. Without the 1-d model, a linked 1-d/2-d model (which is proposed as a final deliverable) cannot be developed. Therefore, it is proposed to develop a 1-d model as forecasting tool as soon topography has been surveyed. In any case, for certain reaches of the river, there will only be a 1-d model, as a 1-d/2-d linked model is not feasible for the entire reach of the river due to higher model run time, and instability in 1d/2d model in steeper reaches. This tool will also give useful feedback on forecasting performance, which then could be translated into the final deliverable. In summary, 1-d model development is not a duplicating tool; it is an essential pre-requisite. Should DoI and ADB decide not to take forward 1-d/2-d linked modelling, then a 1-d model will be the final product. This is the tool which DHM operate in the Bagmati, Koshi and West Rapti basins. The advantage of a 1-d model is that it runs efficiently, which is a key requirement for real time forecasting. However, a 1-d model does not have direct map output for flood risk or hazard and these would require separate and customised GIS development, e.g. as practiced by forecast model in Bangladesh (http://ffwc.gov.bd/). Such a GIS tool is under development within DHM. It will need to be developed in this project for the 1-d only model reaches of the river

● A 1-d/2-d linked model is the final deliverable; such FFEWS models are already in operation in countries like Australia, New Zealand, Malaysia and UK (Syme, 2007; Huxley, 2016). Therefore, developing the next generation of the FFEWS tool would ensure that by the time the project is complete, Nepal won’t fall behind on national standards. The 1-d/2-d linked model can forecast flood levels with better accuracy (as it is linked to 2-d floodplain model). Further, flood risk and hazard maps are direct outputs from such modelling. However, run-time is longer than for the 1-d model; it requires more accurate DEM and

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 51Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

therefore to develop it for all reaches of the river. For selected river reaches, where such modelling will be useful, like in the Lower Terai, this tool shall be developed. To overcome run-time issues for real time forecasting, GPU (graphical processing unit) or HPC (heavily parallelised computing) versions of modelling software shall be used.

In several meetings with DHM, the consultant has proposed the development of a similar FFEWS model, with regards to modelling tools and types of models. We have proposed the same type of advanced 1-d model development for FFEWS, which DHM is presently operating in three different basins (West Rapti, Bagmati and Koshi). The same (or similar) modelling software (e.g. MIKE11 and HEC-RAS), for both hydrological and hydrodynamic modelling, has been recommended (in parallel with other software), thus giving DHM wider options to choose from.

8.3 Gauge-to-gauge correlation The development of a forecasting tool using a gauge-to-gauge correlation has been proposed for the main Mohana and Khutiya rivers. Other tributaries to these two rivers have not been considered. In Mohana, 57km river length, and in Khutiya, 32km river length have been considered for gauge to gauge correlation (Table 19 and Figure 10). Very upstream and the very downstream water level gauges in both the rivers will be used for correlation.

River reaches with relatively mild slope have been considered. The above proposed reach lengths could be changed (decreased or increased) during the development phase after analysing physical data (water level, DEM and cross-sections) when they become available. Reaches with steep slopes have not been considered as in steep slope rivers, a downstream gauge has minimum influence to an upstream gauge as the river’s flow regime is mainly flow dominated from the upstream (due to high Froude1 number, i.e., velocity is relatively high, see Mott MacDonald, 2018b). Further, the benefit of gauge-to-gauge correlation forecasting is very limited in steep slope reaches.

There are existing gauge-to-gauge correlation tools within DHM which are operational in many basins (e.g. Karnali) which will be used. Thus, the new tool can be developed fast and with minimum cost; it will mainly involve analysis and feeding in of the new hydrometric data.

This tool shall be maintained in parallel to advanced 1-d and 1-d/2-d linked models.

Table 19: Proposed river reaches for development of gauge-to-gauge correlation flood forecasting model

River Reach ID Reach Characteristics Channel length

(km)

Slope (%)

FF model type

Mohana 1 Hill 4.97 14.1 - 2 Fan 8.37 1.1 - 3 Peripheral fan (Godawari confluence) 13.87 0.1 gauge-to-gauge 4 Flood plain, meander (India border) 21.29 0.1 gauge-to-gauge 5 Flood plain, meander (India border-

Khutiya) 21.34 0.1 gauge-to-gauge

Godawari 1 Hill 9.26 13.4 - 2 Braided 6.97 1.2 - 3 Flood plain, meander (Mohana 6.52 0.2 -

1 Froude number (Fr) = u/(gh)0.5 where u is flow velocity, h is water depth and g is acceleration due to gravity

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 52Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

River Reach ID Reach Characteristics Channel length

(km)

Slope (%)

FF model type

Confluence) Monohara 1 Hill 6.29 16.2 -

2 Braided 4.80 0.9 - 3 Flood plain, meander (Mohana

confluence) 8.59 0.1 -

Khutiya 1 Hill 19.91 8.3 - 2 Braided 6.51 1.1 gauge-to-gauge 3 Flood plain, meander (Shivaganga

confluence) 15.46 0.2 gauge-to-gauge

4 Meander (Mohana confluence (Nepal-India border)

9.77 0.1 gauge-to-gauge

Shiva Ganga

1 Hill 10.92 14.7 - 2 Braided 3.78 1.4 - 3 Flood, plain, meander Shiva Ganga

(Khutiya) 19.55 0.2 -

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 10: flood forec

Source: Mott

8.4 HydThis sectioninvolves var

● Selection● Identifica

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Proposed rcasting mod

MacDonald

drological mn describes rious system

n of the appration of initia

ration of Priority Rning System: Moh

9 ning System: Moh

river reacheel in Mohan

modelling developmen

matic analyse

ropriate precl estimates o

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

es for develon -Khutiya ba

t of the hydrs:

cipitation-runoof the model

d Risk Managemesin

sin

opment of gasin

rological mo

off module; parameters

ent Project, Nepal

gauge-to-ga

odel. The dev

auge (G2G)

velopment o

correlation

of the model

533

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 54Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Selection of the calibration and validation period covering full range of wet and dry events ● Calibration of the model ● Validation of the model ● Sensitivity analysis of the model parameters required to upgrade the model in future.

The runoff model domain shall cover the entire basin area (702km2) from the Chure Hills to the lower Terai up to the Indian border in the south.

8.4.1 Review of existing data and models

● Use the existing hydrological model (Feasibility Study model, Package 7) to improve sub-catchment delineation, parameter (calibration) improvement; model parameterisation using local/donor data; probably modelling tools include HEC-HMS and NAM; DHM is experienced with both tools

● Review available rainfall data from DHM and other secondary sources to provide a representative areal average rainfall

● Cross-check tipping bucket rain gauge with storage gauge data, including double-mass analysis and / or cumulative-mass time series plots

● Review data against general meteorological records, in particular to identify periods where there may have been snow / snow melt

● Comparison of rainfall radar totals with rain gauge information, investigate spatial and temporal distribution of rainfall for selected calibration and validation events

● Provide a commentary on the suitability of weather radar information to supplement gauge rainfall for rainfall-runoff model development

● Assess the availability of data, and the uncertainties in the accuracy of the data and what effect this could have on the reliability and accuracy of model outputs

● Selection of calibration period using long records of meteorological data (minimum of three to five years). A long period of calibration data is essential due to sensitivity of the runoff model to the initial condition

● Selection of validation period using long records of meteorological data (minimum of three to five years); long period of validation data is essential due to sensitivity of the runoff model to the initial condition.

8.4.2 Catchment delineation

The basin shall be divided into smaller hydrologic sub-catchments to define catchment topology according to geomorphologic homogeneity. The following will be considered while delineating the sub-catchments:

● Topography, DEM based on satellite and their resolution, e.g. SRTM 30m, Cartosat-1 or high resolution Pléiades imageries

● Drainage network based on satellite imageries ● Changes in sub-catchment response, key tributaries/confluences, flood storage reservoirs

etc. ● Catchment delineation shall be verified including use of surface water sewer data in

urbanised sub-catchments ● Permanent snowline and snow cover ● Soil/sediment and land use data ● Urbanisation extents: land use in urban areas

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 55Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Embankment layout and location of flood control sluices and structures ● Location of hydrological monitoring sites ● Average annual precipitation over the basin with reasonably good resolution.

8.4.3 Hydrological input: Rainfall, temperature and PET

Key input data for hydrological modelling are:

● Rainfall ● Temperature ● PET

Initially the model development shall start with data available at DHM, if quality data within the basin is unavailable, the nearest met station data shall be used. Gridded rainfall data could be used from satellite-based sources (IMD, APH etc.) where available.

As soon as data from new proposed gauging stations become available (should be available after the first monsoon during the development phase of the project), they shall be used for improving both calibration and validation of the hydrological model.

8.4.4 Bias correction

Gridded rainfall data cannot be directly used for runoff modelling. Bias correction on historical precipitation series shall be developed for using such data in the FFEWS model.

TRMM-P has the highest possible temporal resolution (three hours) of all gridded precipitation data sources, it is freely available, there is a long record of historical archives, and it is probably the most accurate Satellite-based Precipitation Estimate (SPE) available globally. However, SPE requires bias correction. Region-specific bias in TRMM-P exists and the bias increases with smaller spatial scales, higher temporal resolution and higher magnitude of precipitation values. It has also been observed that capacity of TRMM-P in resolving orographic precipitation in Himalaya is limited. Thereby, raw TRMM-P has to be bias corrected (eQM) before it can be used for rainfall-runoff modelling.

8.4.5 Calibration

The calibration period shall cover hydrological data of at least three hydrological years, but preferably five or more. Availability of data, particularly rainfall, from different sources has been shown in Section 2.8, Table 9. The years shall be selected judiciously so that observed rainfall and discharge are available at most observation stations, if not at all stations. Missing, inconsistent and erroneous data, and non-availability of data are generally an issue in data collected from existing sources and secondary sources; examples of such data are point and gridded rainfall, temperature, PET, observed discharges. Therefore, these factors shall be considered while selecting the calibration period. In the hydrological model, as the initial condition is sensitive, the first 3 to 6 months of simulation period shall be considered as initialisation time, and thus shall be ignored while using the model runoff to hydrological model. One key factor of considering longer simulation period for validation (or calibration) is due to this initial condition effect; other factors are to cover wide range of flow condition, which will allow low to high range of flows to the hydrodynamic model. Calibration of runoff model against log record helps finalising catchment parameters, where most hydrological modelling tools, e.g. HEC-HMS, NAM, are conceptual model, and thus, finalising catchment parameters from model runs of short hydrological events (on a scale of weeks to months) can generate mis-leading catchment parameters in the hydrological model.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 56Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

To assess calibration performance between the modelled and observed discharges, scattered plots shall be constructed at all observed gauging stations on matching of peak magnitude and time to peak, and on the overall shape of modelled and observed hydrographs. In addition to visual comparison of these graphs, statistical methods shall be used to measure the model’s performance. Examples of such statistical methods are:

● Nash–Sutcliffe efficiency ● Coefficient of determination (R2) ● Volumetric error

8.4.6 Validation

Model validation is a process of testing the model’s ability to simulate observed data for a different set of rainfall events than those used in calibration, within accuracy agreed with the client. In model validation, calibrated model parameters shall not be changed; the same set of parameter values used in calibration shall be used during validation. The validation period shall cover hydrological data of at least three hydrological years but preferably five (please see preceding section on criteria and issues on selecting period of validation).

To assess validation performance, the same procedure as for the calibration shall be followed; performance shall be checked by comparing the graphics of the peak magnitude and time to peak, as well as comparing the overall shape of modelled and observed hydrographs. The statistical parameters listed above for calibration shall also be checked.

8.5 Combined rainfall-runoff and gauge-to-gauge correlation Combined rainfall runoff and gauge-to-gauge correlation will also cover the same river reaches as were covered in the gauge-to-gauge correlation. In Mohana, 57km river length, and in Khutiya, 32km river length will be covered (Table 20).

Once the hydrological models are ready, they can be combined with gauge-to-gauge correlation. A ready hydrological model means a calibrated and validated model; if there is need for re-delineation off catchment than in the Package 7 model, then satellite imageries and in-built GIS tool will be used to redefine the watershed boundary; calibration will be carried out using new discharge data proposed for measurement in this study. This will increase lead time to 72 hours (as in the rainfall forecast). However, to utlilise the benefit of increased lead time, the stage-discharge rating curve will be required to convert runoff from the hydrological model into the river level at the upstream base station in gauge-to-gauge correlation.

Table 20: Proposed river reaches for development of combined rainfall-runoff and gauge-to-gauge correlation flood forecasting model

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model type

Mohana 1 Hill 4.97 14.1 RR 2 Fan 8.37 1.1 RR 3 Peripheral fan (Godawari Confluence) 13.87 0.1 RR+gauge-to-

gauge 4 Flood plain, meander (India border) 21.29 0.1 RR+gauge-to-

gauge 5 Flood plain, meander (India border-

Khutiya) 21.34 0.1 RR+gauge-to-

gauge

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 57Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model type

Godawari 1 Hill 9.26 13.4 RR 2 Braided 6.97 1.2 RR 3 Flood plain, Meander (Mohana

Confluence) 6.52 0.2 RR

Monohara 1 Hill 6.29 16.2 RR 2 Braided 4.80 0.9 RR 3 Flood plain, meander (Mohana

confluence) 8.59 0.1 RR

Khutiya 1 Hill 19.91 8.3 RR 2 Braided 6.51 1.1 RR+gauge-to-

gauge 3 Flood plain, Meander (Shivaganga

confluence) 15.46 0.2 RR+gauge-to-

gauge 4 Meander (Mohana confluence (Nepal-

India border) 9.77 0.1 RR+gauge-to-

gauge Shiva Ganga

1 Hill 10.92 14.7 RR 2 Braided 3.78 1.4 RR 3 Flood, plain, meander Shiva Ganga

(Khutiya) 19.55 0.2 RR

Source: Mott MacDonald

8.6 Pilot pure 2-d modelling Pure 2-d modelling has been proposed only in this basin and in the Mawa-Ratuwa basin. In other basins, 2-d model will be developed through 1d/2d linked modelling. The 2-d model domain will only be the very flat region in the Terai where flood water spreads very easily. This will be a 21km reach in Mohana and a 10km reach in Khutiya (Table 21 and Figure 11).

DHM, up to the present time, has not applied any 2d or 1d/2d linked model for flood forecasting. Thus, on-the-job training will be provided on 2-d modelling to DHM Forecasting Experts through this 2-d model development. This 2-d model will afterwards be transformed into a 1-d/2-d linked model.

The model shall be calibrated and validated for the same hydrological events as mentioned for hydrological modelling (section 8.4), and also see data availability in Table 9.

Table 21: Proposed river reaches for development of combined rainfall-runoff and 1-d modelling for flood forecasting model

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model type

Mohana 1 Hill 4.97 14.1 - 2 Fan 8.37 1.1 - 3 Peripheral fan (Godawari Confluence) 13.87 0.1 - 4 Flood plain, meander (India border) 21.29 0.1 - 5 Flood plain, meander (India border-

Khutiya) 21.34 0.1 Pure 2-d model

Godawari 1 Hill 9.26 13.4 -

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 58Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model type

2 Braided 6.97 1.2 - 3 Flood plain, Meander (Mohana

Confluence) 6.52 0.2 -

Monohara 1 Hill 6.29 16.2 - 2 Braided 4.80 0.9 - 3 Flood plain, meander (Mohana

confluence) 8.59 0.1 -

Khutiya 1 Hill 19.91 8.3 - 2 Braided 6.51 1.1 - 3 Flood plain, meander (Shivaganga

confluence) 15.46 0.2 -

4 Meander (Mohana confluence (Nepal-India border)

9.77 0.1 Pure 2-d model

Shiva Ganga

1 Hill 10.92 14.7 - 2 Braided 3.78 1.4 - 3 Flood plain, meander Shiva Ganga

(Khutiya) 19.55 0.2 -

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 11:proposed r

Source: Mott

8.7 1-d All reaches Siwalik hillstotal of 131of Godawaincluded (Fi

The model hydrologica

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

: Two-dimereaches with

MacDonald

modelling of the Moh

s up to the Inkm has been

ari (7km), Migure 12).

shall be calil modelling (

ration of Priority Rning System: Moh

9 ning System: Moh

ensional (2dhin 2d doma

ana River andia-Nepal bon proposed: onohara (9k

brated and v(section 8.4)

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

d) model dain

nd the Khutorder in the s65km in Mo

km) and Sh

validated for and also see

d Risk Managemesin

sin

domain in

tiya River, frosouth, have hana and 32

hivaganga (2

the same he data availa

ent Project, Nepal

Mohana-Kh

om immediabeen propos

2km in Khutiy20km) have

ydrological eability in Tabl

hutiya basin

ately downstrsed for 1-d mya. Only par

been propo

events as mee 9.

n showing

ream of the modelling. A rtial reaches osed to be

entioned for

599

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 60Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

8.7.1 River network

A forecasting tool in the 1-d modelling approach has been proposed to be developed for the Mohana and Khutiya rivers. Three other tributaries have also been proposed for inclusion in the 1-d model. The Godawari and the Monohara Rivers are two important tributaries of Mohana, and Shivaganga River is also an important tributary for Khutiya River. Thus, these three tributaries shall also be considered during the 1-d modelling for this basin.

The river reaches in the Terai as shown in Figure 1 and Table 7 shall be considered in 1-d river modelling. Runoff from the catchment in the Siwalik Hills shall be routed to the 1-d river model by hydrological modelling. Details of reaches and catchments considered in rainfall runoff (RR) and 1-d river modelling are presented in Table 22. If during the development phase it will be deemed appropriate, depending on topography and channel density in the Siwaliks, flood routing, e.g. based on Muskingum-Cunge, will also be considered. Such hydrological routing may improve flood attenuation, flood volume and flood travel time to the downstream 1-d model in the Terai. The above proposed reach lengths in the 1-d model shall be fine-tuned during the development phase depending on the field conditions as more physical data (water level, discharge, DEM and cross-sections) become available.

A single fluvial model shall be built considering all the main rivers and their tributaries included in the same model set-up. The model, with all interconnected branches, will deliver better results. None of the tributaries/branches shall be built as a separate 1-d model.

Modelling approach shall be submitted for acceptance by the client (i.e., DHM) before model build commences.

Key characteristics of the model shall include:

● ● Distributed inflows to reflect the key hydrological characteristics of the catchment ● All structures which influence flood flows/levels between 50% (generally bank-full discharge)

and 0.1% AEP plus climate change allowance ● All flood defence work ● All model nodes and units are to be geo-referenced, to true geographic co-ordinates (i.e.,

schematic set-up of model units shall not be accepted) ● Channel (1-d), bank to bank ● Floodplain as extended section in 1-d, as flood cells connected to the river (1-d) or as

separate channel and connected to main channel.

Table 22: Proposed river reaches for development of combined rainfall-runoff and 1-d modelling for flood forecasting model

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model: gauge-to-

gauge correlation

Mohana 1 Hill 4.97 14.1 RR 2 Fan 8.37 1.1 RR+1-d model 3 Peripheral fan (Godawari

Confluence) 13.87 0.1 RR+1-d model

4 Flood plain, meander (India border)

21.29 0.1 RR+1-d model

5 Flood plain, meander (India 21.34 0.1 RR+1-d model

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 61Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

River Reach ID

Reach Characteristics Channel length (km)

Slope (%)

FF model: gauge-to-

gauge correlation

border-Khutiya) Godawari 1 Hill 9.26 13.4 RR

2 Braided 6.97 1.2 RR 3 Flood plain, Meander (Mohana

Confluence) 6.52 0.2 RR+1-d model

Monohara 1 Hill 6.29 16.2 RR 2 Braided 4.80 0.9 RR 3 Flood plain, meander (Mohana

confluence) 8.59 0.1 RR+1-d model

Khutiya 1 Hill 19.91 8.3 RR 2 Braided 6.51 1.1 RR+1-d model 3 Flood plain, Meander

(Shivaganga confluence) 15.46 0.2 RR+1-d model

4 Meander (Mohana confluence (Nepal-India border)

9.77 0.1 RR+1-d model

Shiva Ganga

1 Hill 10.92 14.7 RR 2 Braided 3.78 1.4 RR 3 Flood plain, meander Shiva

Ganga (Khutiya) 19.55 0.2 RR+1-d model

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 12:proposed r

Source: Mott

8.7.2 C

The model hydrologicaand observapplied.

8.8 1-d/The floodplaTerai shall length for ththe pure 2-d

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

One-dimereaches with

MacDonald

Calibration a

shall be calil modelling (

ved discharg

/2-d linked ain which is be transform

he 1-d/2-d lind and 1-d/2-

ration of Priority Rning System: Moh

9 ning System: Moh

nsional (1-hin 1d doma

and validatio

brated and v(section 8.4)es, the sam

modellingmodelled in med into thenked model -d linked mod

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

d) model dain

on

validated for ). To assesse statistical

2-d (see Sec

e 2-d model shall be 52kdel will be g

d Risk Managemesin

sin

domain in

the same hs calibration

methods as

ction 8.6) andand then lin

km (Table 23eneration of

ent Project, Nepal

Mohana-Kh

ydrological eperformance

s mentioned

d some addinked with th3 and Figuref flood outline

hutiya basin

events as mee between th

in section 8

tional flat reahe 1-d modee 13). The ades as direct

n showing

entioned for he modelled 8.4 shall be

aches in the el. The total dvantage of output from

622

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 63Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

the model results, whereas in 1-d modelling, flood outlines have to be generated separately using 1-d model results and floodplain DEM.

Table 23: Proposed river reaches for development of combined rainfall-runoff and 1-d/2-d linked modelling for flood forecasting model

River Reach ID Reach Characteristics Channel length (km)

Slope (%)

FF model: gauge-to-

gauge correlation

Mohana 1 Hill 4.97 14.1 RR 2 Fan 8.37 1.1 RR+1-d model 3 Peripheral fan (Godawari

Confluence) 13.87 0.1 RR+1-d model

4 Flood plain, meander (India border)

21.29 0.1 RR+1-d/2-d model

5 Flood plain, meander (India border-Khutiya)

21.34 0.1 RR+1-d/2-d model

Godawari 1 Hill 9.26 13.4 RR 2 Braided 6.97 1.2 RR 3 Flood plain, Meander

(Mohana Confluence) 6.52 0.2 RR+1-d model

Monohara 1 Hill 6.29 16.2 RR 2 Braided 4.80 0.9 RR 3 Flood plain, meander

(Mohana confluence) 8.59 0.1 RR+1-d model

Khutiya 1 Hill 19.91 8.3 RR 2 Braided 6.51 1.1 RR+1-d model 3 Flood plain, Meander

(Shivaganga confluence) 15.46 0.2 RR+1-d model

4 Meander (Mohana confluence (Nepal-India border)

9.77 0.1 RR+1-d/2-d model

Shiva Ganga 1 Hill 10.92 14.7 RR 2 Braided 3.78 1.4 RR 3 Flood plain, meander Shiva

Ganga (Khutiya) 19.55 0.2 RR+1-d model

Source: Mott MacDonald

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 13: basin show

Source: Mott

8.9 Ope

8.9.1 K

Operating twould proba

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

One-dimenwing propos

MacDonald

eration of f

Key tasks

he forecast mably include:

ration of Priority Rning System: Moh

9 ning System: Moh

nsional (1dsed reaches

forecasting

model on a

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

) and 1d/2dof model do

g model

continuous b

d Risk Managemesin

sin

d linked momain

basis will inv

ent Project, Nepal

odel domai

volve automa

in in Moha

ation; the ma

ana-Khutiya

ain activities

644

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 65Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

● Writing scripts/program which will automatically download forecasted rainfall from WSF/WRF/IMD; the frequency of downloading shall be discussed and agreed with DHM; if forecast is issued three times during the day, the downloading will be at the same frequency

● Writing scripts/program which will convert forecasted rainfall into format which is compatible as input to runoff modelling tool, e.g., in case of NAM, it should be a dfs0 file while in HEC-HMS, it should be a HEC-HMS-DSS output file

● Writing scripts/program which will download real time discharge and water level from DHM’s central server, convert the data into HEC-RAS or MIKE11 compatible format; this data will be used for the forecast; forecast run is usually for 7 days duration – 4 days of hindcast whose performance is verified using real time water level and discharge and 3 days (72 hours) of forecast run.

● Writing a scripts, which will identify missing and erroneous data, particularly for rainfall data, which is used as input to hydrological model for generating run-off, and which are then input to the hydrodynamic model; both erroneous and missing rainfall record shall be replaced with data from other sources (e.g., from neighbouring station/grid)

● Writing script/program which will trigger automatic run of hydrological and the hydrodynamic models, for the same number of times during a day as agreed with DHM

● Writing script/program which will extract output (water level and discharge, and flood inundation map) in graphical formats at all forecasting points.

8.9.2 Real-time data transmission and maintenance

Maintaining a central database server for telemetric data and also for near real time data is essential. This study will uitlise the existing real time data management system (Figure 14) within DHM for data transmission to the central server, analysis and preparation of input for the model run. Input by International and National Experts have been kept for integration of telemetric data from the proposed new telemetric gauge network to DHM’s system.

Key elements for real time data transmission and management involves:

Operation of telecommunication system: this is outsourced and supervised by DHM Processing of data received from telemetered gauges by Flood Forecasting Centre,

DHM Processing of data received from manual gauges by Flood Forecasting Centre, DHM

During a forecast run, the hydrological model and hydrodynamic models use the following data:

Forecasted rainfall from weather forecast model Real time rainfall, water level and discharge data from the telemetric gauging network

During each forecast run, once daily (or more), the model will run for a 4-day hindcast period and a 3-day forecast period. For the hindcast part of the simulation, input data (rainfall, water level and discharge) should be real-time data, which may also be supplemented by TRMM gridded rainfall data. Running of the hydrological and hydraulic models during the forecast season is carried out by the Flood Forecasting Centre, DHM. The 7-day run time only in 1d only model will be quite fast requiring about 30 minutes for pre-processing, model run time and post processing. In case of 1d/2d linked model, the run time is expected to be higher, between 70 and 75 minutes. However, this run time will depend on the type of computer being used, and spatial resolution of both 1-d and 2-d model.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 14:

Source: DHM

8.9.3 E

DHM’s exisnew models

To operate & MIKE11, derivation aand hydrodyfor these tw

However, thmodels, e.gAustralia. Inpreparing inare also tooDelft-FEWS

In this studoperational input for theexisting systo integrate

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Real time da

M (Courtesy Mr B

Existing fore

ting operatins.

a forecastinthe pre-pro

are relativelyynamic mod

wo systems fo

here are forg., PDM and n cases like nput data anols available

S (See Sectio

y, as DHM in their two e consultant

stem will be the new bas

ration of Priority Rning System: Moh

9 ning System: Moh

ata manage

Binod Parajuli, H

ecast model

ng system (F

g system usocessing any easy. In theel (HEC-RASor pre-post p

recasting moFlood Modethese, relativd deriving o

e for such pon 8.9.4)

has already major basin

ts for develoused. Input

sin models fr

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

ement and tr

Hydrologist/Fore

l operating s

igure 15) wil

sing models dd post procese two caseS or MIKE11

processing pu

odels which ller Pro usedvely consideutputs while

purpose, e.g.

HEC-HMS (Kosi and B

oping an opeby internatio

rom this stud

d Risk Managemesin

sin

ransmission

ecaster, Flood F

system with

ll be used fo

developed incessing workes, the hydro1) are coupleurpose requi

use uncoupd widely in Uerable works

running the. Delft-FEWS

and HEC-RABabai). As a erating systeonal and naty to DHM’s e

ent Project, Nepal

n system wit

Forecasting Sec

hin DHM

r operation a

n HEC-HMS k for input ological mod

ed. Thus, wrires minimum

pled hydrologK, and URBSare required model in reS. A separa

AS, and NAresult, we h

em for the foional consultexisting forec

thin DHM

ction)

and forecasti

& HEC-RASpreparation del (HEC-HMting any cod

m effort.

gical and hyS and TUFLOd for such aual time. How

ate section is

AM and MIKEave proposeorecast modtant has onlycasting syste

ing from the

S, and NAM and output

MS or NAM) de/script/tool

ydrodynamic OW used in utomation in wever, there s added on

E11 models ed minimum dels. DHM’s y been kept em.

666

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 15:

Source: DHM

8.9.4 D

Delft-FEWShandling tDelft-FEWShydraulic m(rainfall-runon third pastructure ofseveral comof Delft-FEDelft-FEWSfully autom

DHM has ahydrologica(DHM, 201developed operational

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

DHM’s exist

M / ICIMOD

Delft-FEWS

S provides ime series S forecastin

models usednoff and hydarty modellinf the Delft-Fmponents fo

EWS is showS can eitherated distribu

an existing al and hydra18). Similarl

in NAM al system.

ration of Priority Rning System: Moh

9 ning System: Moh

ting flood fo

an open shdata (http

g system wa (Werner et

drodynamic ng componeFEWS includor importing,wn in Figurer be deployeuted client-s

operation syaulic modelliy, Banglade

and MIKE11

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

orecasting a

hell system ps://publicwikas essentialal., 2012). Tmodelling) w

ents for raindes a data s, manipulatin

e 16. Currened in a standerver enviro

ystem for Fng, and in Kesh and Ind1 and HEC

d Risk Managemesin

sin

and dissemi

for managki.deltares.nlly built as aThe system within its confall-runoff astorage layeng, viewing

ntly Delft-FEd-alone, ma

onment.

FEWS for mKarnali and Ndia (Bihar), C-HMS and

ent Project, Nepal

ination syst

ing forecasnl/display/FEa shell aroun

contains noode base. Inand hydrodyer, a data acand exporti

EWS is usedanually drive

models, e.g.Narayani by in their flooHEC-RAS

em

ting procesEWSDOC/Hond the hydroo modelling cnstead, it enynamic modccess layer,ng data. Th

d in over 40en environm

. Kosi and Bprobabilistic

od forecasti, also use

sses and/or ome). The

ological and capabilities

ntirely relies delling. The as well as

he structure 0 countries.

ment, or in a

Bagmati by c modelling ing models

their own

677

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 16: Delft-FEWS

Source: Wern

8.9.5 D

Disseminati

● Writing designat

8.9.6 D

Data assimwith the motechnique cinto the hyuncertainty

The flood fshould havewater level.the forecast

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Schematic S and links t

ner et al. (2012)

Disseminatio

ion of foreca

script/prograted recipients

Data assimil

milation is a todel dynamiccan be applieydrodynamic assessment

forecasting me the ability t This will allt.

ration of Priority Rning System: Moh

9 ning System: Moh

structure oto other prim

)

on of foreca

st will involve

am, which ws including C

ation

technique fos in order to ed by incorp

model. Furt.

methodologyto assimilatelow the upda

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

of a flood fomary system

ast

e automation

will disseminCBDRM com

r combining improve the

porating obserthermore, th

y to be appe real time/neating of mod

d Risk Managemesin

sin

forecasting ms within th

n. The main a

ate forecastmittee and o

any measue knowledge erved water he data ass

plied in the ear real timedel results in

ent Project, Nepal

system, shhe operation

activities sha

t including gother stake h

rements of tof the systemlevel and di

similation mo

FFEWS deve telemetry o real time an

owing the pnal environm

all include:

graphical ouolders

the state of m. The data ascharge meaodule can b

velopment inobservations nd improve a

position ofment

utputs to all

the system assimilation asurements

be used for

n this basin of flow and accuracy of

688

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 69Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Data assimilation needs to be applied on the following forecast model on discharge and water level where real time telemetry data are available:

● Combined rainfall-runoff and gauge-to-gauge correlation model ● 1-d hydrodynamic model ● 1-d and 2-d linked hydrodynamic model.

8.10 Evaluation of forecast Based on the evaluation, follow-up (continuous) model update shall be recommended and implemented within the duration of this project, while the knowledge transfer to DHM shall be ensured for future update, operation and maintenance of the following forecast models:

● Gauge-to-gauge correlation ● Combined rainfall-runoff and gauge-to-gauge correlation model, within the duration of this

project ● 1-d hydrodynamic model, within the duration of this project ● 1-d and 2-d linked hydrodynamic model, within the duration of this project.

Forecast evaluation should be carried out using the Skill Scores as per the criteria described in WMO’s Manual on Flood Forecasting and Warning (WMO, 2011).

8.11 Model development schedule The suite of forecasting models shall be developed and made operational over a period of three years (Figure 17), represented by sub-programmes to be completed in nine periods (PR1 to PR9, one period is 4 months); this 36 month period is deemed essential as the model will use new data from the proposed gauging network and topography including DEM from high resolution (50cm) satellite imagery.

● Gauge-to-gauge correlation can start fairly early as soon as some water level data are available from the new proposed water level gauges; it is noted again that there is no existing hydrometric network in this basin.

● Hydrological modelling will also be started from PR2 by using third party and existing rainfall data; as soon as rainfall data from proposed new rainfall gauges are available. The model will be updated, calibrated, validated and improved with new rainfall data and discharge data.

● Forecast issuing will immediately be started using the gauge-to-gauge correlation approach (which has limited lead time). In parallel, as soon as the RR model is ready, the RR model will be combined with gauge-to-gauge correlation forecasting; combining with the RR model will give power to gauge-to-gauge correlation to forecast with much higher lead time (up to 72 hours).

● Parallel to the above modelling, 1-d, 2-d and 1-d/2-d linked model development will continue; this advanced modelling is more data dependent, particularly on topographic data.

● 2-d modelling will be carried out as a pilot exercise and for capability development. This experience will be used in 1-d/2-d linked model development. Forecasts, however, will be issued using the 2-d model for the domain where the model is developed. The 1-d forecast model will be operational from PR5 and the 1-d/2-d linked model will be operational from the middle of PR7.

● 1-2-d linked model will be operational from the middle of PR5.

Mott MacDonaldFlood Forecastin

383877 | REP | 0Flood Forecastin

Figure 17:

Source: Mott

8.12 ModBased on ri(Table 24).

Table 24: FCategories

Data: collectiprocessing, aHydrological modelling Gauge-to-gacorrelation Pure 2-d mod1-d modelling1-d/2-d linkedmodelling Modelling soTotal Note: Modelllicence, and cSource: Mott

In the UK, cost consid

| WRPPF: Prepang and Early Warn

0040 | 4 April 2019ng and Early Warn

Flood forec

MacDonald

del developver length (cSimilar unit c

Forecasting P

ion, analysis

P

Ca

uge R

delling Rg Rd R

oftware S

ing software licost shown heMacDonald

hydrologicalered here is

ration of Priority Rning System: Moh

9 ning System: Moh

casting mod

pment budconsidered focosts are ap

model deveParameter

Per basin

Catchment rea

River length

River length River length River length

Suite

icence cost is ere is per basi

model deves relatively lo

River Basins Floodana – Khutiya Bas

ana – Khutiya Bas

el developm

get or modelling)plicable in th

elopment buUnit Quan

No.

km2

km

km km km

No.

distributed ovin. West Rapti

elopment coow because

d Risk Managemesin

sin

ment progra

) and catchmhe UK and Ire

udget for Montity Capit

opme

1

702

88

31 131 52

1

ver five basinsi is excluded f

st is $500 tothe hydrolo

ent Project, Nepal

mme for Mo

ment size, uneland (Enviro

ohana-Khutial/Develent cost

($)

O

48,500

137,500

80,500

93,000 144,500 169,500

13,000 686,500

s; software willfrom software

o $1500 perogical model

ohana-Khuti

it cost has beonment Agen

iya basin Operation

cost ($) D

-

-

32,500

37,500 73,500 64,500

- 208,000

l have multi uscost

r km2 of catcalready exis

iya basin

een derived ncy, 2015).

Dissemi-nation

cost ($) - 48

-

27,500

27,500 535,000 26,250 4

- 13116,250 ser network

chment; the sts in these

70

Unit cost

($) 8,500

196

1,597

5,097 1,931 4,967

3,000

0

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 71Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

five sub-basins (from pre-feasibility and feasibility level). This advantage will reduce development cost during FFEWS).

8.13 Person-months for experts The suite of FFEWS tools shall be developed first in the Mohana-Khutiya basin, and thus expert input in this basin is higher than for other basins (Table 25).

The next basin shall be Mawa-Ratuwa. Thus, the other basins will benefit from the experience gained from these two basins; as a result, the other three basins will require fewer person-months. The catchment size and river lengths were also key factors in deciding person-months. Gradually with experience, the input from international experts will decrease in other basins than Mohana-Khutiya and Mawa-Ratuwa.

In developing FFEWS in five basins, the inputs of three international experts and four national experts have been considered over a period of three years; one GIS cum data expert (National) has also been considered to support the team. The discipline of both international and national experts shall be:

International

Senior/Principal Hydraulic & flood forecasting modelling expert Hydraulic & flood forecasting modelling expert Hydrologist & flood forecasting modelling expert

National

Hydraulic & flood forecasting modelling expert-I and expert-II Hydrologist & flood forecasting modelling expert-I and expert-II GIS cum data analysis expert

Following activities will be first carried out in this basin. As these activities could easily be applied/customised to other basins, so the cost for these activities will be relatively low in the other basins than this basin.

● A detailed model development conceptualisation document (according to feasibility document) shall be developed for this basin and shall be copied to other basins

● Training and capacity building for 1-d, 2-d and 1-d/2-d linked modelling will be offered to national experts and DHM professionals during development of tools in this basin

● A script for automation of FF model runs in real time shall be developed for this basin, and can be adopted to other basins with very minimum input

● A script for automation of forecast dissemination in real time shall be developed for this basin and can be adopted to other basins with very minimum input.

Further, Mohana-Khutiya is actually two separate basins with separate catchments and river systems, although in this study, both basins were considered as one unit. Thus, cost, in general, is higher in this basin than the other basins.

Considering the above factors, the person-months for Mohana-Khutiya have been calculated as below:

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 72Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Table 25: Experts’ person-months for the Mohana-Khutiya basin Models: Mohana-Khutiya

Development phase

Operational phase

Dissemination phase

Inter-

national National Inter-

national National Inter-

national National

Data: collection, processing and analysis

1 2

Hydrological modelling 4 6 Gauge-to-gauge correlation

2 4 0.5 1 0.3 1

1-d modelling 4 8 2 2 0.6 1 Pure 2-d modelling 2.5 4 0.7 1 0.3 1 1-d/2-d linked modelling 5 8 1.5 3 0.25 1 Total 18.5 32 4.7 7 1.45 4

Source: Mott MacDonald

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 73Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

References

[1] Allot, T (2010), The British Rainfall Network in 2010, https://www.rmets.org/sites/default/files/pdf/presentation/20100417-allott.pdf

[2] Berndtsson, R and Niemczynowicz, J (1988), Spatial And Temporal Scales in Rainfall Analysis Some Aspects And Future Perspectives, Journal of Hydrology, 100 (1988) 293-313

[3] DHM (2018), Standard Operating Procedure for Flood Early Warning System in Nepal

[4] DoI (2016), Package 3: Flood Hazard Mapping and Preliminary Preparation of Flood Risk Management Projects, Final Report – VOLUME 1, Prepared by Lahmeyer International in association with Total Management Services

[5] EA/Defra (2013), Benchmarking the latest generation of 2D hydraulic modelling packages, Report – SC120002

[6] EA/Defra (2004), Benchmarking of hydraulic river modelling software packages, Project Overview, R&D Technical Report: W5-105/TR0, URL for this research is below:

(https://consult.environment-agency.gov.uk/engagement/bostonbarriertwao/results/appendix-6---neelz--s.---pender--g.--2013--benchmarking-the-latest-generation-of-2-d-hydraulic-modelling-packages.-bristol_environment-agency.pdf)

[7] Environment Agency (2015), Cost estimation for flood warning and forecasting – summary of evidence, Report –SC080039/R13

[8] Huxley C (2016), GPU – Next Generation Modelling for catchment Floodplain Management, BMT-WBM, ASFPM Conference

[9] Lopez M/G. et al, (2015). Location and Density of Rain Gauges for the Estimation of Spatial Varying Precipitation. Geografiska Annaler: Ser. A Physical Geography. 97, (1) 167-179

[10] Lengfeld et al. (undated), Pattern: Advantages of High Resolution Weather Radar Network, American Meteorological Society 36th Conference on Weather Radar Networks

[11] Mott MacDonald (2018a), Morphology Assessment: Mohana – Khutiya basin, WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal

[12] Mott MacDonald (2018b), River Hydrology Assessment: Mohana – Khutiya basin, WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 74Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

[13] NIWA (2014), Climate Manual, National Institute of Water & Atmospheric Research Ltd

[14] Schaake, J., 2004. Application of prism climatologies for hydrologic modeling and forecasting in the western U.S. In Proceedings of 18th Conference on Hydrology. Seattle, Washington, 2004. American Meteorological Society

[15] Smith, P.J., Brown, S and Dugar, S (2017), Community-based early warning systems for flood risk mitigation in Nepal, Nat. Hazards Earth Syst. Sci., 17, 423–437

[16] Syme, B (2007), 2-d and 1-d/2-d modelling, BMT WBM

[17] Volkman T. H.M., Lyon, S. W., Gupta, H. V. and Troch, P. A. (2010), Multicriteria design of rain gauge networks for flash flood prediction in semiarid catchments with complex terrain. Water Resources Research 46, W11554, doi:10.1029/2010WR009145, 16pp

[18] Werner et al. (2012), The Delft-FEWS Flow Forecasting System, Environmental Modelling and Software, 40 (2013), 65-77

[19] WMO (2011), Manual on Flood Forecasting and Warning, WMO-No. 1072

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 75Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

Appendices

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

76

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

A. M

odel

ling

softw

are

com

paris

on

FEAT

UR

E D

ESC

RIP

TIO

N

MIK

E Fl

ood

Mod

elle

r Pr

o S

OB

EK

H

EC

TUFL

OW

W

EAP

MO

DSI

M

RIB

ASIM

In

fow

orks

IC

M

GE

NE

RA

L

Sing

le s

oftw

are

suite

N

o in

terfa

ce p

robl

ems;

on

e su

pplie

r for

sup

port

x x

x x

x x

x

Trac

k re

cord

of

supp

ort

Ensu

re fo

r for

esee

able

fu

ture

GIS

Bas

ed

Spat

ial i

nfor

mat

ion

esse

ntia

l

x

Ope

nMI c

ompl

iant

Li

nkag

es to

ext

erna

l so

ftwar

e

x

x x

x x

x x

Loca

l fam

iliar

ity

Loca

l sup

port

in N

epal

x

x

x

x x

x x

Use

r and

Ref

eren

ce

Man

uals

Sc

ient

ific

back

grou

nd a

nd

user

inte

rface

.

Esta

blis

hed

trai

ning

co

urse

s R

egul

ar tr

aini

ng c

ours

es

x x

Gra

phic

al in

terf

ace

D

ata

entry

and

vi

sual

isat

ion

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

77

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

FEAT

UR

E D

ESC

RIP

TIO

N

MIK

E Fl

ood

Mod

elle

r Pr

o S

OB

EK

H

EC

TUFL

OW

W

EAP

MO

DSI

M

RIB

ASIM

In

fow

orks

IC

M

HYD

RO

LOG

Y N

AM

- -

HEC

-HM

S -

WEA

P M

OD

SIM

Snow

and

Gla

cier

Mel

t R

unof

f fro

m s

now

and

gl

acia

l mel

t

x

x

x

x x

x

Rai

nfal

l-Run

off

Run

off f

rom

rain

fall.

x

x

x

Auto

-cal

ibra

tion

Auto

mat

ic a

djus

tmen

t of

para

met

ers

x x

x

x

HYD

RAU

LIC

S M

IKE

11

ISIS

SO

BEK

H

EC-R

AS

TUFL

OW

Info

wor

ks

ICM

Full

hydr

odyn

amic

s Fu

ll hy

drod

ynam

ic

anal

ysis

Stru

ctur

e op

erat

ions

St

ruct

ures

and

con

trols

Inflo

w a

nd F

lood

fo

reca

stin

g Ad

vanc

ed d

ata

assi

mila

tion

x x

x x

x x

x x

Opt

imis

atio

n O

ptim

al o

pera

tion

of

syst

em c

ontro

ls

x x

x x

x x

x x

Auto

-cal

ibra

tion

Auto

mat

ic a

djus

tmen

t of

para

met

ers

x x

x x

x x

x x

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

78

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

FEAT

UR

E D

ESC

RIP

TIO

N

MIK

E Fl

ood

Mod

elle

r Pr

o S

OB

EK

H

EC

TUFL

OW

W

EAP

MO

DSI

M

RIB

ASIM

In

fow

orks

IC

M

Sedi

men

t tra

nspo

rt

(opt

iona

l) Se

dim

ent t

rans

port

Wat

er q

ualit

y (o

ptio

nal)

Tran

spor

t and

dec

ay o

f su

bsta

nces

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

79

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

B. C

omm

ents

and

resp

onse

s

Tabl

e B

.1: C

omm

ents

on

FFEW

S R

epor

ts fr

om A

DB

C

omm

ents

fro

m A

DB

wer

e ge

neric

for

the

rep

orts

fiv

e ba

sins

: M

ohan

a-Kh

utiy

a, M

awa-

Rat

uwa,

Lak

hand

ei,

Bakr

aha

and

East

Rap

ti an

d

Lakh

ande

i.

C

omm

ents

wer

e re

ceiv

ed in

a M

S W

ord

file

whi

ch a

re p

rese

nted

in T

able

bel

ow

Com

men

ts w

ere

also

rece

ived

on

the

hard

cop

y of

the

Moh

ana-

Khut

iya

(M-K

) Rep

ort o

n ea

ch c

hapt

er (c

hapt

er 0

to 8

). Th

ose

com

men

ts,

thou

gh m

ade

on M

-K re

port,

are

mos

tly g

ener

ic a

nd a

pplic

able

for a

ll th

e ot

her f

our b

asin

s. A

ll co

mm

ents

mad

e on

the

hard

cop

y ha

ve

been

add

ress

ed in

all

five

basi

n re

ports

and

repo

rts h

ave

been

upd

ated

acc

ordi

ngly

. The

cha

nges

mad

e ar

e av

aila

ble

on tr

ack

chan

ges

mod

e. F

or s

ome

com

men

ts, t

here

was

nee

d fo

r a re

spon

se fo

r the

AD

B re

view

er; t

hose

resp

onse

s ar

e m

ade

on th

e pd

f ver

sion

of e

ach

chap

ter.

Res

pons

es fo

r the

se c

omm

ents

are

pre

sent

ed b

elow

.

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

1 La

ngua

ge:

En

glis

h ne

eds

to

be

impr

oved

. R

epet

ition

of

text

is o

bser

ved

at m

any

plac

es..

Sugg

estio

ns h

ave

been

pro

vide

d on

a m

arke

d-up

cop

y of

the

Moh

ana-

Khut

iya

FFEW

S re

port.

All s

ugge

stio

ns m

ade

on th

e m

arke

d up

cop

y of

M-K

Rep

ort h

ave

been

add

ress

ed;

rem

oved

repe

titio

n at

pla

ces

and

have

als

o im

prov

ed E

nglis

h.

2 Ex

istin

g hy

drom

et d

ata

: Pro

vide

an

over

view

ta

ble

of th

e ex

istin

g an

d fo

reca

st h

ydro

met

dat

a (ra

infa

ll, w

ater

-leve

ls,

flow

s, t

empe

ratu

re a

nd

evap

otra

nspi

ratio

n), i

nclu

ding

; o

Ty

pe &

sou

rce

We

have

pro

vide

d an

ove

rvie

w in

Tab

le 9

in c

hapt

er 2

.

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

80

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

o

Lo

catio

n o

Pe

riod

avai

labl

e o

C

olle

ctio

n m

etho

d o

C

olle

ctio

n fre

quen

cy

o

Tran

smis

sion

met

hod

o

Late

ncy

o

Whe

re it

is s

tore

d, e

tc.

3 Pr

ovis

ion

of X

-ban

d R

adar

s: It

has

bee

n fo

und

that

D

HM

(u

nder

th

e W

orld

Ba

nk’s

PP

CR

pr

ojec

t) ha

s in

stal

led

a C

-ban

d ra

dar a

t Sur

khet

, w

hich

cov

ers

an a

rea

with

a r

adiu

s of

250

km

. Si

mila

rly,

DH

M is

in t

he p

roce

ss o

f in

stal

ling

2 m

ore

C-b

and

rada

rs a

t C

entra

l an

d Ea

ster

n N

epal

, an

d se

vera

l x-b

and

rada

rs.

The

C-b

and

rada

rs w

ill co

ver t

he e

ntire

Ter

ai a

nd th

e X-

band

ra

dars

will

cove

r inn

er v

alle

ys.

Ther

efor

e, th

ere

is n

o ne

ed o

f pro

curin

g th

e 5

X-ba

d ra

dars

.

Agre

ed;

We

have

dro

pped

this

item

from

bud

get;

how

ever

, we

have

kep

t the

writ

e-up

in S

ectio

n 5.

3, if

D

HM

wis

hes

to c

onsi

der t

hem

in n

ear f

utur

e. W

ithin

the

text

s, w

e ha

ve m

ade

this

cle

ar th

at X

-ba

nd ra

dar w

ill no

t be

cons

ider

ed in

this

pro

ject

, and

thus

, no

budg

et h

as b

een

incl

uded

.

How

ever

, we

wan

t to

men

tion

that

the

C-B

and

long

rang

e ra

dar,

whi

ch D

HM

is in

the

proc

ess

of

inst

alla

tion

at th

ree

loca

tions

, may

not

be

oper

atio

nal o

r dat

a m

ay n

ot b

e av

aila

ble

durin

g ne

xt 2

to

3 y

ears

, by

whi

ch ti

me

this

pro

ject

may

be

com

plet

ed.

We

also

wan

t to

men

tion

that

tota

l 3 n

os. o

f C-b

and

rada

r acr

oss

Nep

al w

ill pr

ovid

e a

dens

ity o

f on

e ra

dar p

er 4

9,06

0 km

2 in N

epal

, whi

le s

uch

dens

ity, f

or e

xam

ple

in U

K, is

one

C-b

and

rada

r pe

r 14,

264

km

2 .

Ther

efor

e, D

HM

/AD

B, if

wis

hes

in fu

ture

, can

als

o co

nsid

er s

hort

rang

e ra

dar i

nsta

llatio

n.

4 Pr

ovis

ion

of

ADC

Ps:

DH

Ms

has

alre

ady

awar

ded

the

proc

urem

ent

of 5

AD

CPs

(un

der

the

Wor

ld B

ank’

s PP

CR

pro

ject

). O

ut o

f the

se 5

AD

CPS

, DH

M p

lans

to p

rovi

de o

ne e

ach

to it

s ba

sins

offi

ces

at B

iratn

agar

, Po

khar

a, B

haira

wa

and

Koha

lpur

, so

that

thes

e ba

sin

offic

es w

ill be

re

spon

sibl

e fo

r m

easu

ring

disc

harg

e in

all

the

river

s of

Nep

al. T

o su

pple

men

t DH

M’s

AD

CPS

,

We

have

mod

ified

the

disc

harg

e m

easu

rem

ent e

quip

men

t lis

t, ho

wev

er, a

bit

diffe

rent

th

an A

DB’

s su

gges

tion.

We

have

pro

pose

d th

ree

set o

f equ

ipm

ent;

plea

se s

ee o

ur c

onsi

dera

tions

:

M

ohan

a-Kh

utiy

a: o

ne s

et o

f equ

ipm

ent f

or th

is b

asin

alo

ne

(Thi

s ba

sin,

in fa

r wes

t Nep

al, i

s fa

r aw

ay fr

om th

e ot

her f

ive

basi

ns; t

his

basi

n ha

s 12

0dis

char

ge m

easu

rem

ents

dur

ing

thre

e ye

ars;

thus

it w

ill be

ver

y di

fficu

lt fo

r thi

s ba

sin

to s

hare

its

equi

pmen

t with

ano

ther

bas

in. S

imila

rly, i

t will

also

be

diffi

cult

to

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

81

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

it

mig

ht b

e us

eful

to

prov

ide

two

mor

e AD

CPs

fo

r tra

inin

g an

d as

sp

ares

fo

r m

easu

ring

disc

harg

e in

the

prio

rity

basi

ns.

Ther

efor

e, it

is

sugg

este

d th

at th

e co

nsul

tant

mod

ify th

e lis

t of

equi

pmen

t.

borro

w D

HM

’s e

quip

men

t for

120

mea

sure

men

ts;

M

awa-

Rat

uwa

and

Bakr

aha:

one

set

of e

quip

men

t for

thes

e tw

o ba

sins

M

awa-

Rat

uwa

and

Bakr

aha

will

have

30

mea

sure

men

ts p

er y

ear

for

each

ba

sin

(tota

l 60

mea

sure

men

ts in

two

basi

ns),

and

thes

e tw

o ba

sins

are

sid

e by

sid

e, a

nd th

us s

harin

g on

e eq

uipm

ent s

et in

this

bas

in is

pos

sibl

e. T

hus,

w

e ha

ve p

ropo

sed

one

equi

pmen

t set

for t

hese

two

basi

ns

Wes

t Rap

ti: w

e ar

e no

t pro

posi

ng a

ny e

quip

men

t set

. DH

M’s

equ

ipm

ent w

ill be

use

d in

this

bas

in

Ea

st R

apti

and

Lakh

ande

i: w

e ar

e pr

opos

ing

one

equi

pmen

t set

for t

hese

two

basi

ns; t

his

set w

ill be

sha

red

betw

een

the

two

basi

ns; h

owev

er, o

ccas

iona

lly,

ther

e m

ay b

e ne

ed to

bor

row

DH

M’s

equ

ipm

ent i

n cr

isis

man

agem

ent.

Ther

e w

ill ha

ve 3

0 m

easu

rem

ents

per

yea

r for

eac

h ba

sin

5 D

isch

arge

m

easu

rem

ent

: Th

e co

nsul

tant

is

su

gges

ted

to d

escr

ibe

the

disc

harg

e m

easu

ring

appr

oach

in t

he s

ix b

asin

s (in

clud

ing

the

Wes

t R

apti

Riv

er)

usin

g th

e AD

CPs

. Th

e ap

proa

ch

shou

ld

incl

ude

frequ

ency

, us

e of

bo

ats

or

cabl

eway

s an

d de

velo

pmen

t of

rat

ing

curv

es.

Also

, co

nsul

tant

sh

ould

de

scrib

e th

e in

volv

emen

t of D

HM

’s b

asin

offi

ces

with

a fo

cus

on c

apac

ity b

uild

ing.

We

have

fur

ther

des

crib

ed t

he a

ppro

ach

in S

ectio

n 6.

2.2

for

ADC

P m

easu

rem

ents

, Se

ctio

n 6.

2.3

on c

able

way

dis

char

ge m

easu

rem

ents

, and

Sec

tion

6.3

in m

easu

rem

ent

frequ

ency

. Rat

ing

curv

es in

Sec

tion

6.3,

par

a 3.

Invo

lvem

ent o

f DH

I bas

in o

ffice

:

Agre

ed;

in a

ll di

scha

rge

mea

sure

men

t, t

echn

ical

pro

fess

iona

l fro

m b

asin

/regi

onal

of

fice

of D

HM

will

be in

volv

ed; w

e ha

ve u

pdat

ed te

xts

acco

rdin

gly

in S

ectio

n 6.

2.2,

in

the

para

bel

ow th

e bu

llet p

oint

s

6 D

isch

arge

sta

tions

: S

tream

flow

gau

ging

and

ra

ting

curv

es a

re p

ropo

sed

for o

nly

som

e w

ater

-le

vel

stat

ions

. Pr

ovid

e ex

plan

atio

n fo

r w

hy t

he

wat

er-le

vel-o

nly

stat

ions

will

not b

e ga

uged

.

We

have

incl

uded

suf

ficie

nt n

umbe

r of s

tream

flow

and

ratin

g cu

rve

stat

ions

. Her

e, w

e w

ill ha

ve th

ee d

isch

arge

sta

tions

(on

e ex

istin

g an

d tw

o ne

w)

in a

ppro

xim

atel

y 59

km

reac

h of

the

Moh

ana.

In

Bang

lade

sh a

nd In

dia,

dis

char

ge m

easu

rem

ent s

tatio

ns a

re

at fa

r dis

tanc

es th

an th

is.

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

82

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

If no

new

trib

utar

y is

join

ing

the

mai

n riv

er o

r if t

he in

term

edia

te c

atch

men

t is

not l

arge

en

ough

, the

n th

ere

is n

o ne

ed to

add

new

dis

char

ge s

tatio

n; a

ll w

ater

leve

l sta

tions

do

not n

eed

to h

ave

flow

gau

ging

as

wel

l; th

is is

the

prac

tice

arou

nd th

e co

untri

es, l

ike

I In

dia,

Ban

glad

esh,

UK

and

in m

any

othe

r cou

ntrie

s.

7 C

alib

ratio

n/va

lidat

ion

data

: T

he c

alib

ratio

n &

valid

atio

n se

ctio

ns o

f th

e re

ports

men

tion

the

need

for c

alib

ratio

n/va

lidat

ion

data

, but

for m

any

basi

ns s

uch

data

are

una

vaila

ble.

Lin

ked

with

th

e ab

ove

bulle

t poi

nt, p

rovi

de a

sum

mar

y of

the

exis

ting

data

ava

ilabl

e fo

r ca

libra

tion/

valid

atio

n.

If no

dat

a ar

e av

aila

ble,

then

the

basi

n’s

wor

ks

prog

ram

me

shou

ld

refle

ct

the

need

to

st

art

colle

ctin

g da

ta

early

fo

r su

bseq

uent

us

e in

ca

libra

tion/

valid

atio

n.

We

have

pro

vide

d an

ove

rvie

w o

f da

ta a

vaila

bilit

y in

Tab

le 9

in S

ectio

n 2.

And

all

hydr

omet

net

wor

k ha

s be

en p

ropo

sed

for

inst

alla

tion

with

in f

irst

six

mon

ths,

so

that

pr

opos

ed m

odel

ling

can

use

the

data

fro

m t

he f

irst

mon

soon

for

cal

ibra

tion

and

valid

atio

n.

8 To

po s

urve

y : E

xpla

in w

hy a

dditi

onal

x/s

sur

vey

is re

quire

d. Is

this

requ

ired

for;

o Ac

cura

cy o

f lev

el fo

reca

st

o Ac

cura

cy o

f inu

ndat

ed a

rea

fore

cast

o R

atin

g cu

rve

exte

nsio

n (b

y hy

drau

lic

mod

el)

We

have

exp

lain

ed th

is in

the

repo

rt in

Cha

pter

7. T

his

surv

ey is

requ

ired

for a

ll th

ree

bulle

t po

ints

as

men

tione

d, a

nd a

s w

ell w

e sh

ould

rep

lace

all

cros

s-se

ctio

ns w

hich

w

ere

surv

eyed

in 2

014;

thes

e w

ill be

mor

e th

an 5

yea

rs o

ld. I

n U

K, w

here

riv

ers

are

very

sta

ble,

the

Envi

ronm

ent A

genc

y (re

spon

sibl

e fo

r flo

od fo

reca

stin

g) u

pdat

es th

eir

mod

el if

topo

grap

hy is

mor

e th

an 6

yea

rs o

ld. H

ere

in N

epal

, we

need

suc

h up

date

ea

rlier

as

the

river

s ar

e m

orph

olog

ical

ly d

ynam

ic. W

e al

so n

eed

mor

e cr

oss-

sect

ions

in

Tea

ri to

dev

elop

1d/

2d li

nked

mod

el a

nd 2

d m

odel

.

9 To

po s

urve

y :

Will

the

x/s

surv

ey i

nclu

de t

he

flood

plai

n, if

so

then

sta

te.

Yes,

cro

ss-s

ectio

ns w

ill be

ext

ende

d to

floo

dpla

in. I

t has

alre

ady

been

men

tione

d in

th

e Ex

ecut

ive

sum

mar

y; w

e ha

ve n

ow a

lso

men

tione

d th

is in

cha

pter

7 a

s w

ell.

10

2D m

odel

ling

– D

TM :

The

pro

pose

d hy

drau

lic

mod

el m

etho

dolo

gy i

nclu

des

2D m

odel

ling

for

som

e ar

eas.

Pr

evio

us

wor

k ha

s sh

own

Than

ks fo

r thi

s co

mm

ent a

nd c

omin

g up

with

you

r sup

port

to in

clud

e so

me

purc

hase

of

bet

ter D

TM b

y di

verti

ng th

e fu

nd o

f X-b

and

Rad

ar.

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

83

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

in

com

patib

ility

betw

een

the

avai

labl

e SR

TM-

deriv

ed D

TM a

nd t

he s

urve

yed

cros

s-se

ctio

ns.

The

repo

rt do

es

not

outli

ne

how

th

is

inco

mpa

tibilit

y w

ill be

re

solv

ed

eith

er

by

(i)

obta

inin

g a

new

DTM

fro

m a

ltern

ativ

e sa

tellit

e so

urce

s,

or

(ii)

surv

eyin

g th

e flo

odpl

ain,

by

Li

DAR

or

tradi

tiona

l met

hods

. If i

t’s is

pro

pose

d to

use

the

exi

stin

g SR

TM d

eriv

ed D

TM t

hen

cons

ider

atio

n sh

ould

be

give

n to

whe

ther

the

in

accu

raci

es in

the

DTM

are

com

men

sura

te w

ith

the

impr

oved

acc

urac

y of

the

2D a

ppro

ach.

The

co

nsul

tant

may

rec

omm

end

to c

arry

out

top

o su

rvey

s of

the

flood

affe

cted

are

a us

ing

LiD

AR

or a

noth

er m

oder

n m

etho

d. E

xpla

natio

n m

ay

also

be

incl

uded

on

the

use

of a

ccur

ate

DTM

S fo

r Irr

igat

ion

infra

stru

ctur

e pl

anni

ng

by

the

Gov

ernm

ent

of N

epal

. Ad

ditio

nal

cost

of

topo

su

rvey

s m

ay b

e co

vere

d fro

m th

e co

st a

lloca

ted

to X

-ban

d ra

dars

and

AD

CPS

.

We

have

incl

uded

now

the

purc

hase

of P

LEIA

DES

sat

ellit

e im

ager

y, w

hich

DH

M h

as

alre

ady

used

(inf

orm

ed b

y D

HM

Flo

od fo

reca

ster

); th

is im

ager

y is

ava

ilabl

e up

to 5

0cm

re

solu

tion.

The

prop

osed

top

o an

d cr

oss-

sect

ion

surv

ey i

n ea

ch b

asin

will

have

mor

e cr

oss-

sect

ions

in th

e fla

t Ter

ai re

gion

. Thi

s w

as p

artic

ular

ly p

lann

ed w

here

2-d

mod

el w

ill be

bu

ilt a

nd 1

-d/2

-d m

odel

will

be li

nked

. Thi

s w

as a

lread

y m

entio

ned

in th

e re

port.

The

purc

hase

of t

he n

ew D

TM (P

LEIA

DES

) will

help

dev

elop

ing

a be

tter a

nd a

ccur

ate

2-d

mod

el in

com

bina

tion

with

cro

ss-s

ectio

nal d

ata

11

2D m

odel

ling

– co

mpu

tatio

nal t

ime

of r

eal t

ime

inun

datio

n fo

reca

stin

g: M

any

of t

he b

asin

s ar

e fa

st re

spon

ding

(12-

24 h

ours

) and

fore

cast

s w

ill ne

ed t

o be

iss

ued

with

out

unne

cess

ary

dela

y.

The

repo

rt do

es

not

outli

ne

the

estim

ated

in

crea

se

in

com

puta

tiona

l tim

e re

quire

d to

un

derta

ke t

he 2

D m

odel

ling

and

whe

ther

the

tim

e in

crea

se

is

prac

tical

fo

r th

ese

fast

re

spon

ding

cat

chm

ents

.

Sum

mar

y: t

otal

tim

e fo

r is

suin

g fo

reca

st w

ill be

abo

ut 7

0 to

75

min

utes

for

a b

asin

. Fo

reca

st m

odel

ope

ratio

ns a

re d

escr

ibed

in d

etai

ls in

Sec

tion

8.9.

30 m

inut

es o

f da

ta p

roce

ssin

g an

d an

alys

is,

30 m

inut

es o

f m

odel

run

tim

e an

d 15

m

inut

es fo

r dis

sem

inat

ing

the

fore

cast

. Ope

ratio

nal p

hase

will

be re

lativ

ely

easi

er, f

or

whi

ch D

HM

has

alre

ady

exis

ting

syst

em (

sim

ilar

to D

elft-

FEW

S);

so h

opef

ully

the

re

will

not m

uch

issu

es a

t ope

ratio

nal a

nd d

isse

min

atio

n ph

ase.

The

re s

houl

d ha

ve 2

to 3

op

erat

ors

(tech

nici

ans)

to

do t

his

rout

ine

proc

ess

each

day

dur

ing

fore

cast

sea

son

(mon

soon

). An

d co

nsul

tant

(at

lea

st o

ne i

nter

natio

nal

and

one

natio

nal)

from

thi

s pr

ojec

t will

rem

ain

avai

labl

e fu

ll tim

e fo

r thr

ee y

ears

.

The

core

wor

k of

thi

s pr

ojec

t is

the

dev

elop

men

t of

the

mod

els

(runo

ff, 1

d, 2

d an

d

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

84

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

1d/2

d lin

ked

mod

els)

.

In th

is n

ote,

we

wan

t to

men

tion

that

ther

e ar

e 2d

mod

ellin

g, fo

r flo

od fo

reca

stin

g, in

us

e in

Ind

ia (

Bagm

ati

Riv

er i

n Bi

har)

and

Aust

ralia

, w

here

run

-tim

e in

rea

l-tim

e is

ar

ound

30

min

utes

(we

have

refe

rred

the

liter

atur

e, h

ave

men

tione

d it

at a

num

ber o

f pl

aces

with

in th

e re

port

(Hux

ley,

201

6).

Onc

e th

e fo

reca

sted

rai

nfal

l will

be r

ecei

ved

on a

day

and

rea

l tim

e w

ater

leve

l and

di

scha

rge

data

will

be r

ecei

ved,

afte

r pr

oces

sing

the

dat

a (w

hich

will

also

be

auto

mis

ed li

ke in

Del

ft-FE

WS)

for m

odel

run

(runo

ff m

odel

and

hyd

rody

nam

ic m

odel

), th

e m

odel

s to

com

plet

e ru

n w

ill ta

ke a

bout

30

to 4

0 m

inut

es (

for

all

five

basi

ns,

runn

ing

from

a b

atch

file

).

12

Floo

d fo

reca

stin

g sy

stem

:

The

repo

rts

(esp

ecia

lly A

pp A

) don

’t m

ake

a cl

ear d

istin

ctio

n be

twee

n a

hydr

olog

ical

/hyd

raul

ic

mod

ellin

g sy

stem

and

a f

lood

for

ecas

ting

syst

em.

The

flood

fo

reca

stin

g sy

stem

is

th

e to

ol

whi

ch

inte

grat

es r

eal t

ime

data

, co

nduc

ts m

odel

run

s an

d

crea

tes

flood

fo

reca

sts

and

war

ning

in

clud

ing

Web

pub

licat

ions

and

SM

S al

erts

. A

Fore

cast

ing

syst

em g

ener

ally

nee

d to

car

ry o

ut

the

follo

win

g ac

tiviti

es;

o R

ead

obse

rved

hyd

rom

et a

nd r

ainf

all

fore

cast

.

o Q

ualit

y as

sura

nce

on

obse

rved

an

d fo

reca

st in

put d

ata

o D

eter

min

e ho

w to

inte

rpre

t poo

r qu

ality

or

mis

sing

dat

a (ie

rain

fall

hier

arch

y)

Dis

tinct

ion

betw

een

Hyd

rolo

gica

l and

hyd

raul

ic m

odel

:

Yes,

thi

s is

cor

rect

, in

App

endi

x A,

we

mai

nly

wan

ted

to l

ist

the

hydr

odyn

amic

m

odel

ling

softw

are,

and

then

hav

e ad

ded

in a

ny o

f the

hyd

rody

nam

ic s

oftw

are,

ther

e is

a c

oupl

ed h

ydro

logi

cal s

oftw

are.

In th

e lis

t of m

odel

ling

softw

are

(see

Tab

le 1

1), w

e w

ante

d to

men

tion

key

and

benc

hmar

ked

hydr

odyn

amic

mod

ellin

g so

ftwar

e on

ly, a

nd

wan

ted

to in

clud

e th

ose

whi

ch D

HM

use

s at

pre

sent

.

Ther

e is

no

benc

h m

arki

ng r

esea

rch

(to m

y kn

owle

dge)

for

hyd

rolo

gica

l m

odel

ling

softw

are.

How

ever

, w

e ha

ve d

escr

ibed

thr

ee k

ey h

ydro

logi

cal

mod

ellin

g so

ftwar

e:

NAM

, H

EC-H

MS

and

PDM

am

ong

whi

ch N

AM a

nd H

EC-H

MS

are

bein

g us

ed b

y D

HM

. W

e di

d no

t ai

m t

o de

scrib

e al

l hy

drol

ogic

al a

nd h

ydro

dyna

mic

mod

ellin

g so

ftwar

e av

aila

ble

arou

nd th

e w

orld

. We

are

afra

id, w

e w

ill st

rugg

le w

ith o

ur a

lloca

ted

inpu

t.

Floo

d fo

reca

stin

g sy

stem

:

We

have

now

add

ed o

n Fo

reca

stin

g sy

stem

/Too

l in

Sect

ion

8.9.

2 to

8.9

.4 a

nd D

elft-

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

85

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

o Pr

epar

e m

odel

in

put

files

, in

clud

ing

boun

dary

con

ditio

n an

d ho

tsta

rt fil

es

o Sc

hedu

le, d

istri

bute

am

ongs

t com

putin

g re

sour

ces

and

laun

ch s

imul

atio

ns

o C

arry

out

dat

a as

sim

ilatio

n

o Ex

tract

rele

vant

sim

ulat

ion

resu

lts

o D

eter

min

e st

atus

dur

ing

fore

cast

per

iod

o Pr

epar

e an

d is

sue

war

ning

s

o D

isse

min

ate

war

ning

s to

W

eb

and

crea

te S

MS

aler

ts.

o Ar

chiv

e re

sults

FEW

S in

Sec

tion

8.9.

4

Bulle

ted

item

s m

entio

ned

here

, lik

e : R

ead

obse

rved

hyd

rom

et a

nd r

ainf

all f

orec

ast,

Q

ualit

y as

sura

nce

on

obse

rved

an

d fo

reca

st

inpu

t da

ta,

data

as

sim

ilatio

n ar

e m

entio

ned

in S

ectio

n 8.

9.

Now

, ple

ase

see

belo

w b

ulle

t wis

e re

spon

se:

R

ead

obse

rved

hyd

rom

et a

nd ra

infa

ll fo

reca

st.

Plea

se s

ee b

ulle

t 1, 2

and

3 in

Sec

tion

8.9.

1

Q

ualit

y as

sura

nce

on o

bser

ved

and

fore

cast

inpu

t dat

a

Plea

se s

ee b

ulle

t 4 in

Sec

tion

8.9.

1

Q

ualit

y as

sura

nce

on o

bser

ved

and

fore

cast

inpu

t dat

a

Plea

se s

ee b

ulle

t 4 in

Sec

tion

8.9.

1

Pr

epar

e m

odel

inpu

t file

s, in

clud

ing

boun

dary

con

ditio

n an

d ho

tsta

rt fil

es

Plea

se s

ee b

ulle

t 2 to

5 in

Sec

tion

8.9.

1

C

arry

out

dat

a as

sim

ilatio

n

Plea

se s

ee S

ectio

n 8.

9.6

Ex

tract

rele

vant

sim

ulat

ion

resu

lts

Plea

se s

ee b

ulle

t 6 in

Sec

tion

8.9.

1

D

eter

min

e st

atus

dur

ing

fore

cast

per

iod

Not

und

erst

ood,

wha

t is

mea

nt b

y th

is c

omm

ent

Pr

epar

e an

d is

sue

war

ning

s

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

86

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

Plea

se s

ee b

ulle

t 1 in

Sec

tion

8.9.

5

D

isse

min

ate

war

ning

s to

Web

and

cre

ate

SMS

aler

ts.

Plea

se s

ee b

ulle

t 1 in

Sec

tion

8.9.

5

13

Floo

d fo

reca

stin

g sy

stem

:

Out

line

the

diffe

renc

e be

twee

n an

ope

n so

urce

sof

twar

e w

hich

req

uire

s co

ding

the

se f

eatu

res,

and

a

prop

rieta

ry s

oftw

are

whi

ch in

clud

es m

ost o

f the

in

-bui

lt fu

nctio

nalit

y fo

r the

se fe

atur

es

We

have

now

men

tione

d th

e fu

nctio

nalit

y of

Del

ft-FE

WS

and

use

of t

he e

xist

ing

fore

cast

mod

el o

pera

tion

syst

em w

ithin

DH

M (

see

sect

ion

8.9.

3 an

d 8.

9.4)

14

Floo

d fo

reca

stin

g sy

stem

: Pr

ovid

e a

shor

t ov

ervi

ew o

f Del

ft FE

WS

and

the

fact

that

it c

an

inco

rpor

ate

man

y di

ffere

nt

hydr

olog

ical

an

d hy

drau

lic m

odel

s, in

clud

ing

HEC

and

MIK

E.

Prov

ided

, see

Sec

tion

8.9.

4

Also

pro

vide

d be

low

for r

eady

refe

renc

e

Del

ft-FE

WS

prov

ides

an

open

she

ll sy

stem

for

man

agin

g fo

reca

stin

g pr

oces

ses

and/

or

hand

ling

time

serie

s da

ta

(http

s://p

ublic

wik

i.del

tare

s.nl

/dis

play

/FEW

SDO

C/H

ome)

. Th

e fo

reca

stin

g sy

stem

was

es

sent

ially

bui

lt as

a s

hell

arou

nd th

e hy

drol

ogic

al a

nd h

ydra

ulic

mod

els

used

(Wer

ner

et a

l., 2

012)

. Th

e sy

stem

con

tain

s no

mod

ellin

g ca

pabi

litie

s (ra

infa

ll-ru

noff

and

hydr

odyn

amic

mod

ellin

g) w

ithin

its

code

bas

e. In

stea

d, it

ent

irely

rel

ies

on th

ird p

arty

m

odel

ling

com

pone

nts

for r

ainf

all-r

unof

f and

hyd

rody

nam

ic m

odel

ling.

The

stru

ctur

e of

th

e D

elft-

FEW

S in

clud

es a

dat

a st

orag

e la

yer,

a da

ta a

cces

s la

yer,

as w

ell a

s se

vera

l co

mpo

nent

s fo

r im

porti

ng, m

anip

ulat

ing,

vie

win

g an

d ex

porti

ng d

ata.

The

stru

ctur

e of

D

elft-

FEW

S is

sho

wn

in F

igur

e 16

.

Cur

rent

ly D

elft-

FEW

S is

use

d in

ove

r 40

cou

ntrie

s ov

er t

he w

orld

. D

elft-

FEW

S ca

n ei

ther

be

depl

oyed

in

a st

and-

alon

e, m

anua

lly d

riven

env

ironm

ent,

or i

n a

fully

au

tom

ated

dis

tribu

ted

clie

nt-s

erve

r env

ironm

ent.

DH

M, i

n th

eir e

xist

ing

FFEW

S, (e

.g.,

in K

osi

and

Bagm

ati

by h

ydro

logi

cal

and

hydr

aulic

mod

ellin

g, a

nd i

n Ka

rnal

i an

d N

aray

ani b

y pr

obab

ilistic

mod

ellin

g) u

ses

thei

r ow

n op

erat

iona

l sys

tem

(D

HM

, 201

8).

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

87

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

Sim

ilarly

, Ban

glad

esh

and

Indi

a (B

ihar

), in

thei

r flo

od fo

reca

stin

g m

odel

s de

velo

ped

in

NAM

and

MIK

E11

and

HEC

-HM

S an

d H

EC-R

AS,

also

use

the

ir ow

n op

erat

iona

l sy

stem

.

15

Ope

ratio

nal

FFEW

S :

Prov

ide

a su

mm

ary

over

view

of

how

eac

h FF

EWS

will

oper

ate,

in

term

s of

;

o C

over

age

(hyd

rolo

gica

l, 1D

hyd

raul

ic,

linke

d 1D

-2D

hyd

raul

ic)

o In

put

data

(ob

serv

ed)

and

topo

(x/

s &

2D).

Wha

t w

ill th

e gr

idde

d m

eteo

rolo

gica

l dat

a be

use

d fo

r ?

o Q

uant

itativ

e pr

ecip

itatio

n fo

reca

st

(QPF

), al

so

cons

ider

ing

DH

M’s

W

RF

mod

el

resu

lts

o Fr

eque

ncy

of fo

reca

st,

o H

indc

ast a

nd fo

reca

st h

oriz

on,

o Fo

reca

st p

oint

s,

o Fo

reca

st d

eliv

erab

les

o Ti

min

g,

incl

udin

g la

tenc

y,

pre-

proc

essi

ng,

runt

ime,

pos

t-pro

cess

ing,

for

ecas

t pr

epar

atio

n,

fore

cast

ap

prov

al,

fore

cast

is

suan

ce (n

otin

g th

at y

ou w

ill ne

ed to

coo

rdin

ate

five

FFEW

S sy

stem

s).

We

have

inco

rpor

ated

this

com

men

t in

chap

ter 8

in d

etai

ls.

C

over

age

(hyd

rolo

gica

l, 1D

hyd

raul

ic, l

inke

d 1D

-2D

hyd

raul

ic)

Plea

se s

ee F

igur

e 10

to 1

3

All w

ater

leve

l and

wat

er a

nd d

isch

arge

gau

ges

are

fore

cast

poi

nts;

thes

e w

ill be

the

fore

cast

poi

nts

at g

auge

d lo

catio

ns. A

t un-

gaug

ed lo

catio

ns, e

ach

com

puta

tiona

l nod

e w

ill be

a fo

reca

st p

oint

, app

roxi

mat

ely

300

to 4

00m

apa

rt al

ong

the

river

In

put d

ata

(obs

erve

d) a

nd to

po (x

/s &

2D

). W

hat w

ill th

e gr

idde

d m

eteo

rolo

gica

l dat

a be

use

d fo

r ?

Grid

ded

data

will

be u

sed

from

APH

RO

DIT

E, T

RM

M a

nd IM

D (

in c

ase

of IM

D, D

HM

w

ill re

quire

a tr

eaty

with

Indi

a fo

r usi

ng th

eir d

ata)

Fr

eque

ncy

of fo

reca

st

Dai

ly o

nce

or m

ore

durin

g hi

gh o

r mul

tiple

pea

k, p

leas

e bu

llet 5

in S

ectio

n 8.

9.1

H

indc

ast a

nd fo

reca

st h

oriz

on

Seve

n da

ys: 4

day

s fo

r hin

dcas

t and

thre

e da

ys fo

r for

ecas

t. Pl

ease

bul

let 3

in S

ectio

n 8.

9.1

Fore

cast

del

iver

able

s

Wat

er le

vel,

disc

harg

e at

all

fore

cast

poi

nts

and

floo

d in

unda

tion

map

(Ple

ase

bulle

t 6

in S

ectio

n 8.

9..1

), an

d ye

arly

eva

luat

ion

repo

rt on

for

ecas

ting

perfo

rman

ce (

Plea

se

see

Sect

ion

8.1)

Ti

min

g, in

clud

ing

late

ncy,

pre

-pro

cess

ing,

runt

ime,

pos

t-pro

cess

ing,

fore

cast

pr

epar

atio

n, fo

reca

st a

ppro

val,

fore

cast

issu

ance

(not

ing

that

you

will

need

to

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

88

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

co

ordi

nate

five

FFE

WS

syst

ems)

.

Thes

e ar

e co

vere

d be

twee

n bu

llet 1

to b

ulle

t 6 in

Sec

tion

8.9.

1 an

d al

so s

ee 8

.9.2

last

pa

ra

16

Cos

t of

m

odel

ling

softw

are

: Th

e co

st

of

mod

ellin

g so

ftwar

e is

list

ed a

s U

SD13

,000

per

ba

sin,

as

sum

ing

the

sam

e so

ftwar

e ca

n be

us

ed f

or t

he f

ive

basi

ns,

tota

ling

USD

65,0

00.

Wha

t abo

ut a

nnua

l mai

nten

ance

cos

ts?

Shou

ld

freew

are

softw

are

be u

sed

then

this

pric

e w

ould

be

ni

l, bu

t th

ere

may

be

ex

tens

ive

codi

ng

requ

irem

ents

. Th

e w

ay t

he c

ost

of m

odel

ling

softw

are

is

pres

ente

d do

es

not

refle

ct

this

. Be

caus

e of

thi

s, p

erha

ps it

sho

uld

be in

clud

ed

as a

sep

arat

e ite

m.

If H

EC-H

MS

and

HEC

-RAS

are

use

d, th

ey w

ill be

free

If M

IKE

is u

sed,

DH

M h

as a

lread

y ha

ve L

icen

ses;

the

y w

ill ne

ed s

ome

addi

tiona

l bu

dget

to

get

mul

tiple

Lic

ense

and

get

the

late

st r

elea

se o

f M

IKE

(rele

ase

2016

or

late

r if a

lread

y av

aila

ble)

In c

ase

of T

UFL

OW

, Flo

od M

odel

ler p

ro o

r Inf

owor

ks IC

M o

r SO

BEK,

the

softw

are

will

have

to

be p

urch

ased

, an

d th

us U

SD 6

5,00

0 ha

ve b

een

kept

; w

e ha

ve t

aken

thi

s pr

ice

from

the

WBM

hom

e pa

ge

17

Cos

ts:

The

exec

utiv

e su

mm

ary

shou

ld i

nclu

de

all c

ost

com

pone

nts,

ie t

opo

surv

ey,

hydr

omet

an

d ra

infa

ll pr

ocur

emen

t an

d O

&M,

mod

ellin

g so

ftwar

e pr

ocur

emen

t and

cod

ing

Incl

uded

in E

S, p

leas

e se

e Ta

ble

1 to

6

18

Res

pons

e tim

e :

Sect

ion

1 sh

ould

out

line

the

resp

onse

tim

es

(eg

time

of

conc

entra

tion,

la

gtim

e) a

t var

ious

loca

tions

in th

e ba

sin.

Thi

s is

im

porta

nt

to

appr

ecia

te

the

need

fo

r ra

pid

fore

cast

s.

Hav

e in

clud

ed it

with

refe

renc

e in

Sec

tion

1.5,

par

a 2.

19

Dat

a flo

w :

A d

iagr

am o

utlin

ing

the

data

flo

ws

(from

ob

serv

ed

to

harv

este

d by

FF

EWS

to

diss

emin

atio

n) w

ould

be

help

ful.

Prov

ided

, see

Fig

ure

14 a

nd s

uppo

rting

text

s in

Sec

tion

8.9.

2, b

ulle

t 1, 2

and

3

20

Wat

er-le

vel g

auge

: D

HM

will

take

ow

ners

hip

of

We

have

pro

pose

d th

ree

type

s of

gau

ges;

dep

endi

ng o

n th

e si

te c

ondi

tion

and

DH

M’s

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

89

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

th

e ga

uges

. Li

aise

with

DH

M o

n w

hat

type

of

gaug

e th

ey w

ill ac

cept

, si

nce

they

mig

ht o

nly

wan

t do

wnw

ard-

look

ing

rada

r co

mpa

tible

to

th

eir r

ecen

tly in

stal

led

real

tim

e sy

stem

in m

any

basi

ns.

pref

eren

ce,

gaug

e ty

pe w

ill be

sel

ecte

d at

eac

h si

te.

The

thre

e ty

pes

of g

auge

co

nsid

erat

ion

is a

lso

base

d on

spe

cific

atio

ns g

iven

by

RTS

, th

e Ka

thm

andu

bas

ed

hydr

omet

equ

ipm

ent

prov

ider

. Bo

th A

DB

and

DH

M a

dvis

ed t

o co

ntac

t R

TS;

DH

M

how

ever

men

tione

d us

to ta

ke n

ote

that

RTS

is o

nly

thei

r ven

dor

21

O&

M c

osts

: O

utlin

e w

heth

er th

is is

per

yea

r or

over

a n

umbe

r of y

ears

(how

man

y?).

Hav

e m

ade

them

cle

ar in

eve

ry b

asin

rep

ort,

see

Tabl

e 1

to 6

and

als

o pl

ease

see

Fe

asib

ility

Rep

ort,

Cha

pter

6 a

nd A

ppen

dix

E

22

Num

ber

of h

ydro

met

gau

ges

: En

sure

rep

orts

in

dica

te

inte

rnal

ly-c

onsi

sten

t nu

mbe

rs

of

hydr

omet

gau

ges.

The

Moh

ana-

Khut

iya

repo

rt is

inco

nsis

tent

.

Hav

e co

rrect

ed t

hem

in

each

bas

in r

epor

t. So

rry t

hat

we

mad

e so

me

mis

take

s in

nu

mbe

rs, m

ainl

y du

e to

cop

y an

d pa

stin

g

23

Wat

er-le

vel

and

disc

harg

e bu

dget

:

mea

sure

men

t co

st

and

an

O&M

co

st

are

sepa

rate

, w

hat’s

th

e di

ffere

nce

betw

een

mea

sure

men

t and

ope

ratio

n ?

Hav

e cl

arifi

ed it

, Men

tione

d in

not

e (c

) in

Tabl

e 3

and

Tabl

e 16

.

24

Rea

l tim

e da

taba

se (

SCAD

A): I

nclu

de th

e ne

ed

to

deve

lop

a re

al

time

data

ba

se

syst

em

inte

grat

ed

to

DH

M’s

ex

istin

g da

taba

se.

Upd

atin

g an

d m

aint

enan

ce o

f th

e da

taba

se b

y D

HM

sho

uld

also

des

crib

ed

We

will

use

DH

M e

xist

ing

real

tim

e da

taba

se s

yste

m.

We

have

add

ed a

sec

tion

( see

sec

tion

8.9.

2) t

hat d

ata

from

new

tele

met

ric g

auge

s w

ill be

tra

nsm

itted

to

DH

M’s

ser

ver;

how

ever

, co

nsul

tant

for

thi

s pr

ojec

t w

ill do

the

ch

ecki

ng, a

naly

sis

and

qual

ity a

ssur

ance

for

thes

e ne

w g

auge

s; w

e ha

ve c

onsi

dere

d bu

dget

for t

his

25

Floo

d fo

reca

stin

g ap

proa

ches

: T

he t

abul

ated

flo

od f

orec

astin

g ap

proa

ches

(Ta

bles

12,

13,

14

, 15

& 1

6) a

re n

ot c

lear

. Pl

an/s

chem

atic

di

agra

ms

wou

ld m

ake

this

cle

arer

.

Plan

map

add

ed, p

leas

e se

e Fi

gure

s (m

aps)

10

to 1

3

26

Dis

sem

inat

ion

: pr

ovid

e a

sect

ion

on f

orec

ast

diss

emin

atio

n, ie

wha

t, w

hen,

how

, etc

Ad

ded,

ple

ase

see

Sect

ion

8.9,

bul

let 1

to 5

and

als

o Se

ctio

n 8.

9.5

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

90

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

27

FFEW

S pr

ogra

mm

e : D

o no

t rep

lace

gau

ge-to

-ga

uge

fore

cast

ing,

kee

p it

as a

n al

tern

ativ

e (b

acku

p)

Com

plie

d; g

auge

to g

auge

cor

rela

tion

will

be m

aint

aine

d in

par

alle

l to

othe

r too

l

28

FFEW

S pr

ogra

mm

e : t

o Fi

gure

8 a

dd/c

larif

y;

o G

auge

inst

alla

tion

perio

d

o D

efin

e w

hat Q

1 to

Q8

are

o R

ainf

all-r

unof

f is

a st

anda

lone

act

ivity

o En

sure

cal

ibra

tion

data

are

ava

ilabl

e fo

r ca

libra

tion

activ

ity.

Ther

e m

ay

need

to

be

an

othe

r ca

libra

tion/

valid

atio

n ac

tivity

nea

r th

e en

d w

hen

mor

e da

ta is

ava

ilabl

e.

o En

sure

al

l fiv

e pr

ogra

mm

es

are

not

coin

cide

nt, t

here

sho

uld

be s

ome

stag

ger i

n th

e pr

ogra

mm

es

Cla

rifie

d,

Reg

ardi

ng m

odel

dev

elop

men

t for

five

bas

ins,

we

have

con

side

red

suffi

cien

t res

ourc

e in

put (

3 in

tern

atio

nal a

nd fi

ve n

atio

nal e

xper

ts. T

hus,

the

stag

gerin

g of

act

iviti

es w

ill re

mai

n up

to h

e co

nsul

tant

. The

aim

in th

e pr

ogra

mm

e th

at m

odel

for

each

bas

in w

ill go

in p

aral

lel a

nd s

houl

d be

cal

ibra

ted

and

valid

ated

for

all

thre

e ye

ars,

and

eac

h ba

sin

mod

el w

ill be

han

ded

over

to D

HM

at t

he e

nd o

f 36th

mon

th

29

FFEW

S m

aint

enan

ce

: pr

ogra

mm

e sh

ould

in

clud

e at

lea

st a

3-y

ear

mai

nten

ance

per

iod.

M

aybe

ev

en

a 5-

year

pe

riod

mai

nten

ance

pe

riod.

O

utlin

e w

hat

is

requ

ired

in

the

mai

nten

ance

per

iod;

o Pr

e-se

ason

set

up (2

wee

ks)

o Ea

rly s

easo

n as

sist

ance

(2 w

eeks

)

o O

n-ca

ll tro

uble

-sho

otin

g du

ring

flood

se

ason

o Po

st-s

easo

n re

view

(2 w

eeks

)

Dur

ing

the

perio

d of

dev

elop

men

t of

thi

s pr

ojec

t, w

hich

is t

hree

yea

rs,

ther

e w

ill be

op

erat

ion

and

mai

nten

ance

wor

k an

d co

nsul

tant

will

rem

ain

avai

labl

e fu

ll tim

e fo

r any

su

ppor

t inc

ludi

ng tr

oubl

e sh

ootin

g

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

91

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Ref

eren

ce

Com

men

ts fr

om A

DB

R

eply

con

sulta

nt

30

Coo

rdin

atio

n of

fiv

e FF

EWS

syst

ems

: ex

plai

n ho

w

the

over

all

syst

em

will

coor

dina

te

five

indi

vidu

al F

FEW

S sy

stem

s.

Thes

e ar

e fiv

e di

ffere

nt b

asin

s an

d fiv

e di

ffere

nt m

odel

s. In

our

sch

edul

e, a

ll fiv

e ba

sin

mod

el w

ill be

dev

elop

ed in

par

alle

l. W

e ha

ve in

clud

ed s

uffic

ient

and

logi

cal h

uman

re

sour

ce (i

nput

) for

this

(see

Sec

tion

8.13

).

For o

pera

tion

and

diss

emin

atio

n, a

s th

ese

will

be a

utom

ated

sys

tem

(exi

stin

g sy

stem

of

DH

M –

the

new

mod

els

will

be c

usto

mis

ed t

o th

e sy

stem

), th

e op

erat

ion

and

diss

emin

atio

n tim

e is

ver

y m

inim

al,

for

five

mod

els

(or

10 m

odel

s),

the

run

will

be

mad

e th

roug

h a

batc

h fil

e. F

or t

he f

irst

thre

e ye

ars,

all

inte

rnat

iona

l an

d na

tiona

l ex

perts

to

geth

er

with

DH

M

expe

rts

will

rem

ain

avai

labl

e. A

fter

3rd

year

, D

HM

co

ntin

ues.

The

y ha

ve s

peci

alis

ts w

ho w

ork

on s

hifts

(DH

M, 2

018)

31

CBR

DM

: Inc

lude

CBR

DM

act

iviti

es in

the

Wes

t R

apti

basi

n ,

as

this

as

pect

w

as

not

fully

co

vere

d in

th

e W

orld

Ba

nk

PPC

R

proj

ect.

CBR

DM

sho

uld

also

foc

us o

n th

e ro

le o

f lo

cal

mun

icip

aliti

es.

Incl

uded

. It i

s in

a s

epar

ate

repo

rt on

CBD

RM

for a

ll si

x ba

sins

32

Floo

d Sh

elte

rs:

Incl

ude

Floo

d Sh

elte

rs in

Wes

t R

apti

Basi

n.

Incl

uded

, ple

ase

see

CBD

RM

Rep

ort,

it is

a s

epar

ate

repo

rt

33

Evac

uatio

n ro

ute:

Inc

lude

the

nee

d to

pre

pare

ev

acua

tion

rout

es in

all

the

basi

ns b

ased

on

2D

mod

el re

sults

and

road

net

wor

k.

Incl

uded

, ple

ase

see

CBD

RM

Rep

ort;

it is

a s

epar

ate

repo

rt

34

Floo

d R

isk:

Inc

lude

the

pro

visi

on o

f flo

od r

isk

asso

ciat

ed w

ith a

n flo

od fo

reca

st e

vent

, so

that

co

mm

uniti

es

and

Gov

ernm

ent

agen

cies

ar

e aw

are

of th

e ris

ks.

Yes,

this

is in

-bui

lt de

liver

able

; flo

od r

isk

map

s w

ill be

issu

ed d

aily

, irr

espe

ctiv

e of

a

maj

or e

vent

or

regu

lar

flow

; Pl

ease

see

bul

let

6 in

Sec

tion

8.9.

1 w

here

we

have

m

entio

ned

flood

map

as

deliv

erab

le in

eac

h fo

reca

st ru

n

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

92

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Tabl

e B

.2: C

omm

ents

on

FFEW

S R

epor

ts fr

om W

RPPF

(Dat

e of

rece

ipt:

13/0

1/20

19)

Com

men

ts fr

om W

RPP

F w

ere

gene

ric fo

r the

repo

rts o

n fiv

e ba

sins

: Moh

ana-

Khut

iya,

Maw

a-R

atuw

a, L

akha

dei,

Bakr

aha

and

East

Rap

ti

Ref

eren

ce

Com

men

ts

from

: W

RPP

F;

Rec

eive

d on

: 13

/01/

2019

R

eply

con

sulta

nt

Bul

let 1

Th

e la

ngua

ge

of

the

repo

rt ne

eds

to

be

impr

oved

ed

ited

in

stan

dard

fo

rmat

as

te

xt

writ

ing

in a

dditi

on t

o gr

amm

atic

ally

cor

rect

. Al

l th

e te

xt is

to b

e th

orou

ghly

che

cked

.

We

have

gon

e th

roug

h th

e re

port

thor

ough

ly,

have

im

prov

ed t

exts

and

cor

rect

ed

gram

mat

ical

erro

r

Reg

ardi

ng s

tand

ard

form

at a

s te

xt w

ritin

g, t

he r

epor

t ha

s be

en w

ritte

n in

Mot

t M

acD

onal

d’s

stan

dard

tem

plat

e; s

o w

e ho

pe th

e fo

rmat

is a

lrigh

t

Bul

let 2

Ab

brev

iate

d w

ords

are

to b

e st

anda

rdis

ed

We

have

sta

ndar

dise

d ab

brev

iate

d w

ords

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

93

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

Bul

let 3

Th

e re

ports

are

nic

e bu

t I

foun

d so

me

typi

ng

erro

rs a

nd s

ome

tech

nica

l iss

ues

like

inst

alla

tion

of l

ot o

f ra

in g

auge

inc

ludi

ng X

-ban

d ra

dar

inst

alla

tion

and

wat

er

leve

l se

nsor

st

atio

ns

sim

ilar

to t

hat

of U

K an

d Au

stra

lian

stan

dard

. Te

chni

cal

disc

ussi

ons

are

requ

ired

abou

t pr

opos

ed lo

catio

ns a

nd n

umbe

r of s

tatio

ns.

Mai

n is

sue

is s

usta

inab

ility

of t

he s

tatio

ns a

nd

syst

em fo

r ope

ratio

n an

d m

aint

enan

ce a

fter

the

proj

ect.

Prop

osed

wat

er l

evel

sen

sors

lik

e flo

atin

g in

st

illing

wel

l and

pre

ssur

e ai

r bub

ble

sens

ors

are

not o

pera

ble

in h

igh

sedi

men

t loa

d riv

er.

The

hydr

o-m

etric

st

atio

ns

may

ne

ed

spec

ial

stru

ctur

es li

ke fl

ood

pilla

r in

river

cha

nnel

hav

ing

high

fluc

tuat

ions

of w

ater

leve

ls a

nd c

hang

e of

riv

er c

hann

els

as w

ell.

The

deta

il de

sign

of

cabl

eway

is

not

incl

uded

Than

ks fo

r you

r app

reci

atio

n ab

out t

he re

port,

we

have

rem

oved

typi

ng e

rrors

.

Num

ber

of g

roun

d ba

sed

rain

gau

ge s

tatio

ns h

ave

been

dec

ided

bas

ed o

n re

sear

ch

reco

mm

enda

tions

and

bas

ed o

n pr

actic

e, e

.g.,

in E

urop

e (p

leas

e se

e Se

ctio

n 2.

2 of

th

e re

port

for r

efer

ence

). Ke

epin

g in

min

d th

e fu

ture

mai

nten

ance

, the

tota

l num

ber o

f st

atio

ns h

ave

been

cho

sen

on th

e hi

gher

end

of t

he re

com

men

ded

spat

ial d

istri

butio

n.

For f

lood

fore

cast

ing

purp

ose,

one

sta

tion

with

in 1

0 to

100

km

2 is re

com

men

ded.

Our

pr

opos

ed d

istri

butio

n is

one

in

arou

nd 1

00 k

m2 .

The

area

is

even

hig

her

per

one

gaug

e in

big

ger

basi

n lik

e Ea

st a

nd W

est

Rap

ti. B

ased

on

the

perfo

rman

ce o

f th

e FF

EWS

mod

els,

whi

ch w

ill be

dev

elop

ed in

this

stu

dy, w

e ho

pe th

at D

HM

, in

futu

re,

can

add

few

mor

e st

atio

ns b

ringi

ng th

e st

atio

n de

nsity

like

in th

e U

K.

We

have

app

rised

the

num

bers

of g

auge

s an

d th

eir l

ocat

ions

to D

HM

thro

ugh

seve

ral

mee

tings

; w

e al

so

sent

th

ese

docu

men

ts

(Cha

pter

5

and

6)

to

DH

M;

they

re

com

men

ded

to d

ecid

e on

the

num

bers

bas

ed o

n hy

dro-

met

eoro

logi

cal c

limat

e in

N

epal

, w

hich

we

have

com

plie

d th

roug

h lit

erat

ure

and

rese

arch

rev

iew

s. F

urth

er t

o th

is,

we

also

pub

lishe

d tw

o ba

sins

rep

orts

(FS

Rep

ort)

for

M-K

and

M-R

bas

ins

in

adva

nce

in J

uly

2018

sho

win

g th

is s

patia

l dis

tribu

tion

of r

ain

gaug

es,

and

rece

ived

fe

edba

cks

from

AD

B an

d D

OI

and

impl

emen

ted

thos

e in

fin

alis

ing

rain

gau

ge

num

bers

in a

ll si

x ba

sins

.

Reg

ardi

ng X

-Rad

ar r

ain

gaug

e, w

e ha

ve n

ow d

ropp

ed t

his

item

fol

low

ing

ADB;

s ad

vice

(see

AD

B’s

com

men

t 3 in

Tab

le B

.1)

As o

f the

pro

pose

d ne

w s

tatio

ns a

re a

utom

ated

tele

met

ric s

tatio

ns, t

hey

have

min

imal

op

erat

ion

cost

. Th

e m

aint

enan

ce c

ost

is a

lso

min

imal

, w

hich

we

belie

ve N

epal

G

over

nmen

t/DH

M w

ill co

ntin

ue fr

om th

eir a

nnua

l bud

get a

fter t

his

proj

ect i

s co

mpl

eted

We

have

pro

pose

d th

ree

type

s: s

enso

r in

stil

ling

wel

l, ai

r bu

bble

s se

nsor

s an

d ra

dar

sens

or;

deep

enin

g on

site

con

ditio

n an

d D

HM

’s p

refe

renc

e, t

he g

auge

typ

e w

ill be

se

lect

ed d

urin

g pr

ocur

emen

t and

inst

alla

tion.

The

budg

et in

clud

es a

ll in

frast

ruct

ures

for i

nsta

llatio

n of

a g

auge

, rai

n ga

uge

or w

ater

le

vel g

auge

. We

have

men

tione

d th

is in

the

repo

rt (p

leas

e se

e Se

ctio

n 5.

5, p

ara

1) a

s fo

llow

s: th

e bu

dget

incl

udes

pro

cure

men

t, in

stal

latio

n, te

stin

g, c

alib

ratio

n, m

onito

ring,

an

d op

erat

ion

and

mai

nten

ance

for 3

yea

rs

Mot

t Mac

Don

ald

| WR

PPF:

Pre

para

tion

of P

riorit

y R

iver

Bas

ins

Floo

d R

isk

Man

agem

ent P

roje

ct, N

epal

94

Floo

d Fo

reca

stin

g an

d Ea

rly W

arni

ng S

yste

m: M

ohan

a –

Khut

iya

Basi

n 38

3877

| R

EP |

0040

| 4

April

201

9 Fl

ood

Fore

cast

ing

and

Early

War

ning

Sys

tem

: Moh

ana

– Kh

utiy

a Ba

sin

4 R

egar

ding

hy

drol

ogic

al

mod

ellin

g an

d flo

od

early

w

arni

ng

syst

em

deve

lopm

ent,

list

of

mod

els

are

prov

ided

bu

t no

t re

com

men

ded

final

ly a

lthou

gh t

hey

have

est

imat

ed c

ost

for

softw

are

purc

hase

. The

dev

elop

men

t cos

t of t

he

hydr

olog

ic a

nd h

ydra

ulic

mod

el s

hall

redu

ce

sign

ifica

ntly

onc

e it

is b

uilt

for

a ba

sin

and

just

re

plic

ate

with

oth

er w

ith s

mal

l ch

ange

s ba

sed

on d

ata

avai

labi

lity.

The

cost

of

land

acq

uisi

tion

and

fenc

ing

and

civi

l wor

ks e

tc. n

ot m

entio

ned.

Cha

pter

8, i

n Ta

ble

19 to

23,

spe

cific

reac

h-w

ise

mod

els

have

bee

n pr

opos

ed..

Mod

ellin

g co

st:

The

hydr

olog

ical

and

hyd

raul

ic m

odel

s fo

r ea

ch b

asin

are

sep

arat

e m

odel

; i.e

., se

para

te m

odel

set

-up

prep

arat

ion,

sep

arat

e ca

libra

tion

and

valid

atio

n; th

us, t

he c

ost

is a

lmos

t si

mila

r am

ong

the

basi

ns;

how

ever

, as

exp

ert

will

beco

me

expe

rienc

ed

thro

ugh

wor

k in

one

bas

in, t

he c

ost w

ill sl

ight

ly r

educ

e in

oth

er b

asin

s, a

nd w

e ha

ve

cons

ider

ed th

is fa

ctor

, ple

ase

see

text

s ex

plai

ning

this

in S

ectio

n 8.

13

All c

ivil

wor

ks a

nd fe

ncin

g ar

e in

clus

ive

in th

e co

st; p

leas

e se

e Se

ctio

n 5.

5, p

ara

1. W

e as

sum

e th

at a

ll ra

in g

auge

s w

ill be

ins

talle

d at

Gov

ernm

ent/P

ublic

pre

mis

es,

e.g.

, D

HM

O

ffice

pr

emis

e,

DW

IDM

of

fice

prem

ise,

Lo

cal

mun

icip

ality

pr

emis

e,

Adm

inis

tratio

n O

ffice

pre

mis

es, G

ram

Pan

chay

at e

tc.

Mott MacDonald | WRPPF: Preparation of Priority River Basins Flood Risk Management Project, Nepal 1Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

383877 | REP | 0040 | 4 April 2019 Flood Forecasting and Early Warning System: Mohana – Khutiya Basin

mottmac.com