MAP D-PHASE Forecast Demonstration Project Instrument of WWRP Last phase of MAP Mathias Rotach,...

MAP D-PHASE

• Forecast Demonstration Project• Instrument of WWRP • Last phase of MAP

Mathias Rotach, MeteoSwiss

What is an FDP?

WWRP instrument> Forecast of weather of international relevance

high-impact weather

> Clear evaluation protocols

> Expectation of success

> Clear advance in operational forecast. Proposal to WWRP

Components of MAP FDP

Forecast heavy precipitation events in the alps Include assimilation schemes (e.g., for radar

data) Coupling hydro/meteo models End-user specific products

develop and verify on MAP cases experience with operational MC-2 (3km resolution during MAP SOP)

Forecast Demonstration period

General Goals

Demonstrate ability for improved forecast of heavy precipitation in the alps

High-resolution atmospheric modelling ensemble forecast technique

Radar data (assimilation)

Hydrological modeling End users involved (end user needs, e.g.

probabilistic forecasts) Evaluation protocols (yet to be determined)

D-PHASE….

Demonstration of Probabilistic Hydrological and Atmospheric

Simulation of Flooding Events in the Alps

D-PHASE essentials

Forecast (warning) chain for heavy precipitation events in the alps

Test in pre-defined ‘Forecast Period’ MAP season (Sept to November): 2006 (7?) extension to COPS (DFG field program on QNV in summer 07)?

Joint Target area: Lago Maggiore – various sub-areas

Strategy for D-PHASE

Atmosphere - Multi-component approach:

1. Local EPS systems (COSMO-LEPS, LAMEPS, PEPS,..) 3-5 days probabilistic forecast likelihood of ‘event’

Probability maps for 24 h precipitation

André Walser & COSMO-LEPS

Strategy for D-PHASE

Atmosphere - Multi-component approach:

1. Local EPS systems (COSMO-LEPS, LAMEPS, PEPS,..) 3-5 days probabilistic forecast likelihood of ‘event’

2. ‘standard’ deterministic models at high resolution (1-3km) short-range, targeted coupled hydrological models latest radar information assimilated

3. A possible ‘micro-LEPS’ made up as a poor man’s EPS from the above probabilistic information on hydrol. patterns.

Strategy for D-PHASE

Hydrosphere:

1. Hydrological models distributed coupled

2. Assimilation of latest information radar composites rain gauges

3. Probabilistic forcing from atmospheric models from radar (obs) uncertainty

Using COSMO-LEPS for hydrological forecasting

COSMO-LEPS Setup:

quasi-operational since November 2002

initial and boundary cond. from ECMWF EPS

Integration period: 120 h

Model: Lokal Modell (LM)

Grid-spacing: 10 km, 32 levels

Hydrological Model Setup:

Rhine upstream of Rheinfelden (34550 km2)

Model: PREVAH (Gurtz et al. 1999)

Driven by LM / COSMO-LEPS

A few test cases with 51 EPS members

AARE(Thun) ILL

(Gisingen)

Mark Verbunt, Christoph Schär ETH Zürich & COSMO-LEPS crew

Deterministic Runoff Forecast

observedrunoff

Ru

no

ff [

mm

/h]

Aare - Thun (2490 km2)

1970:~400 m3/s

1999:~570 m3/s

initialization

deterministic forecast

hydrological model driven bysurface observations

surface observations

simulated runoff

Mark Verbunt, Christoph Schär ETH Zürich & COSMO-LEPS crew

Deterministic versus Probabilistic Forecast

Ru

no

ff [

mm

/h]

AARE - Thun (2490 km2)

51 member LEPS forecast

hydrological model driven bysurface observations

surface observations

no precip

initialization

(Mark Verbunt, ETH Zürich)Mark Verbunt, Christoph Schär ETH Zürich & COSMO-LEPS crew

Deterministic versus Probabilistic Forecast

Ru

no

ff [

mm

/h]

ILL – Gisingen

51 member LEPS forecast

hydrological model driven bysurface observations

surface observations

initialization

Mark Verbunt, Christoph Schär ETH Zürich & COSMO-LEPS crew

Representative membersR

un

off

[m

m/h

]

5 EPS members (COSMO-LEPS clustering)

10 EPS members (COSMO-LEPS clustering)

51 EPS members

Ru

no

ff [

mm

/h]

Ru

no

ff [

mm

/h]

(Mark Verbunt, ETH Zürich)

ILL – Gisingen

Mark Verbunt, Christoph Schär ETH Zürich & COSMO-LEPS crew

Strategy for D-PHASE

End users:

1. Authorities civil protection river/lake management

2. THEIR needs thresholds cost/loss

--> end-to-end flood forecasting system

Event?Where?

yes

HI-ressimulation

Alert

Day -4

ECMWF EPS

LAMEPS

Day 0Latest obs.

Nowcasting

Latest obs.Latest obs.

Action

Tim

e

regionalLEPS(COSMO) LEPS

Days -3, -2, -1Sub-domain X

no

no

HI-ressimulationHI-res

simulationHI-res

simulationHI-res

simulation

Confir-mation

ECMWF EPS

LAMEPSregionalLEPS(COSMO) LEPS

HI-resLEPS

Hydrol. simulation

Contributions

Many (more than one..) LEPS SYSTEMS Many high-res (1-4km) deterministic models

--> most covering ‚the alps‘

Contributions

Many (more than one..) LEPS SYSTEMS Many high-res (1-4km) deterministic models

--> most covering ‚the alps‘ Many hydrol. Models

--> each for one (more) catchment

Contributions

Many (more than one..) LEPS SYSTEMS Many high-res (1-4km) deterministic models

--> most covering ‚the alps‘ Quite a few hydrol. Models

--> each for one (more) catchments End users: to be determined & involved

Time table

Evaluation etc. 2007 Demonstration period: Aug to Nov 2006 (Jun/Jul

desirable) Test chain ready summer 2006 High res. models tested: end 2005 Hydrol. Models tested: end 2005 Test cases defined: May 2005 Definition of user needs: May 2005 Set up of organisation end 2005 (done) Financial resources:

Start summer 2005application asap (dependent on

what/where/etc)

Proposal WWRP: spring 2005 Start: May 2004

Time

June to Nov 2007 ??

Thank you for your attention