154
GLOWA Volta Phase II completion report Report period: 01.06.2003 – 30.05.2006 (Submitted in March 2007) Förderkennzeichen 01/LW0302A Charles Rodgers, Paul L.G. Vlek, Irit Eguavoen and Claudia Arntz (Eds.) submitted to Bundesministerium für Bildung und Forschung by the Center for Development Research, Rheinische Friedrich-Wilhelms-Universität Bonn

GLOWA Volta - ZEF...GLOWA Volta Phase II completion report Report period: 01.06.2003 – 30.05.2006 (Submitted in March 2007) Förderkennzeichen 01/LW0302A Charles Rodgers, Paul L.G

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

  • GLOWA VoltaPhase II completion report

    Report period: 01.06.2003 – 30.05.2006

    (Submitted in March 2007)

    Förderkennzeichen 01/LW0302A

    Charles Rodgers, Paul L.G. Vlek, Irit Eguavoen and Claudia Arntz (Eds.)

    submitted to

    Bundesministerium für Bildung und Forschung

    by the

    Center for Development Research, Rheinische Friedrich-Wilhelms-Universität

    Bonn

  • 2

    Outline

    Abbreviations ...........................................................................................................................4

    List of figures .................................................................................................................................10

    1. Introduction: The second project phase of GLOWA Volta: From concepts to application .......13

    1.1. The GLOWA Volta Project.................................................................................................13

    1.2. Planning and progress of the project ...................................................................................16

    1.3. Achievements in the establishment of infrastructure and capacity building.......................18

    1.3.1. Technical infrastructure and research network ............................................................18

    1.3.2. Doctoral studies............................................................................................................22

    1.3.4. Cooperation with local stakeholders ............................................................................23

    2. Achievements and research results of the sub-projects..............................................................25

    2.1. The Atmosphere Cluster..................................................................................................25

    2.1.1. Sub-project A1: Regional Climate Simulations ...........................................................25

    2.1.2. Sub-project A2: Hydro-Meteorological Monitoring System .......................................32

    2.1.3. Sub-project A3: Onset of the Rainy Season.................................................................45

    2.2. The Land Use Cluster..........................................................................................................53

    2.2.1. Sub-Project L1: Land Use Change Detection and Quantification ...............................54

    2.2.2. Sub-Project L2: Soil Characterization..........................................................................60

    2.2.3. Sub-Project L3: Vegetation Dynamics.........................................................................64

    2.2.4. Sub-Project L4: Modelling and Spatial and Temporal Up-scaling of Erosion and

    Hydrological Processes ..........................................................................................................68

    2.2.5. Sub-project L5: Land Use Change Prediction..............................................................75

    2.3. The Water Use Cluster ........................................................................................................81

    2.3.1. Sub-Project W1: Runoff and Hydraulic Routing .........................................................82

    2.3.2. Sub-Project W2: Water and Livelihood .......................................................................93

    2.3.3. Sub-Project W3: Institutional Analysis ........................................................................96

    2.4. Technical Integration and Decision Support .....................................................................102

    2.4.1. Sub-Project D1: Technical Integration of Socio-Economic and Environmental Models

    ..............................................................................................................................................102

    2.4.2. Sub-Project D2: Household Decision-Making and Policy Response ........................111

    2.4.3. Sub-Project D3: Experimental Application of Scientific Knowledge (Policy Pilot

    Study) ...................................................................................................................................113

    3. Internal progress evaluation, analysis and outlook to GVP phase III ......................................118

    3.1. Mechanisms for internal project evaluation ......................................................................118

    3.2. Analysis of achievements in project phase II ....................................................................121

  • 3

    3.3. Outlook to project phase III ..............................................................................................122

    4. Completed Ph.D. projects, project publications and research documentation .........................125

    4.1. Completed Ph.D., M.Sc., M.Phil. and Diploma theses .....................................................125

    4.2. Publications, presentations and research documentation in project phase II ....................128

    4.3. Contributions at Status Conference 2005 ..........................................................................139

    5. Appendix ..................................................................................................................................141

    5.1. Researchers and staff of the GLOWA-Volta project, Phase II .........................................141

    5.2. Selected publications (see appendix) ................................................................................153

  • 4

    Abbreviations

    AARSE African Association of Remote Sensing of

    the Environment

    AEJ African Easterly Jet

    AMMA African Monsoon Multidisciplinary Analysis

    ANN Artificial Neural Network

    APSIM Agricultural Production Systems Simulator

    ATCOR Atmospheric & Topographic Correction

    AVN Aviation Model

    BF Burkina Faso

    BIGS-DR Bonn International Graduate School for

    Development Research

    BIOTA Biodiversity Monitoring Transect Analysis

    BMBF Federal Ministry of Education and Resarch -

    Germany

    BON Biophysical Observation Network

    CA Cellular Automata

    CEC Cation Exchange Capacity

    CGIAR Consultative Group on International

    Agricultural Research

    CIDA Canadian International Development Agency

    CIRAD Centre de Coopération Internationale en

    Recherche Agronomique pour le

    Développement

    COBIDS Component-Based Integration of Data and

    Services

    CP Circulation Pattern

    CPU Central Processing Unit

    CPWF Challenge Program for Water and Food

    CR Clay Ratio

    CRFS Centre de Recherches et de Formation

    Scientifique

    CRSN Centre de Recherche en Santé de Nouna

    CsDM Diffusion and Migration Calibration Model

    CSF GVP Common Sampling Frame

  • 5

    CSIR Council for Scientific and Industrial

    Research

    CsMB1 Mass Balance 1 Calibration Model

    CsMB2 Mass Balance 2 Calibration Model

    CsPM Proportional Calibration Model

    CWSA Community Water and Sanitation Agency

    DA District Assembly

    DAAC NASA Distributed Active Archive Center

    DANIDA Danish International Development Agency

    DBM Data-based mechanistic

    DEM Digital Elevation Map

    DGIRH Direction Générale de l'Investissement dans

    les Ressources Humaines

    DLR German Aerospace Center

    DOY Day of Year

    DSS Decision Support System

    ECHAM4 Atmospheric General Circulation Model

    ECMWF European Center for Medium Range Weather

    Forecasting

    ECOWAS Economic Community of West African

    States

    EM Expectation Maximization

    EPA Environmental Protection Agency - Ghana

    ESA European Space Agency

    ET Evapotranspiration

    etr Evapotranspiration (in WasSiM)

    ETref Evapotranspiration (reference)

    FAO Food and Agriculture Organization

    GAMS General Algebraic Modelling System

    GCM General Circulation Model

    GEF Global Environment Facility

    GIRE Gestion Intégrée des Ressources en Eau

    GIS Geographic Information System

    GLOWA Global Change in the Hydrological Cycle

    GMP Governance and Modelling Project

    GPS Global Positioning System

    GUI Graphic User Interface

  • 6

    GV LUDAS GLOWA Volta Land Use Dynamics

    Simulator

    GVP GLOWA Volta Project

    GWCL Ghana Water Company Limited

    GWP Global Water Partnership

    GWP-WAWP Global Water Partnership West Africa

    HMMS Hydro-Meteorological Monitoring System

    HSD Hydrological Services Division - Ghana

    ICOUR Irrigation Committee of Upper East Region

    IFPRI International Food Policy Research Institute

    IGBP International Geosphere-Biosphere

    Programme

    IMF International Monetary Fund

    IMK-IFU Institute of Meteorology and Climate

    Research - Atmospheric Environmental

    Research Division

    INERA Institut de l’Environnement et de Recherche

    Agricoles

    ITD Inter-Tropical Discontinuity

    IUCN World Conservation Union

    IWMI International Water Management Insitute

    IWRM Integrated Water Resources Management

    KACE Kofi Annan Center of Excellence in

    Information and Communications

    Technology

    LAI Leaf Area Index

    LANDSAT ETM LANDSAT Enhanced Thematic Mapper

    LANDSAT TM LANDSAT Thematic Mapper

    LAS Large-Aperture Scintillometry

    LC Land Cover

    LCCS Land Cover Classification System

    LPJ-model Lund – Potsdam - Jena Dynamic Global

    Vegetation Model

    LSA-SAF Land Surface Analysis Satellite Application

    Facility

    LST Land Surface Temperature

    LUC Land Use / Cover

  • 7

    LUDAS Land Use Dynamics Simulator

    LULC Land Use and Land Cover

    MAS Multi-Agent System

    MATA Multi-Level Analysis Tool for Agriculture

    METOP Meteorological Operational Satellite

    MM5 Mesoscale Meteorology Model 5

    MODIS Moderate Resolution Imaging

    Spectroradiometer

    MoFA Ministry of Food and Agriculture - Ghana

    MoWH Ministry of Works and Housing -Ghana

    MSG Meteosat Second Generation

    NAO North Atlantic Oscillation

    NARMAX Nonlinear Autoregressive-moving Averages

    Model with Exogenous Input

    NASA National Aeronautics and Space

    Administration

    NCEP National Center for Environmental

    Prediction

    NCWSP National Community Water and Sanitation

    Program - Ghana

    NDVI Normalised Difference Vegetation Index

    NGO Non-Governmental Organization

    NOAA-AVHRR National Oceanic & Atmospheric

    Administration - Advanced Very High

    Resolution Radiometer

    NUSLE Nomograph Universal Soil Loss Equation

    PAGEV Projet pour l’Amélioration de la

    Gouvernance de l’Eau dans le Bassin de la

    Volta

    PbDM Diffusion and Migration Model for 210Pb

    PbMB Mass Balance Model for 210Pb

    PERL Practical Extraction and Report Language

    PEST Parameter Estimation Tool

    PPP Public Private Partnership

    prec Precipitation

    PSF Particle Size Fractions

    PSP Private Sector Participation

  • 8

    PTF Pedo-Transfer Function

    PURC Public Utilities Regulatory Commission -

    Ghana

    qdir Direct runoff (in WaSiM)

    qifl Interflow (in WaSiM)

    qtot Total runoff (in WaSiM)

    RCM Regional Climate Model

    RMSE Root Mean Square Error

    RN Net Radiation

    RS Remote Sensing

    SARI Savannah Agricultural Research Institute

    SEBAL Surface Energy Balance Algorithm for Land

    SMOS Soil Moisture and Ocean Salinity

    SOFM Second-Order-First-Moment

    SPOT Satellites Pour l’Observation de la Terre

    SR Sand ratio

    SRP Small Reservoirs Project

    SRTM Shuttle Radar Topography Mission

    SVAT Surface-Vegetation-Atmosphere-Transfer

    SVN Social Venture Network

    SVRT SMOS Validation and Retrieval Team

    SWAT Soil & Water Assessment Tool

    TEJ Tropical Easterly Jet

    TU Technical University

    UER Upper East Region - Ghana

    UML Unified Modeling Language

    UN United Nations

    USP Unit Stream Power

    USAID Unites States Agency for International

    Development

    USDS-ARS United States Department of Agriculture –

    Agricultural Research Service

    USGS United States Geological Survey

    USLE Universal Soil Loss Equation

    UTM Universal Transverse Mercator

    VBA Volta Basin Authority - Ghana

    VBTC Volta Basin Technical Committee

  • 9

    VF Vegetation Fraction

    VRA Volta River Authority -Ghana

    WAPP West African Power Pool

    WaSiM-ETH Water Balance Simulation Model

    WaTEM/ SEDEM Water and Tillage Erosion Model / Sediment

    Delivery Model

    WATSAN Water and Sanitation Committee - Ghana

    WEPP Water Erosion Prediction Project

    WRC Water Resources Commission - Ghana

    WRI Water Research Institute

    WUA Water User Association - Ghana

    WVPP White Volta Policy Pilot Project

    ZEF Center for Development Research

  • 10

    List of figures

    Figure 1 Sap flow measurement equipment ...................................................................................21

    Figure 2 Location of sub-catchments and gauges in the White Volta catchment (resolution: 1km)

    ................................................................................................................................................22

    Figure 3 Completed Ph.D., M.Sc. and M.Phil. theses in phase II ..................................................23

    Figure 4 Calculation of evaporation in 24h time step, 3h time step simulation and with improved

    calculation with 24h time step, including an empirical factor ...............................................27

    Figure 5 Model setup: Digital elevation model, sub-catchment boundaries, river network and

    location of the Sourou depression ..........................................................................................28

    Figure 6 Calibration run, coupled simulation and observed discharge values for the gauges

    Bamboi and Saboba, 1968......................................................................................................28

    Figure 7 Mean latitudal displacement (2030-2039 vs. 1991-2000) of the Inter Tropical

    Discontinuity (ITD): southward (northward) shift in April (September) connected to a

    decrease (increase) in precipitation ........................................................................................29

    Figure 8 Climate change signal in precipitation (prec), direct runoff (qdir), interflow (qifl), total

    runoff (qtot) and evapotranspiration (etr), 2030-2039 vs. 1991-2000, left: percentage change,

    right: absolute change.............................................................................................................30

    Figure 9 Monthly signal-to-noise ratio of precipitation (prec), total runoff (qtot) and

    evapotranspiration (etr) ..........................................................................................................31

    Figure 10 Measured (blue) and simulated (green) discharge [m³/s] and precipitation (black) in

    [mm] for the gauges Nakong and Nawuni for the calibration period 1968 ...........................34

    Figure 11 Simulated (green) vs. measured (blue) discharge [m³/s] at Pwalugu and Nawuni for the

    validation period 1961-67. .....................................................................................................34

    Figure 12 Simulated annual precipitation sum for 2004 (Domain 3) for the Volta Basin .............35

    Figure 13 Comparison of simulated and measured monthly precipitation sums in 2004 for

    gauging stations (Kaburi, Kpasenkpe, Pwalugu, Babile, Zuarungu) in Ghana......................35

    Figure 14 Coupled simulated (green) and measured (blue) discharge [m³/s] and precipitation

    (black) in [mm] for three gauging stations along the White Volta in Ghana (Kaburi, Pwalugu

    and Nawuni) ...........................................................................................................................36

    Figure 15 Leaf Area Index (LAI) for the White Volta catchment: mean MODIS LAI-grids for

    2002........................................................................................................................................37

    Figure 16 Annual evapotranspiration map simulated with a) standard literature and b) MODIS

    derived LAI- grids; c) difference of annual evapotranspiration: a) - b) .................................37

    Figure 17 Different components of actual evapo-transpiration, example for November 15, 2002.

    Incoming solar radiation (top left), vegetation fraction (top right), reference temperature

  • 11

    (bottom left) and resulting actual evapotranspiration (incl. evaporation from interception

    (bottom right) .........................................................................................................................40

    Figure 18 Maps of ET, sensible heat flux, and soil moisture (degree of saturation) around Tono

    dam at the end of the dry season (March 7, 2003) derived from LANDSAT image. The size

    of these maps is approximately 10 by 15 km. Spatial resolution of heat fluxes and soil

    moisture maps is 30 m............................................................................................................43

    Figure 19 Map of ET at the end of the dry season (March 7, 2003) derived from MODIS image.

    The spatial resolution of this map is 1000 by 1000 m ...........................................................44

    Figure 20 ET histograms of LANDSAT and MODIS ET maps on March 7, 2003.......................45

    Figure 21 Spatial distribution of five derived regions within the Volta Basin ..............................47

    Figure 22 Mean Onset-Date, Julian Day (left); standard deviation of onset dates (right) .............48

    Figure 23 Mean normalized MF_U distribution in 500hPa of CP5 associated to the start of the

    rains in PC1 (left) and mean normalized MF_U distribution in 500hPa of CP4, which is not

    linked to the rainy season’s onset (right) ...............................................................................49

    Figure 24 Zero- and first order Markov chain of Accra.................................................................50

    Figure 25 Observed (*) and fitted (-) proportion of a dry spell of > 6 days at Bole for each day of

    the year, based on rainfall data (1961-2001) with a threshold of 1 mm for a rainy day ........51

    Figure 26 Supplementary information to assess the date of optimal planting [DOY] for the major

    rainy season, presenting the minimum probability of the occurrence of a dry spell of > 6

    consecutive days within the following 30 days......................................................................52

    Figure 27 Sensing databases and products used within the L1 sub-project ...................................55

    Figure 28 Workplan for the LC mapping period from January 2004 – May 2006 ........................57

    Figure 29 LCCS based LC map using 26 LANDSAT images from 1990 (left) and 26

    corresponding LANDSAT tiles from 2000/2001 (right). The GLOWA Volta Basin is shown

    as a black solid line ................................................................................................................58

    Figure 30 Description of soil sampling sites ..................................................................................61

    Figure 31 Hydrotope map at Tamale study site (left), Histograms of three hydrotope soil moisture

    distributions (right).................................................................................................................63

    Figure 32 Patterns of diurnal anomalies (left), average soil moisture distribution from 1992-2000

    (right)......................................................................................................................................64

    Figure 33 Tree spatial pattern maps in 30x30m2............................................................................66

    Figure 34 Vegetation classification map of Bontiolo natural reserve ............................................67

    Figure 35 Comparison between simulated and observed latent heat fluxes (left); dependency of

    grid-scale effective roughness on subgrid scale mean and standard deviation (right) ...........69

    Figure 36 Land Cover Map of the Ioba- and the Dano catchment based on classified Aster Image

    2004........................................................................................................................................71

    Figure 37 Spatial distribution of small reservoirs in the White Volta Basin .................................72

  • 12

    Figure 38 Soil erosion maps: 210

    Pb (Pb), 137

    Cs (Cs) and Sand ratio (SR) models .........................74

    Figure 39 Users select land conservation options by changing management parameters, and

    visualize consevation plan on slope map ...............................................................................78

    Figure 40 Given the selected parameters, UPS model simulates soil erosion (map and histogram)

    as a consequences of the selected conservation option. .........................................................79

    Figure 41 Geology and groundwater potential in the Volta Basin .................................................86

    Figure 42 Hydrogeological concept model of the study area (Atankwidi catchment)...................87

    Figure 43 Daily groundwater recharge calculated by soil moisture modelling (years 2003 – 2005)

    ................................................................................................................................................89

    Figure 44 Spatial Infiltration Variation in Pwalugu and Tindama Wetlands .................................91

    Figure 45 Interaction between sub-surface and surface water in Pwalugu site .............................91

    Figure 46 Schematic of GVP Model Linkages and DSS .............................................................104

    Figure 47 Interface for linkage of simulation systems .................................................................108

    Figure 48 COBIDS-based model ensemble and interfaces ..........................................................109

    Figure 49 Proposed Geo-database, web portal and file server .....................................................111

    Figure 50 Sub-catchments and outlets delineated for MIKE BASIN ..........................................117

    Figure 51 GLOWA Volta Thematic Timeline .............................................................................122

    Figure 52 GLOWA Volta Project Phase III Planning Timeline ..................................................124

  • 13

    1. Introduction: The second project phase of GLOWA Volta:

    From concepts to application

    1.1. The GLOWA Volta Project

    The Volta River Basin, encompassing regions within six West African nations, is in many

    respects broadly representative of large basins within the “developing” world. It embodies a wide

    range of challenges to water resources management, including static or declining water

    availability as influenced by global and regional climatic and land cover processes; increasing

    demand for water resources over the coming decades in the domestic, agricultural, industrial and

    hydropower generation sectors, leading to increasing intersectoral competition for water, and the

    deterioration of water quality and supply of water to serve critical ecosystem support functions.

    Other water management challenges include a persistence of weak, inefficient or conflicting

    water rights and governance structures, an insufficient hydro-climatic database, inadequate

    scientific capacity to support effective political decision-making, and low levels of investment

    reflecting limitation in access to financial resources. Much of the research conducted within the

    framework of the GLOWA Volta Project has been designed to address, or to work around these

    constraints. Thus, our investigation of the physical and socio-economic determinants of the

    hydrological cycle within the Volta Basin, and the development of model-based decision-making

    tools, may strengthen efforts to manage water resources at river basin level not only within the

    study area, but more broadly throughout the region.

    The two dominant riparian countries of the Volta Basin, Ghana and Burkina Faso, provide the

    research context for the GLOWA Volta Project. They occupy 40.18 % and 42.65 % of the total

    400,000 square kilometre territory of the basin, respectively. They also embody the full range of

    climatic and land cover variability characterizing the Basin, spanning at least four climatic zones

    (sub-humid in the South to semi-arid in the North of the basin), annual precipitation gradient

    (2,000 mm in the southern lowland forest to less then 600 mm in the northern Sahel) and

    vegetation cover, as well as the full range of socio-cultural and economic variation. Global

    climate change is perceived as a factor likely to enhance the existing, erratic and unreliable

    character of rainfall within the basin. Projected annual population growth of 2.8 % and likely

    future expansion of irrigation schemes in northern parts of the basin will lead to a situation in

    which domestic water uses and irrigation compete directly with water use for the generation of

    hydropower at newly established dams in Burkina Faso, and at the existing downstream

    Akosombo hydropower dam on Volta Lake, which provides Ghana with over 90 % of its

  • 14

    electrical power. Reflecting on these conditions, it should be clear that we are in a position not

    only to contribute to the scientific understanding of linked climate, hydrology and land use-land

    cover patterns in a major West African basin, but also to make a substantial contribution to the

    solution of water sector- and related problems within the basin.

    The GLOWA Volta Project (GVP), conducted by the Center for Development Research (ZEF),

    Bonn University and research partners in Europe and West Africa, is conceived as a nine-year,

    multi- and trans-disciplinary study organized in three phases. Phase I (June 2000 – May 2003)

    emphasized the identification of methods and models; the establishment of research infrastructure

    on the ground in Ghana, the collection of baseline climatic, hydrological and socio-economic

    data, and the training of the first cadre of physical and social scientists from the Volta Basin and

    surrounding regions of West Africa. In phase II (June 2003 – May 2006), emphasis shifted from

    scoping studies and data collection to the development, testing and application of a range of

    numerical simulation models; as well as process-oriented studies of important bio-physical and

    socio-economic phenomena. Phase III, currently underway, will focus on integration of phase I

    and II outputs, emphasis on aggregate economic analysis, operationalization of Decision Support

    System (DSS) components and transfer of activities and responsibilities to institutions within the

    Volta Basin.

    This report is intended to provide a summary of activities conducted, and completed, during the

    second phase of the GVP. Given the size, scope and complexity of the GVP, it is impractical to

    attempt an exhaustive description of research activities and outputs realized over a three-year

    period, so this review is necessarily selective. We have attempted to highlight the most

    significant findings, and products, and we have in addition pointed to the broader range of

    contributions through the inventory of theses, publications and presentations. Similarly, although

    this completion report is organized according to the four research clusters (Atmosphere, Land

    Use, Water Use, Technical Integration) used to organize the GVP phase II proposal (2003), an

    additional range of research has been conducted or initiated during phase II that cannot easily be

    subordinated to these clusters. Examples include the contributions to land use and land cover

    change by the Wuerzburg project team, which bridge the structure of the Land Use cluster sub-

    projects; the coupled atmospheric-hydrological modelling, (original work packages A1 and W1);

    and the Multi-Agent Systems (MAS) model development conducted by Dr. Quang Bao Le,

    which spans Land Use and Water Use clusters, as originally defined. The presentation of results

    in this document still follows the structure of the phase II proposal, but many activities described

    link to several of these clusters, or occasionally fall outside of them, as in the case of the

    development of a GVP Geo-database, described under the Technical Integration (D1) activities.

  • 15

    As in any project as ambitious in scope and design as GVP, there have been project areas in

    which the quality and utility of research activities and outputs exceeded expectation; areas where

    expectations were largely met, and areas in which progress fell short of expectation. Examples of

    each will be documented, and the picture that emerges is that setbacks relative to the original

    research design have, on balance, been compensated by a range of innovative activities that serve

    to justify fully the time and resources committed to this program. In taking stock of the progress

    and accomplishments realized during phase II, as well as the challenges and setbacks, it is useful

    to consider the factors required to achieve success in meeting project objectives. A list of such

    factors would certainly include (i) financial resources; (ii) human resources; (iii) data and/or

    knowledge to support scientific enquiry; (iv) appropriate tools and technologies, including field

    instrumentation, numerical simulation tools, software engineering capacity, modes of

    communication and data storage and manipulation; (v) supporting infrastructure, including

    housing, transportation, logistics and related; (vi) appropriate institutional configurations, linking

    research partners, decision-makers and stakeholders; finally (vii) a reasonably stable political

    matrix within which to conduct the research and disseminate the results.

    Our perception is that GVP phase II was adequately funded, so that availability of funds seldom

    acted as the binding constraint to successful completion of research activities (time was more

    likely to constrain outcomes). Regarding human resources, the greatest constraint to timely

    completion of planned tasks during phase II was often staff turnover. Several cases are described

    below. This is not altogether unexpected, however, for at least two reasons. First, the GVP has

    made extensive use of young scientists, many from the Volta Basin, who have earned their

    Ph.D.’s in the context of GVP. Upon completion, they naturally seek the widest possible set of

    professional opportunities. As Ph.D. programs usually require between three and four years, the

    large cadre of students beginning their research during the first phase of GVP completed their

    work at some point during phase II. Although many have been retained by GVP as postdoctoral

    scientists or continue their collaboration via home institutions in Ghana and Burkina Faso, many

    entered the international job market, and were effectively lost to the project just as their research

    skills had matured. Second, a number of senior scientists moved from research positions within

    GVP to tenure-track positions at major research institutions; or to international research centres

    including the International Water Management Institute (IWMI). Most have continued to

    contribute actively to the GVP, but are no longer “dedicated” to the project. Turnover has clearly

    and regrettably resulted in discontinuities in several research activities during phase II, but also

    serves to demonstrate, via the success of alumni and senior staff in securing increasingly

    responsible positions, the quality and impact of GVP research activities. Access to data – both

    physical and socio-economic – continues to be an issue that affects many areas of GVP research,

    although many of the obvious gaps have been address through phase I and early phase II data

  • 16

    collection efforts. Given that the GVP methodological approach was predicated on sparse or

    nonexistent data, many of the impacts of data scarcity have been fully factored into research

    designs.

    A complete review of project tools and technologies cannot be presented here. Some points

    appear relevant, however: First, decisions made early in the project concerning the selection of

    numerical simulation tools (“models”) have largely been vindicated in terms of model suitability

    and performance, with a few key exceptions. Models were not chosen ex ante on the basis of

    inter-operability, however, so that model linkage and integration tasks encountered for the first

    time during phase II have proven rather demanding. In this context, the component-based

    integration strategy, initiated in mid-phase II and described in greater detail under work package

    D1, appears as a promising approach, and indeed the only approach identified that is likely to

    enable the more ambitious task of full DSS integration underway during phase III. The physical

    infrastructure supporting GVP activities in Ghana and Burkina Faso has served us well, which is

    largely a reflection of the skill and experience of GVP project managers within the Basin. In

    Ghana, Dr. Boubacar Barry has served effectively since June of 2005, when he replaced Dr. Mark

    Andreini, who moved to IWMI’s liaison office in Washington, D.C. In Burkina Faso, Dr. Konrad

    Vielhauer has served since project inception. Their effectiveness has been greatly enhanced

    through the network of participating and collaborating institutions. IWMI, along with the

    Ghanaian Council for Scientific and Industrial Research (CSIR) through affiliated organizations

    have provided strong support for our program in Ghana, as have Foundation Dreyer and INERA

    in Burkina Faso. More recent extensions of the GVP research network are described in the phase

    III Proposal (October 2005). Finally, the political matrix in which the project operates within

    West Africa, specifically the Governments of Ghana and Burkina Faso and a range of

    international actors, continues to provide a favourable setting for project activities.

    1.2. Planning and progress of the project

    The overall objectives of the GLOWA Volta Project (GVP) remain consistent with the initial

    proposal (submitted in March 2000) and the updated Phase II proposal (submitted in October

    2002): (1) to provide an analysis of the physical and socio-economic determinants of the

    hydrological cycle within the Volt Basin, and (2) to develop a scientifically sound Decision

    Support System (DSS) for the assessment, sustainable use and development of the Basin’s water

    resources. The DSS is intended to provide a comprehensive monitoring and simulation

    framework, enabling decision makers to evaluate the impacts of climatic and land use trends

    overlaid on the consequences of deliberate policies, investments and other interventions on the

  • 17

    social, economic, and biological productivity of water resources. Integral to this effort is the

    development of scientific capacity and infrastructure within the Basin to ensure the self-

    sustainability of the DSS through the completion of formal GVP activities. The development of

    human capital via advanced education and training, co-operative research and stakeholder

    participation has been one of the GVP’s most successful activities to date, and is now identified

    as the third overall objective of the Project.

    The GLOWA Volta phase II was initiated in June 2003, with four strategic objectives identified:

    • Successful completion of phase I activities;

    • Expansion of significant project activities to Burkina Faso;

    • Technical integration of disciplinary models and knowledge generation frameworks;

    • Design, testing and preliminary application of a prototype DSS for the White Volta Basin.

    Broadly speaking, the first and second objectives have been completed successfully, and

    substantial progress has been made toward the third and fourth objectives, respectively. These

    accomplishments will be described in detail in the following text. In addition, a number of

    additional activities and accomplishments not explicitly described or anticipated in the phase II

    proposal were initiated. These activities include (i) a multi-disciplinary evaluation of shallow

    groundwater irrigation, an increasingly common practice in Northern Ghana and throughout

    Burkina Faso, conducted jointly with the International Water Management Institute (IWMI),

    Accra, Ghana; (ii) integrated analysis of small reservoirs, found throughout the upper regions of

    the Basin, in collaboration with the CGIAR Challenge Program for Water and Food (CPWF),

    Small Reservoirs Project; and (iii) sector-level analysis of the agricultural and power sectors of

    Ghana and Burkina Faso, respectively, utilizing the CIRAD MATA (Multi-Level Analysis Tool

    for Agriculture) bio-economic model and the West African Power Pool (WAPP), developed via

    USAID and currently under implementation throughout the ECOWAS region. Although these are

    largely intended as GVP phase III activities, and described in detail in the phase III proposal,

    groundwork was successfully laid during Phase II. In addition, these activities have served to link

    the GVP with the nascent Volta River Basin Authority, formally constituted in 2006.

    Although the GLOWA Volta Project is a large, multi-disciplinary and multi-institutional project,

    the presence and influence of a relatively small number of senior scientists nevertheless act to

    direct the GVP research agenda, as well as specific research activities. A number of personnel

    changes occurred during GVP phase II, some of which had implications for the schedule and/or

    orientation of research within the GVP. Among the most important are the following: (1) in late

    2003, Michael Schmidt of DLR – University of Wuerzburg was promoted to a more senior and

    supervisory position within DLR, thus creating a critical vacancy within the lead partner

  • 18

    responsible or retrieving, correcting and interpreting remote sensing imagery for the GVP. A

    suitable replacement for Dr. Schmidt was not identified until late 2004, when Dr. Sasa Fistric

    joined DLR-Wuerzburg. Dr. Fistric was in turn replaced by Dr. Tobias Landmann in late 2005.

    As a consequence of these discontinuities, the status of GVP RS activities is only now, in early

    2007, at the point where we anticipated we would be at the end of Phase II. This has in turn

    delayed completion of the cellular automata-based land use-land conversion (LULC) model, also

    envisioned for completion during Phase II. This will be complete in late 2007. (2) In late 2004,

    Thomas Berger moved from ZEF to University of Hohenheim as a visiting Professor, leaving the

    task of identifying and utilizing multi-agent system (MAS)-based models of community-based

    water resources management incomplete. Dr. Berger has continued active collaboration with the

    GVP through his role in the CGIAR CPWF Governance and Modelling Project, although this

    project is on an independent timeline. Dr. Berger’s position of GVP lead economist was not filled

    until Dr. Victor Afari-Sefa joined ZEF late in the phase II project cycle. Dr. Afari-Sefa is an

    agricultural economist whose expertise extends to the areas of partial equilibrium economics, as

    distinct from MAS, so the economic decision-making component of the project has been re-

    oriented accordingly. (3) Development of agent-based GV-LUDAS, which had proceeded rapidly

    during GVP phase I under Dr. Soojin Park, was interrupted following Dr. Park’s return to Korea

    in 2003, but resumed under Quang Bao Le, a student of Dr. Park’s, in 2005. Dr. Le, a Forester

    and Physical Geographer, has already developed a working prototype community-scale model of

    land and water resources use, and two Ph.D. students are currently implementing this model at

    locations in the Upper East Region (UER) of Ghana and the Dano region of Burkina Faso,

    respectively. Finally (4), Nick van de Giesen, the GVP Scientific Coordinator, accepted a tenure-

    track position at TU Delft in mid-2004. Dr. van de Giesen was replaced by Dr. Charles Rodgers,

    from the International Food Policy Research Institute (IFPRI), Washington, D.C., in July of

    2004. As there was overlap between Dr. van de Giesen and Dr. Rodgers, no hiatus in project

    scientific coordination occurred, and as Dr. van de Giesen continues to participate in GVP

    scientific research, no loss of continuity in project activities or oversight was experienced.

    1.3. Achievements in the establishment of infrastructure and capacity

    building

    1.3.1. Technical infrastructure and research network

    During the first phase of the GVP, research activities were largely confined to the major

    downstream riparian country, Ghana. During phase II, project activities have expanded

    significantly within Burkina Faso, primarily through collaborative research with Institut de l’

  • 19

    Environment et de Recherche Agricoles (INERA) and through the establishment of research

    networks linking GVP with BIOTA-West Africa and the Virtual Institute – Foundation Dreyer.

    The research network within Ghana has expanded as well. In early 2005, the White Volta Pilot

    Project was formally initiated in collaboration with the Ghanaian Water Resources Commission

    (WRC), joined by the International Food Policy Research Institution (IFPRI) via the Challenge

    Program on Food and Water (CPWF), Governance and Modelling Project (GMP). Shared project

    offices were established in Bolgatanga, in the UER, and several joint investigations initiated.

    These include studies of small reservoirs, shallow groundwater irrigation and riverine pump

    irrigation. All are of high interest to the WRC, and to the Ghanaian Ministry of Food and

    Agriculture (MoFA).

    GVP research infrastructure also expanded significantly though the integration of Project

    activities and research sites into regional research networks. Important among these are the

    formation of an integrative multi-scale monitoring concept in Burkina Faso linking GVP with

    BIOTA West Africa (BMBF) and Burkinabé counterpart INERA. The Biophysical Observation

    Network (BON) combines important features of biophysical ground measurement and remote

    sensing techniques in order to enable (a) monitoring of large scale vegetation, hydrologic and

    bio-geophysical dynamics and (b) evaluation of climate dynamics based on observations of

    biosphere – atmosphere interactions. 2004 also saw the inauguration of the Centre de Recherches

    et de Formation Scientifique (CRFS), an international research facility constructed by the Dreyer

    Foundation and shared by ZEF/GVP, BIOTA West Africa, Helmholtz Institute and INERA. This

    innovative research site in western Burkina Faso provides residence and research facilities for

    scholars conducting extended agricultural, environmental and hydrologic studies. The Dreyer

    Foundation field station at Dano (Burkina Faso) was used to support the AMMA (African

    Monsoon Multidisciplinary Analysis) 2006 field campaign. To support the transfer and ongoing

    development of computationally-intensive numerical modelling activities, GVP established a

    formal working relationship with the Kofi Annan Center of Excellence in Information and

    Communications Technology (KACE), a joint India – Ghana facility located in Accra providing

    regional access to high-performance computing and training in hardware and software

    development and utilization. KACE also provides broadband satellite links to the Internet. Our

    strategic objective is to transfer computationally intensive aspects of the Volta Basin DSS,

    including climate forecasting, to the KACE during GVP phase III.

  • 20

    In anticipation of the European Space Agency’s (ESA) pending launch of the Soil Moisture and

    Ocean Salinity (SMOS) mission in early 2008 (http://www.esa.int/esaLP/LPsmos.html), GVP

    submitted a proposal to participate, and we are now part of the SMOS Validation and Retrieval

    Team (SVRT), with Ph.D. student Jan Friesen serving as principal investigator (ESA-SMOS PI

    #3280). GVP has established a soil moisture monitoring network, consisting of soil moisture

    transects at three ongoing GVP research sites within Ghana and Burkina Faso: Boudtenga,

    Tamale, and Ejura. The field stations at Dano and Navrongo, respectively, are also equipped with

    automatic weather stations that record soil moisture levels at fine intervals (Friesen 2005). These

    instrumented sites will provide important in situ measurements required to calibrate and validate

    the SMOS mission, and will also serve to improve the calibration and validation of the GVP

    MM5 meso-scale climate model, operated by IMK-IFU and currently providing near-term

    forecasting for the basin. Remotely-sensed soil moisture fields will become an important

    component of the GVP climate monitoring system.

    To provide additional points of validation for numerical climate simulation (MM5) and

    hydrological (WaSiM-ETH) modelling ensembles, four sap flow measurements sites were

    installed in the Nature Reserve of Bontioli, Burkina Faso in collaboration with BIOTA West

    Africa (Tia Lazare, Ph.D. student from Burkina Faso). The sites were selected to provide

    coverage of a large number of the dominant tree species within the upper Basin, and will be

    valuable in cross-validating wide-area evapotranspiration estimates derived from numerical

    models, Large-Aperture Scintillometry (LAS) and eddy covariance measurements, and SEBAL

    and related energy balance algorithms.

  • 21

    Figure 1 Sap flow measurement equipment

    To strengthen the understanding of groundwater recharge and flow processes in the upper regions

    of the Basin, where groundwater is relied upon extensively to provide domestic water supplies of

    acceptable quality and quantity throughout the year, four stations within the Nakambe (White

    Volta) sub-basin within Burkina Faso were equipped with rain gauges, piezometers and divers

    (Jean-Pierre Sandwidi, Ph.D. student from Burkina Faso). These are monitored daily, and form

    the basis for estimates of annual groundwater recharge. In addition, 15 water quality sampling

    points from 6 localities within the basin are monitored on a daily basis to provide a continuous

    evaluation of groundwater quality and quantity. In early 2006, an additional study of groundwater

    recharge was initiated in the Upper East Region of Ghana (Emmanuel Obuobie, Ph.D. student

    from Ghana) which will also establish a regional network of climate and groundwater sensors,

    supported by numerical modelling of the spatio-temporal patterns of groundwater recharge using

    the USDS-ARS model SWAT (Soil & Water Assessment Tool). These studies build on the work

    of Dr. Nicola Martin, who completed her Ph.D. study of groundwater in the Atankwidi

    catchment, UER Ghana, in 2005.

    To facilitate the calibration and validation of coupled meteorological (MM5) and hydrological

    (WaSiM-ETH) models within the White Volta sub-basin, a dense observational network for

    precipitation and surface runoff was installed in spring 2004 in the Upper East Region in Ghana,

    involving scientists from ZEF-Bonn, IMK-IFU and the Ghanaian Hydrological Services Division

    (HSD). In October 2004, two Hydro Argos systems were installed at locations selected in

    consultation with HSD. The Hydro Argos units transmit water levels measured by in situ divers

    (pressure transducers) by dedicated satellite uplink to an internet platform, so that this data can be

    retrieved remotely and analyzed in near-real time from locations in Germany and/or Ghana.

    Sap Flow sites location

    Sapflowfixedsystem

    Boundaryof NRB

    LEGEND

    b

    ÊÚ

    ð

    EddyCovariancestation

    Micrometeorologystation

    bSapflowmobilesystem

    0 2 4Km

    o

    bb

    b

    bbbð

    ÊÚLoc3

    Loc4

    Loc6

    Loc5

    Loc7

    Loc8

    Source: Fielddata

    L. TIA©

    Sap Flow fixed station on groundSap Flow mobile system on tree

  • 22

    Figure 2 Location of sub-catchments and gauges in the White Volta catchment (resolution: 1km)

    1.3.2. Doctoral studies

    As in phase I, the project contributed extensively to local capacity building in the form of Ph.D.,

    Master and Diploma research. Most of the doctoral students trained in the context of GVP were

    affiliated with the Bonn International Graduate School for Development Research (BIGS-DR),

    located at ZEF – Bonn University. Others worked as student assistants and supported the GVP

    through both research and administrative and organizational services. In return, the research

    institute provided offices and other research infrastructure and also consistent supervision during

    the preparation, research and completion of individual studies. During their research, students had

    access to GVP infrastructure in Ghana und Burkina Faso and to local project researchers who

    provided support, scientific advice and supervision, information, social contacts, and technical

    instrumentation.

    All Ph.D. projects started in the first project phase were completed during phase II, while many

    Ph.D. projects initiated during phase II will be finalized before mid - 2007, during the first year

    of phase III. As documented in the Annual Report for 2005, the timelines for Ph.D. projects are

    not necessarily aligned with project phases, but are rather determined by the timeline of academic

    programmes, the cycles of dry and rainy seasons in the Volta basin (which govern research

  • 23

    logistics as well as researched phenomena), vegetation cycles, the occurrence of environmental

    hazards, as well as by institutional calendars, such as periods reserved for thesis defence and final

    examinations as set by the relevant faculty. In most cases, the three years allocated to the ZEF

    doctoral programme resulted in a tight academic schedule, and not all doctoral students are able

    to complete within the allotted period. Extensions for an additional 4 to 6 months became

    effectively routinized. In spite of extended timelines, the strategy of utilizing doctoral research

    projects as a primary engine of GVP research proved to be successful, and only one Ph.D. student

    supported by the GVP was unable to complete the Ph.D. degree, due largely to family issues. In

    phase II, the project produced twenty-two completed Ph.D. projects, seven MSc. theses, one

    M.Phil., and four Diploma theses. Another fourteen Ph.D. projects of phase II are ongoing and

    will be finished in early GVP phase III.

    Figure 3 Completed Ph.D., M.Sc. and M.Phil. theses in phase II

    0

    2

    4

    6

    8

    10

    12

    14

    16

    Number of Students

    Germany Burkina

    Faso

    Ghana Mali Eritrea Nigeria Netherlands Ethiopia USA

    Country of Origin

    GLOWA Volta Students

    Number of students

    1.3.4. Cooperation with local stakeholders

    As the project proceeds and the DSS becomes increasingly functional, the level of stakeholder

    interest and the local requirements for informed decision-making among water experts serve as

    guidelines for the development of the user platform, the selection of the use cases, as well as for

  • 24

    the identification of applicable knowledge from our scientific data sets, including policy

    recommendations. Several stakeholder workshops were organized in Ghana to elaborate on such

    specific local interest for decision support, as well as for the introduction of the GLOWA Volta

    project and preliminary research results to local water experts and stakeholders. Such workshops

    further enhanced the creation of a local stakeholder - GVP network, and the establishment of

    professional partnerships. For example, during the “The White Volta Basin Pilot Workshop”

    (14.07.2005 in Bolgatanga), agricultural policies in northern Ghana and public participation in

    decision making in the water sector and possible contributions of the GVP were debated with

    local stakeholders, such as representatives of MoFA, WRC, IFPRI and local farmers. During

    2005, GVP scientists also initiated informal collaboration with the Canadian International

    Development Agency (CIDA) for improved understanding of groundwater distribution and

    behaviour in the Northern and Upper East regions of Ghana, where CIDA is extensively involved

    in supporting community water supply projects (for more details on the workshops, see section

    2.4.3. on D3 sub-project).

  • 25

    2. Achievements and research results of the sub-projects

    2.1. The Atmosphere Cluster

    Primary objectives of the Atmosphere Cluster are to estimate the impacts of climatic and land

    use changes on the quantity and timing of precipitation; to quantify feedback mechanisms

    between land processes and climate and to establish a capacity for simulating a range of

    historical and hypothetical climatic conditions as an important component of the Volta Basin

    DSS toolkit. Significant accomplishments during phase II included the successful generation of

    regional climate scenarios corresponding to “present” and “future” (2030-2040) conditions,

    calibration, validation and utilization of distributed physical hydrologic models at three nested

    scales, coupling of mesoscale climate with physical hydrologic models, preliminary analysis of

    factors influencing the onset of the rainy season and initiation of web-based short-term forecasts

    for West Africa.

    2.1.1. Sub-project A1: Regional Climate Simulations

    G. Jung, H. Kunstmann

    Abstract

    The Volta Basin is a semi-arid to sub-humid region in West Africa, believed to be sensitive to broad

    changes in global and regional circulation patterns. The livelihood of much of the rural population

    depends critically on rain-fed agriculture and is therefore highly vulnerable to rainfall variability and

    climate change. Coupled regional climate-hydrology simulations are required to generate the

    numerical data used in assessing a wide range of local strategies for pre-empting, or adapting to

    a changing climate. To investigate the likely impacts of potential global climate change on regional

    climate and surface- and sub-surface hydrological processes in the region of the Volta Basin, coupled

    regional climate-hydrology simulations were performed. Following validation of the meso-scale

    meteorological model MM5, the model was used to generate two 10-year reference time slices: 1991-2000

    representing “present climate” and 2030-2039 representing “future climate’. These regional climate

    simulations were then coupled to the physically based, distributed hydrological model WaSiM-ETH,

    following the calibration and validation of the hydrological model within the study region. This coupled

    model ensemble provides the basis for a wide range of empirical water management studies,

    commencing in 2006 and continued in phase III.

  • 26

    Milestones achieved:

    Ø Validation of the meso-scale meteorological model MM5

    Ø Calibration and validation of the distributed-parameter physically based hydrological

    model WaSiM at three nested levels within the region of the Volta Basin

    Ø Analysis of regional climate simulation results for 10 years of present climate (1991-

    2000) and 10 years of future climate (2030-2039)

    Ø Performance and analysis of coupled regional climate - hydrology simulations

    Ø Preliminary statistical analysis of regional climate change (rainfall, temperature) for the

    recent historical period

    Ø Statistical analysis of change of surface and subsurface water balance components

    Work in progress:

    Ø Coupled simulations of climate and LUC changes

    Research results

    The meso-scale (regional) climate model was specified and evaluated for the rectangular region

    in West Africa containing the Volta Basin. Atmospheric boundary conditions were provided by

    the General Circulation Model (GCM) ECHAM4, which was dynamically downscaled in two

    stages, from the original 2.5 resolution of ECHAM4 (“Domain 1”) to 27 km x 27 km (“Domain

    2”), and finally to 9 km x 9 km pixels within the Volta Basin model domain (“Domain 3”). A

    comparison of the ECHAM4 output, as well as the Regional Climate Model (MM5) output

    corresponding to present-day climatic conditions relative to historical observations indicated a

    wet bias over the Sahel; although simulated temperatures in ECHAM4 corresponded well with

    historical observation (1961-1990). In the regional (MM5) climate simulations, the displacement

    of the Inter-Tropical Discontinuity (ITD) to the North at the beginning of the rainy season, as

    well as its displacement Southward at the end of the rainy season occur early relative to historical

    observation. Simulated rainfall showed a negative deviation relative to observation along the

    coast, but appeared sufficiently accurate within the Volta Basin proper.

    The physically based, distributed parameter hydrological model WaSiM-ETH was specified and

    calibrated for the topography, land surface cover and climate of the Volta Basin. The calibration

    period was 1962-68, although calibration on the basis of daily gauging records was possible only

    for the hydrological year 1967/68 due to limitations in stream gauging records. Modifications

    were required to resolve evapotranspiration when running WaSiM at a daily time step (figure 4)

    and to properly account for the impacts of the Sourou depression (figure 5), a topographically

    depressed region in which extensive water storage occurs during many rainy seasons. The

    calibration run, as well as a coupled MM5-WaSiM run for 1968 indicated good performance for

    different sub-catchments (figure 6).

  • 27

    Figure 4 Calculation of evaporation in 24h time step, 3h time step simulation and with improved

    calculation with 24h time step, including an empirical factor

    Source: Jung (2006)

  • 28

    Figure 5 Model setup: Digital elevation model, sub-catchment boundaries, river network and location of

    the Sourou depression

    Source: Jung (2006)

    Figure 6 Calibration run, coupled simulation and observed discharge values for the gauges Bamboi and

    Saboba, 1968

    Source: Jung (2006)

    The results of the coupled regional climate-hydrology simulations suggest an annual mean

    temperature increase of 1.2-1.3°C in West Africa and the Volta Basin over the period 1991-2000

  • 29

    to 2030-2039. This temperature change is judged to be significant relative to inter-annual

    variability. Simulated mean annual precipitation increases over both the Sahel and the coastal

    regions of West Africa above the Gulf of Guinea. Spatial averages are 840 mm for present

    conditions and 894 for future conditions, an increase of roughly 6% over the 40-year period.

    However, this increase, at basin-level at least, is relatively small given inter-annual variability

    within the region. Only in the Sahel does simulated change in mean annual precipitation exceed

    simulated inter-annual variability. The magnitude of increase is highly heterogeneous spatially,

    ranging from -20 % to +50 %. A dipole pattern of rainfall variability in the Sahel and the Guinea

    Coast region was detected for June and July. An overall increase in precipitation was found for

    September, and a strong decrease for April. This decrease is of particular significance, since April

    is historically the month in which seasonal rains begin over much of the basin, and the simulated

    decrease corresponds closely to anecdotal reports of delays in the onset of rains, and reductions in

    precipitation during the early parts of the rainy season. A partial explanation for simulated

    rainfall variability was found in the dynamics of the Tropical Easterly Jet (TEJ), the African

    Easterly Jet (AEJ) and in the position of the ITD (figure 7).

    Figure 7 Mean latitudal displacement (2030-2039 vs. 1991-2000) of the Inter Tropical Discontinuity

    (ITD): southward (northward) shift in April (September) connected to a decrease (increase) in

    precipitation.

    Source: Jung (2006)

    Simulations for the Volta Basin indicate that the decreases in April (at the beginning of the rainy

    season) are associated both with lower rainfall accumulations and with delays in the onset of the

    rainy season, averaging around nine days within the Sahel region of the basin. In addition,

    simulated inter-annual variability in the Volta Basin increases in the early stage of the rainy

  • 30

    season. The simulations did not show strong changes in aridity, as evaluated using the de

    Martonne aridity index, which depends primarily on annual values of temperature and

    precipitation. The behaviour of this annual index (if subsequently validated) is mildly

    encouraging; since it suggests no major disruption in climatic suitability for agricultural

    activities. However, simulated increases in the inter-annual rainfall variability in the early stage

    of the rainy season, as well as the delay in the onset of the rainy season increases the vulnerability

    of local farming activities, even if annual rainfall amounts do not change significantly.

    Figure 8 Climate change signal in precipitation (prec), direct runoff (qdir), interflow (qifl), total runoff

    (qtot) and evapotranspiration (etr), 2030-2039 vs. 1991-2000, left: percentage change, right: absolute

    change

    Using the MM5-WaSiM-ETH model ensemble, simulated changes in the hydrologic budget can

    be disaggregated into changes in precipitation (prec), evapotranspiration (etr), direct runoff (qdir)

    and inter-flow (qifl). Over the entire basin, simulated increases in precipitation are to some

    degree offset by increases in actual evapotranspiration, which largely reflect increases in potential

    ET via temperature effects. Overall (direct + interflow) runoff is still projected to increase,

    although by a smaller amount in percentage terms than either rainfall or actual

    evaoptranspiration. No significant changes in discharge follow the precipitation decrease at the

    onset of the rainy season. (figure 8).

  • 31

    Figure 9 Monthly signal-to-noise ratio of precipitation (prec), total runoff (qtot) and evapotranspiration

    (etr)

    Source: Jung (2006)

    To summarize the results of integrated climate-hydrology simulations at full basin scale over the

    period 1991-2000 to 2030-2039, temperatures will increase, hence potential evapotranspiration

    will increase. Precipitation will increase modestly (subject to a high level of spatial variability)

    and thus actual evapotranspiration will increase due to simultaneously increasing “supply” and

    “demand” trends. Total runoff will increase only modestly, composed of increases in surface

    runoff overlaid on small decreases in inter-flow. However, with respect to signal-to-noise ratios

    (trends relative to inter-annual variability), only the temperature trend clearly exceeds interannual

    “noise”. The climate change signals of most of the other reference climatic and hydrological

    variables lie essentially within the range of inter-annual variability (figure 9). This largely reflects

    the modest increases in simulated precipitation. The future impacts on water resources; and on

    specific activities including rain-fed agriculture, are therefore somewhat ambiguous, and will

    likely depend on the actions (or inactions) taken in response to a changing environment. Higher

    rainfall is in general favourable for rainfed agriculture, but increasing interannual variability and

    delays in the onset of seasonal rains are not. Together with increasing temperatures (the single

    unambiguous trend) these trends suggest an increased justification for, and a likely increase in the

    coverage of supplemental irrigation as a likely response. Moreover, simultaneous to projected

    changes in climatic and hydrologic conditions, demographic and economic trends will invariably

    work to increase the demand for water resources while doing little to influence overall supply.

    These factors are likely to exacerbate water stress and reduce water availability within the basin,

    manifesting over the same time frame as the climatic processes, and the relative magnitudes of

    these impacts cannot be assumed.

  • 32

    2.1.2. Sub-project A2: Hydro-Meteorological Monitoring System

    J. Friesen, J. Hendrickx; H. Kunstmann, A. Moene, D. Schüttemeyer, S. Wagner

    Abstract

    The primary challenge for the development and testing of the Hydro-Meteorological Monitoring

    System (HMMS) is to develop systems and protocols by which real- or near-real-time data can be

    assimilated into the coupled model framework, so that e.g., predictions of river discharge and

    reservoir storage can be updated with sufficient timeliness to improve water management

    decisions; or forecasts of the onset of seasonal rains can be improved using updated climatic

    indicators. Substantial progress was made within sub-project A2 toward an operational HMMS,

    although much challenging work lies ahead in phase III. A fundamental problem is data itself.

    Although many types of climatological and hydrological data may soon be available in near-real

    time via a range of satellite remote sensing products, at present most RS-based data must be

    calibrated and/or validated before it can be utilized successfully. One possibility for dealing with

    this problem is to provide the required input data with a meteorological model, and using

    simulated climatic data to drive a coupled hydrological model. We investigated the extent to

    which meteorological models are able to provide the required meteorological fields with

    sufficient accuracy for predictive hydrological modelling. In this study, the meso-scale

    meteorological model MM5 and the distributed parameter water balance simulation model

    WaSiM were used in the setting of the White Volta (Nakambe) Catchment. Prior to coupling, both

    the meteorological and hydrological models were calibrated and validated separately. In

    addition to missing meteorological data, gridded information on land surface properties (albedo,

    LAI, etc.) is also difficult to obtain, although it is an essential input to distributed hydrological

    models. Satellite remote sensing can provide global, spatially detailed information on land

    surface properties, particularly valuable in areas such as the Volta Basin where such information

    is otherwise difficult to obtain.

    Milestones achieved:

    Ø Operational web-based climatic forecasts for the Volta Basin region

    Ø Installation and operationalization of real-time river discharge measurements using

    HydroArgos

    Ø Successful testing of the Makkink-Vegetative Fraction method of wide-area, remotely

    sensed surface energy flux estimation

    Ø Successful testing of the Surface Energy Balance Algorithm (SEBAL) method of wide-

    area, remotely sensed surface energy flux estimation using LANDSAT and MODIS

  • 33

    imagery

    Work in progress:

    Ø Energy flux estimates using Large Aperture Scintillometer network; comparison with RS-

    based methods and MM5 simulation

  • 34

    Research results

    To address the limited availability of meteorological and hydrological data, an expanded

    observational network for precipitation and surface runoff was installed in the Upper East Region

    in Ghana in 2004 (see section 1.3.1.) which provides supplemental data for the calibration and

    validation of the coupled meteorological and hydrological models for the White Volta tributary.

    The measurement campaign was conducted in close collaboration with Ghanaian Hydrological

    Services Division (HSD). The distributed parameter hydrological model WaSiM-ETH was

    specified for the White Volta catchment. Results of the hydrological simulations driven by

    measured values of the required meteorological inputs indicate that for the calibration period

    1968 (see figure 10) and validation period 1961-1967 (see figure 11) the overall discharge

    hydrograph could be simulated quite satisfactorily, particularly given the large catchment size

    and available data basis.

    Figure 10 Measured (blue) and simulated (green) discharge [m³/s] and precipitation (black) in [mm] for

    the gauges Nakong and Nawuni for the calibration period 1968

    Source: Wagner et al. (2006)

    Figure 11 Simulated (green) vs. measured (blue) discharge [m³/s] at Pwalugu and Nawuni for the

    validation period 1961-67.

    Source: Wagner et al. (2006)

  • 35

    The year 2004 was chosen as the time period for the coupled meteorological- hydrological

    simulations. Figure 13 shows one output of the MM5 meteorological simulations, specifically the

    spatial distribution of annual precipitation [mm] for Domain 3 (9 x 9 km²) containing the Volta

    Basin. The precipitation distribution shows a strong North-South gradient with values of less than

    300 mm in the North and over 1800 mm in the South. The scatter plot in figure 12 provides a

    comparison between the available observed and simulated monthly precipitation sums. The

    results of the coupled meteorological- hydrological simulations (see figure 13) demonstrate that

    the model ensemble is able to simulate the behaviour of the discharge hydrographs satisfactorily.

    Figure 12 Simulated annual precipitation sum for 2004 (Domain 3) for the Volta Basin

    Source: Wagner et al. (2006)

    Figure 13 Comparison of simulated and measured monthly precipitation sums in 2004 for gauging stations

    (Kaburi, Kpasenkpe, Pwalugu, Babile, Zuarungu) in Ghana

    Source: Wagner et al. (2006)

    2 0 0 4 m i t F D D A

    0

    1 0 0

    2 0 0

    3 0 0

    0 1 0 0 2 0 0 3 0 0

    S i m u l a t e d [ m m ]

    Ob

    se

    rv

    ed

    [

    mm

    ]

    K a b

    K p a

    P w a

    B a b

    Z u a

  • 36

    In physical hydrological simulations, the land surface parameters albedo and leaf area index

    (LAI) are influential. To evaluate the utility of satellite-derived gridded land surface data in

    hydrological modelling, composites of albedo (16 days) and LAI (8 days) grids for 2002 were

    analyzed, aggregated and imported into the hydrological model. The preliminary results indicate

    that the hydrological simulations are relatively insensitive to quarterly changes in the albedo

    grids. However, the assimilation of seasonally-dependant LAI, as shown in figure 15, has a

    pronounced effect on all water balance variables, such as the spatial distribution of annual

    evapotranspiration (figure 16). Particularly within the southern regions of the model domain,

    higher annual evapotranspiration sums were calculated using MODIS-derived LAI-grids.

    Figure 14 Coupled simulated (green) and measured (blue) discharge [m³/s] and precipitation (black) in

    [mm] for three gauging stations along the White Volta in Ghana (Kaburi, Pwalugu and Nawuni)

    Source: Wagner et al. (2006)

  • 37

    Figure 15 Leaf Area Index (LAI) for the White Volta catchment: mean MODIS LAI-grids for 2002

    January – March April – June July – September October - December

    Source: A2 sub-project

    Figure 16 Annual evapotranspiration map simulated with a) standard literature and b) MODIS derived

    LAI- grids; c) difference of annual evapotranspiration: a) - b)

    Source: A2 sub-project

    The objective of developing the coupled climate-hydrology model complex is to develop the

    capacity to simulate the terrestrial water balance in the absence of directly measured

    meteorological data. As it is likely that the situation with regard to measured data is unlikely to

    improve significantly in the near future, the model ensemble will provide and important tool for

    contemporary estimation of the spatial and temporal changes of water balance variables, which

    will in turn provide important information supporting water resources management in the Volta

    Basin. While the model complex described here is currently operational, the various implications

  • 38

    of using satellite-derived land surface parameters (here: albedo, LAI) on coupled hydrological

    simulations continues to be an area of active investigation.

    Research results: Remote Sensing of Latent and Sensible Heat Fluxes

    In order to provide near-real-time forecasts of climate and hydrology for the basin, information

    concerning the accuracy of the wide-area surface energy balance is also essential. As with

    climatic and hydrological variables, direct measurement networks are sparse in West Africa, and

    remote sensing (RS) methods are required for broad scale validation of water balance variables.

    During phase II, we evaluated and applied a range of methods to estimate latent and sensible heat

    fluxes over the Volta Basin. These included (1) Large Aperture Scintillometer (LAS), (2)

    Makkink-Vegetative Fraction method and (3) the surface energy balance algorithm (SEBAL).

    The meso-scale climatic simulations generated by MM5 provide an additional point of reference.

    Research results: Scintillation

    Scintillation refers generally to perceived changes in the brightness of an object when viewed

    through an atmosphere. Common examples include the twinkling of stars, or “heat waves”

    arising from asphalt or other surfaces heated by the sun. The phenomenon occurs due to changes

    in the refractive index of the air caused by fluctuations in heat, humidity and pressure. Large

    Aperture Scintillomenters (LAS) are instruments that measure the turbulent intensity of the

    refractive index fluctuations across an air mass on the basis of fluctuations in the intensity of a

    light beam passing through that air mass. Under suitable conditions, sensible and latent heat

    fluxes can be derived from these fluctuations if net surface radiation, air temperature and vapour

    pressure are known. In practice, the light beam is generated by a light source placed from 0.5 to 5

    km from the light receiver. The transect, or path of the scintillometer beam, is chosen to be

    representative of the zone (pixel, region) for which the heat flux estimate is desired, with respect

    to topography, vegetative cover (LAI), surface roughness, albedo and related parameters. During

    phase II, LAS pairs were installed and operated at the GVP field sites lying along a south-north

    transect in Ghana, at Ejura, Tamale and Navrongo, respectively; each installation in proximity to

    the micro-meteorological stations that were intended to provide the required surface radiation and

    climate data.

    Research conducted during phase I indicated that LAS were capable of providing ground-truth

    validation of RS-based estimates of latent and sensible heat fluxes. In these experiments, satellite-

    derived H was computed from NOAA-AVHRR images for December 2001 under favourable

    cloud conditions, and compared to LAS and MM5 simulation results. An algorithm for

    estimating actual evaporation from Meteosat-derived global radiation and temperature and

    MODIS-derived vegetation fraction was also applied and validated for a complete drying-up

  • 39

    period. The LAS-derived evapotranspiration (ET) and sensible heat fluxes performed well with

    respect to energy-balance closure, and in comparison with eddy-covariance data, suggesting that

    they can be used with confidence for validation of RS-derived fluxes. Satellite derived H showed

    good agreement with LAS data, with RMSE of 39 W/m² for the Tamale site. Satellite derived H

    showed substantial disagreement relative to MM5 due to overestimation of latent heat flux in

    MM5, with RMSE of 167 W/m² for the Tamale site. Uncertainty analysis of satellite derived H

    indicated relative uncertainties of 21 % for the Tamale site and 32 % for the Ejura site.

    Uncertainty in LAS data is much lower, at 8 % for Tamale and 7 % for Ejura. The Meteosat-

    based procedure provides hourly estimates of actual ET under both cloudless and cloudy

    conditions. The agreement with ground-based data is generally good, but in wet periods (higher

    fluxes) the RS-based algorithm tends to underestimate real ET. Long term bias of this method is

    less than 0.5 mm/day. A significant advantage to this method is that it can be used to monitor ET

    at seasonal time scale, irrespective of cloud conditions.

    Based on these favourable preliminary results, the intention was to use the South-North transect

    of LAS installations during phase II to provide near-real-time validation for MM5-generated

    energy flux estimates and ultimately, to update and correct these fields as a component of the

    HMMS. This objective has not realized to date, due primarily to poor performance of the data-

    logging modules and procedures associated with both LAS units and the accompanying micro-

    meteo stations. A number of problems were observed and documented, including (i) data format

    errors; (ii) data over-write errors; (iii) date and time errors; (iv) implausible and/or out-of-range

    values for key climatic variables; and (v) power supply failures resulting in data loss. The

    presence of multiple failure modes creates a serious problem, since the effective use of LAS data

    requires the simultaneous availability of the corresponding radiation and climatic data. Thus, if

    any of the required data series are missing, corrupt or suspect on a given day, the LAS data itself

    may be unusable. In addition, in order for the LAS-derived flux estimates to be useful in

    validating RS-based flux estimates, atmospheric conditions (particularly cloud cover) must also

    be favourable for RS interpretation. An assessment of the LAS and supporting data collected at

    the three Ghanaian sites during phase II (Moene, et al., 2006) concluded that for each site, there

    were only a limited number of dates (weeks in total) that met the multiple criteria required for

    successful interpretation and application of LAS results. As a consequence, the objective of

    utilizing the LAS transects for calibration and validation of MM5 flux fields was not completed

    during phase II. On balance, an extensive body of information was obtained that should permit

    the improved operation of the LAS network through phase III of GVP. In addition, a number of

    dates were identified for which all component data required for successful energy flux estimation

    and RS validation are present, inclusive of RS imagery. This analysis is being conducted

    currently by Dr. Jan Hendrickx.

  • 40

    Research results: Optimal estimation of evapotranspiration at the basin scale using

    Makkink-Vegetation Fraction

    Wageningen MAQ’s contribution to the development of the Hydro-meteorological Monitoring

    System (HMMS) is the development of algorithms to estimate actual evapotranspiration at

    regional scale using Meteosat imagery. The baseline algorithm utilizes a combination of the

    Makkink formula for reference evapotranspiration (ETref) and the vegetation fraction (VF) in

    order to scale down ETref to actual ET as: ET = VF ETref. The underlying assumption is that

    vegetation cover adjusts to the extent of available water; and that this existing vegetation then

    transpires at its maximum rate.

    Figure 17 Different components of actual evapo-transpiration, example for November 15, 2002. Incoming

    solar radiation (top left), vegetation fraction (top right), reference temperature (bottom left) and resulting

    actual evapotranspiration (incl. evaporation from interception (bottom right).

    A first test of this algorithm was performed in 2005 for two sites in Ghana (Schüttemeyer et al.,

    2006). Although results for the baseline method are promising, a number of limitations have also

    been identified. Errors can occur on a limited number of days due to ignoring bare-soil

    evaporation and/or evaporation from canopy interception; and consistent errors can occur on all

    days due to the assumption that the crop factor equals one. Moreover, the Makkink method does

    not incorporate the effects of wind and atmospheric humidity on evapotranspiration, and the

  • 41

    baseline method is not able to account for the reduction in evapotranspiration due to water stress

    (in the short term, this stress has not yet resulted in a decrease of vegetation fraction).

    Based on these findings, the baseline algorithm has been extended to incorporate the effects of

    interception and bare soil evaporation, as well as the variability of the crop factor. The inclusion

    of rainfall information from climate models has also been considered, but a validation study of

    ECMWF re-analysis rainfall revealed that the timing, amount and location of the modelled

    rainfall were problematic. The same appears to hold for the MM5 simulated rainfall. Therefore

    near real-time rainfall information obtained from remote sensing data (RFE 2.0, a product of

    NOAA) has been included to serve as input for the parameterization of evaporation from

    interception and bare soil.

    The second modification involved the inclusion of different crop factors for high and low

    vegetation, as well as open water, based on the USGS land cover database. The enhanced

    baseline algorithm can now produce maps of daily evapotranspiration (based on hourly global

    radiation and temperature data) for the entire Volta Basin. The resulting evapotranspiration map

    for one day in November 2002 is shown in Figure 18. Incoming solar radiation and near-surface

    temperature are needed to determine ETref, and the vegetation fraction is needed to scale ETref.

    The resulting ET map reveals the signature of the North-South gradient of vegetation fraction,

    and the (small, on this particular day) variation in incoming solar radiation. In the South-West, a

    spot with high ET can also be identified which results from evaporation of the previous day’s

    intercepted rainfall. This spot of enhanced ET is also visible via enhanced cloudiness at that

    location in the global radiation image.

    In order to study the feasibility of including atmospheric information from a climate model , we

    performed extensive experiments with a single-column boundary layer model. This demonstrated

    that in principle, useful information is available in the boundary layer air temperature, which can

    be combined with a remotely sensed surface temperature to quantify water stress. However, this

    simulated air temperature proved to be very sensitive to the amount of soil moisture. Given the

    difficulties of numerical weather models in predicting the correct amount of rainfall at a specific

    location (particularly in the West-African climate), it is likely that the boundary layer

    temperature, as well as its diurnal cycle, will be incorrect in the model as well. Consequently, the

    exploration of including model data in the remote sensing algorithm was not pursued further.

    The algorithm described above has been developed using products for incoming solar radiation

    and near-surface temperature provided by DLR-Stuttgart, in combination with VF derived from

    MODIS vegetation products, and RFE 2.0 from NOAA. However, data from other sources can be

  • 42

    utilized as well. Specifically, incoming solar radiation and vegetation fraction products freely

    available from the Land Surface Analysis Satellite Application Facility (LSA-SAF) could be

    utilized. The near-surface temperature data used in Makkink ET (where errors of a few degrees

    do not matter) could be replaced by data from an atmospheric model (MM5) or freely available

    NCEP forecasts.

    Research results: SEBAL: Mapping Energy Balance Fluxes and Root Zone Soil Moisture in

    the White Volta Basin using Optical Imagery

    The overall goal of this activity is to demonstrate that important background information on the

    components of the energy balance and soil moisture can be obtained from optical imagery even in

    data-scarce areas such as the Volta, without relying on ground measurements. We evaluated the

    method for the heterogeneous arid lands of the White Volta Basin in Ghana, which represents a

    challenging environment. The specific objectives were (i) to test SEBALNM with LANDSAT

    imagery in the White Volta Basin of Ghana without using any ground measurements and (ii) to

    compare evapotranspiration maps generated from LANDSAT images with spatial resolution 60

    m to those generated from MODIS images with spatial resolution 1,000 m.

    Model Approach: SEBAL is a remote sensing flux algorit