Experimental drought early warning system in the Inner Delta

2013 International SWAT Conference, Toulouse, France Samuel Fournet1,2, Stefan Liersch1, Valentin Aich1, Léo Zwarts4, Bakary Koné3, Fred F. Hattermann1

1 Potsdam Institute for Climate Impact Research 2 UMR G-eau, Montpellier Supagro 3 Wetlands International 4 Altenburg & Wymenga

Wednesday 17th of July 2013 Impacts of climate change and upstream river management on the flood regime in the Inner Outline

1. The : case study characteristics

2. SWIM setup, development and calibration

3. Climate change and upstream water management scenario

4. Hydrological change and trends

5. Integration of the results in an operationnal drought early warning system in the Inner Niger Delta

2 Case study introduction Inner Niger Delta 3,27 M inhabitants

Large wetland inundation plain (40.000 km²) in the Sahelian climate zone

Drastic seasonal and inter-annual variation in discharge (30 to 50 hm3/y), flood extent (5 to 25.000 km²)

Flood peak delay ~2 months 30 to 50% water losses

Zone crucial for fishing, livestock, agriculture in free submersion and the biodiversity

High vulnerability from upstream management

Source: http://earthobservatory.nasa.gov/ Zwarts et al., 2005, Niger the lifeline, Wetlands International SWIM Soil and Water Integrated Model Development of Inundation module

Pre-processing in the delta floodplain: upstream of each sub-basin´s outlets, inundated area and the water volume accumulated and trapped in ponds are identified into sequential layers

Processes Parameters

1. Flooding > Flooding: flow-threshold 2. Routing, backwater 3. Evaporation (water surface) 4. Percolation 5. Release > Flood release (linear)

1 2 5

Source: Liersch et al., 2011, SWAT conference SWIM setup (1) Topography, Land-use, Soil, Sub-basins

Digital Elevation Model Shuttle Radar Topographical Mission SRTM Version 4, 90m resolution

Land-use classification

Global Land Cover 2000 GLC

Soil classification and parameterization Hydrotope FAO Digital Soil Map of the World Harmonized World Soil Database Hydrological Response

Sub-basin delineation Unit Number of sub-basins: 1923 Sub-basin average area: 1150km²

5 SWIM setup (2) Climate inputs

Watch Forcing Data (WFD) • ECMWF reanalysis ERA40

• from 1960-2001 at daily time step

• 0.5° resolution

• Bias corrected with Global Precipitation Climatology Centre (GPCC v4) and Climate Research Unit (CRU TS2.1)

Source; Weedon et al., 2010 , Watch tech report6 22 Aich and Fournet 2013 in Dewfora D4.6, PIK SWIM calibration Discharge Global Runoff Data Centre (GRDC) ID Monitored Gauge Calibration period NSE 1 Koulikoro 1964-1974 0.93 2 Douna 1964-1974 0.88 3 Ibi 1975-1995 0.87 6 4 1964-1974 0.86 10 1972-1982 0.85 7 8 5 2 11 6 Dire 1964-1974 0.83 1 7 Kirango Aval 1975-1981 0.82 9 13 8 Kandadji 1976-1986 0.82 4 14 9 Selingue 1965-1975 0.8 3 10 Ansongo 1968-1979 0.76 5 11 1975-1985 0.76 12 Tossaye 1968-1979 0.75 13 Malanville 1976-1986 0.54 14 Yidere Bode 1985-1995 0.18

Calibration Koulikoro gauge Source; Aich and Fournet, 2013 in Dewfora D4.6,7 PIK

Climate change projections Air temperature trends in the Upper Niger Basin

4 Earth System Models (ESMs) Downscaled and bias corrected by Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) •GFDL-ESM2M (GFDL) • IPSL-5 CM5A-LR (IPSL) •HadGEM2-ES (Had) • NorESM1-M (Nor) Use of 2 Representative Concentration Pathways underlying assumptions about radiative forcing •2.6 - “moderate” • 8.5 - “extreme”

Source; Liersch et al., 2013, AFROMAISON internal report,8 PIK Climate change projections Precipitation trends in the Upper Niger Basin

4 Earth System Models (ESMs) Downscaled and bias corrected by Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) •GFDL-ESM2M (GFDL) • IPSL-5 CM5A-LR (IPSL) •HadGEM2-ES (Had) • NorESM1-M (Nor) Use of 2 Representative Concentration Pathways underlying assumptions about radiative forcing •2.6 - “moderate” • 8.5 - “extreme”

Source; Liersch et al., 2013, AFROMAISON internal report,9 PIK Upstream river management Reservoirs and Irrigation schemes

1. Current and future irrigation water uptake with restricted minimal flows were setup in line with the development plan of the and the future dams in line with engineering technical report.

2. The scenario matrix was defined with local stakeholder representatives from the IND region

Source; Liersch et al., 2013, AFROMAISON internal report,10 PIK Results: climate change projections Impact on discharge at the combined IND´s inlet

11 Results: river management scenario Impact on annual maximum inundated area

3 dams Irrig. Irrig. 3 dams + Irrig. Ef.high Ef. Med. Irrig. Ef. High Ef. Low

Source; Liersch et al., 2013, AFROMAISON internal report,12 PIK Results: Scenario 3 dams + irrigation (: 250.000) Impact on discharge at the IND´s oultet

Source; Liersch et al., 2013, AFROMAISON internal report,13 PIK Conclusion Summary and planned research

• Climate change projections: increase of interrannual variability but trend agreement on flow in/de-crease remain unclear 1 >> Use RCMs projection from CORDEX project to enlarge the spectrum and the state of art for regional climate change impact • Upstream water management: results shows clear gradual impact on the flood propagation and extent 2 >> Test the impacts with other managerial options for dams 3 >> Vulnerability assessment of flood-dependent water uses

14 Conclusion Dissemination with OPIDIN Drought early warning system

Tool to predict flood peak and retreat water level and timing Range of early warning signals in use for the based on statistical regression function from historical water annual flood peak in level time series Annual flood peak Annual flood peak Mopti Akka Mopti Akka water level water timing Classes Range Freq. Range Freq. Classes Range Freq. Range Freq. Really low (80´s to 00´s) 440-550 cm. 10 330-410 cm. 10 Really early (70´s to 00´s) before 9 Oct. 11 in Oct. 10 Low (70´s to 00´s) 551-590 cm. 9 411-450 cm. 11 Early (70´s to 00´s) 9 to 16 Oct. 10 1 to 9 Nov. 10 Normal (70´s to 00´s) 591-640 cm. 11 451-500 cm. 9 Normal (70´s to 00´s) 17 to 24 Oct. 11 10 to 19 Nov. 12 High (80´s) 641-680 cm. 10 501-550 cm. 10 Late (80´s) 24 to 31 Oct. 8 20 Nov. to 30 Dec. 8 Really high (50´s to 60 ´s) 681-730 cm. 12 551-625 cm. 12 Really late (50´s to 60 ´s) in Nov. 12 in Dec. 12

OPIDIN stakeholder platform: dissemination via key persons, Workshop with sheperds to interprate and radio, bulletin disseminate the results of OPIDIN prediction

Source: Fournet , 2013 in Dewfora D4.8,15 PIK Koné Bakary, Wetlands International Thank you for your attention !

16 Questions ?

17 Spare slides

18 Case study introduction

3rd longest river in Africa watercourse 4200km 9th biggest fluvial system area 2.1M.km2 ~ 25% located in 9 countries , , , , , , Mali, Niger and Major cities Tembakounda, , , Niamey, Lokoja, 4 climate zones • Humid tropical zone • Tropical zone with dry seasons • Sahelian zone • Desert zone

UNEP. 2010 , Africa Water Atlas Zwarts et al., 2005, Niger the lifeline, Wetlands International Case study introduction Upstream river basin management The Upper Niger The zone of the Offices The Bani catchment 2,43 M. inhabitants 1.44 M. inhabitants 0.53 M inhabitants

• Covers the Guinean part of • Intensive irrigated rice • Reservoir of Talo and the basin and stretchs to production with Office du Djenné (planned Selingué dam included. Niger (Markala dam), Office extension) de Ségou and Office de • Crucial for the generation of Baguinéda with a high • High potential of rural water ressources with the potential to extend development of more than mountains agricultural area 100.000 ha (agriculture, fishing and livestock) • Regulation and storage • Bamako and the infrastructure with Selingué hydropower dam of Sotuba • Projects of minor dams in and the future Fomi dams Baoulé, Gbado and • High potential for navigation Bagoué • 5 RAMSAR sites

Source: NBA, PADD, 2010 Scenario matrix Reservoirs and Irrigation schemes

1. Current and Irrigation Efficiency Provision future Irrigation water Irrigation Scheme in ha Rice Rice CS Gardening Sugar Cane uptake and efficiency m³/ha/y in l/ha/s Sélingué 1600 31000 1.5 were setup in line with the Baguinéda 3000 400 71500 2.2 development plan of the Markala (ON) 77000 7700 15400 5000 30000 [SC:71200] 2.7 [SC:3.4] Niger Basin Authority and Sélingué planned 3200 31000 1.5 the future dams in line with Djenné planned 68000 13276 2.4 engineering technical Talo planned 20000 13276 2.4 Fomi planned 3000 10000 11500 1 report. Markala ON extension 1) 220000 22000 44000 30000 13500 [SC:71200] 1.2 [SC:3.4] 2. Water uptake Markala ON extension 2) 220000 22000 44000 30000 20000 [SC:71200] 1.8 [SC:3.4] was restricted to minimal Markala ON extension 3) 600000 60000 120000 30000 24000 [SC:71200] 2.2 [SC:3.4] flows (40m3/s at Markala 3. The scenario matrix was defined with local stakeholder and 10m3/s at Fomi, representatives from the IND region Sélingué, and Djenné)

21 SWIM Soil and Water Integrated Model

Process based eco-hydrological model, simulates runoff generation, nutrient and carbon cycling, plant growth and crop yield, river discharge and erosion as interrelated processes with a daily time step on the river basin scale

New Features • Reservoir-model to simulate effects of reservoir management, including Hydropower production • Conditionnal irrigation uptake in the river routing • Inundation-model to simulate effects of wetlands (flood propagation, evapotranspiration and discharge from wetland area)

Source: Krysanova et al., 2000, SWIM manual, PIK report 22n° 69 SWIM

Soil and Water Integrated Model transpiration

net radiation land use precipitation air temperature evaporation soil texture relative humidity management slope wind speed retention coefficient

surface roughness

LAI surface drainage field capacity

passage time t subsurface drainage drainable water per layer from the saturated zone hydraulic soil water conductivity content drainage porosity percolation slope length saturated rise capillary

conductivity

Groundwater (shallow aquifer) groundwater flow

23 OPIDIN (Flood prediction tool for the Inner Niger Delta) Statistical tool

Example of regression curves for annual peak flood water level from Mopti to Mopti the 30th of September

Source: Zwarts Léo, 2009, A&W report 241254 Fournet , 2013 in Dewfora D4.8, PIK Calibration (with WFD_ERA40) Scenario A Flood propagation in the IND: SWIM simulation vs. Remote sensing

25 Results Scenario B Climate change impact on annual maximum inundated area

26