Modeling the hydrological regime of Turkana Lake (Kenya, Ethiopia) by combining spatially distributed hydrological model and remote sensing datasets
c) Gibe I
The model (resolution: 1 km x 1 km, daily) can fairly
• flow time series observed at stations representative of
the northern catchment dynamics (Fig. 4b);
• lake level time series, representative of the whole
catchment dynamics (Fig. 4c)
The lake water budget is almost at equilibrium with
ET = P+Q, where Q is the dominant component (Tab. 1).
A change in the total inflow due to the hydropower reservoir
operations is likely to significantly affect the lake dynamics
(in line with Avery, 2010, Velpuri and Senay, 2012).
The simulated total water storage at the catchment level
(i.e., sub-surface water content and lake) shows fluctuations
similar to GRACE GRGS and CRS mascon (Fig. 4d),
although these estimates are uncertain because of the small
area and shape of the Turkana catchment.
MERRA-2 cloud cover
GRACE total water storage
2m soil moisture
satellite altimetry data
1st sub-surf. layer
2nd sub-surf. layer
1. MOTIVATION AND OBJECTIVES
• Understand and model the hydrological processes of the poorly gauged Lake Turkana catchment.
• Investigate how long-term water budget of the Lake Turkana catchment is affected by the operations of existing and
planned large hydropower reservoirs.
• Understand the Water-Energy-Food nexus in the Lake Turkana catchment, to investigate the impact of planned
hydropower and irrigated agriculture expansion.
Modeling the hydrological regime of Turkana Lake (Kenya, Ethiopia)
by combining spatially distributed hydrological model and remote sensing datasets
D. Anghileri1, A. Kaelin1, N. Peleg1, S. Fatichi1, P. Molnar1, C. Roques1, L. Longuevergne2, and P. Burlando1
1 ETH Zurich; 2 CNRS
3. DATA & METHODS
Ground datasets (northern catchment only):
• Precipitation (14 stations, 1998-2008, daily)
• Temperature (6 stations, 1998-2008, daily)
• Flow (3 stations, 1998-2007, daily)
Remote sensing and reanalysis datasets:
• TRMM-V7 (0.25°x0.25°, 3 hours)
• MERRA-2 (0.5°x0.625°, hourly)
• TOPEX/Poseidon (≈1 per month)
• GRACE (GRGS and CRS mascon)
(300 km X 300 km, daily or monthly)
• GLDAS-NOAH (0.25°x0.25°, hourly)
• Technical features (e.g., capacity)
• Target release/production
Funded under the H2020 Framework
Program of the EU, GA No. 690268
Hydrological model Topkapi-ETH (Fatichi et al., 2015):
• Physically-based, fully distributed model
• Vertical discretization of the subsurface in three layers
• Kinematic approximation to route subsurface, overland, and river flow
• Description of the anthropic component (reservoir operations)
• Endorheic or terminal lake
• Area = 7’500 km2 (approx. 240 km x 30 km)
• Volume = approx. 200 km3
• Area = 148’000 km2
• Elevation = 320 – 2’700 m a.s.l.
• Mean annual precip.= 250 – 1’800 mm/y
• Mean annual temp.= 14 – 30 °C
Hydropower reservoirs in operation:
• Gibe I: 184 MW, 0.7 km3 (commissioned in
• Gibe II: 420 MW, (commissioned in 2010)
• Gibe III: 1870 MW, 14.7 km3 (filling phase
started in 2016)
Hydropower reservoirs planned either:
• Gibe IV and V: 1472 and 560 MW
• Koysha: 2160 MW, 6 km3
5. RESULTS OF HYDROLOGICAL MODEL
• Strong precipitation gradient over the catchment.
• Different rainy patterns depending on the latitude.
• Uncorrected MERRA-2 significantly overestimates
• TRMM-V7 compares better than MERRA-2 to the
observations in the Northern Omo sub-catchment (both
at the monthly and daily time scale).
• TRMM-V7 overestimates precipitation over the lake
(corrected by linear interpolation for the hydrological
Figure 3: a) Map of TRMM-V7 mean annual precipitation over the
catchment; b-e) Comparison of mean monthly precipitation between
observations (when available), TRMM-V7, and MERRA-2 (uncorrected
and corrected) datasets.
Figure 2: Modeling framework: ground, remote sensing, and reanalysis datasets
and schematization of the hydrological processes modeled in Topkapi-ETH (for one grid cell).
Figure 1: a) Satellite image of the Turkana catchment (source: Google map),
and b) map of the catchment including the lake, hydropower reservoirs, major rivers, and available gauges .
2. LAKE TURKANA CATCHMENT
Land 2.85 2.44 0.37 634 (2)
Lake 0.87 8.32 (1) 7.30 634 (2)
Omo 3.86 3.19 0.67 562 (3)
Kerio 1.75 1.62 0.18 35 (4)
Turkwel 1.94 1.81 0.14 37 (5)
(1) 7.6 mm/d Avery, 2010
(2) 650 m3/s Velpuri, 2012
(3) 555 m3/s Avery, 2013
(4) 5 m3/s Avery, 2010
(5) 26 m3/s Sogreah, 1982
Figure 4: a) Map of simulated mean annual actual evapotranspiration over the catchment; b) Comparison of daily simulated and observed flow at Abelti;
c) Comparison of simulated and observed lake level; d) Comparison of simulated sub-surface water storage and GRACE datasets.
Table 1: Water budget components over the whole catchment
(without the lake), lake, and sub-basins of major lake tributaries.
• Simulation of the current and planned hydropower and irrigated
agriculture expansion to estimate the effects on the lake level
• Simulation of the hydrological budget when considering climate
CONTACT: Daniela Anghileri | email@example.com
4. ANALYSIS OF PRECIPITATION INPUT DATA
• Avery (2010). “Hydrological Impacts of Ethiopia’s Omo Basin on Kenya’s Lake Turkana Water
Levels & Fisheries. Final report prepared for the African Development Bank, Tunis.”
• Avery (2013). “What Future For Lake Turkana”. In: African Studies Centre, Oxford University.
• DAFNE: http://dafne-project.eu/
• Fatichi et al. (2015), JoH 525, 362–382.
• Sogreah, Consulting Engineers (1982). “Regional Development Plan for the Kerio Valley,
Water Resources Study, for Kerio Valley Development Authority, Kenya”.
• Velpuri et al. (2012). HESS, 16, 1-18.
• Velpuri and Senay (2012). HESS, 16.10, 3561–3578.
DAFNE poster presented at the 2017 AGU Fall Meeting