UNEP: Ethiopia’s Gilgel Gibe III Dam its Potential Impact on Lake Turkana Water Levels

ACKNOWLEDGEMENTS

RESEARCHERS:
Naga Manohar Velpuri (consultant)
Gabriel Senay (USGS)

REVIEWERS:
Anil Hooda (consultant)
Ashbindu Singh (UNEP)
Charles Sebukeera (UNEP)
Patrick Mmayi (UNEP)
Bruce Pengra (ARTS)

EDITORS:
Tom Adamson (SGT, Inc)
Jane Barr (consultant)
Arshia Chander (SGT, Inc)

DESIGN AND LAYOUT:
Kim Giese (SGT, Inc)

We sincerely thank Mr. Mehret Debebe and Mrs. Azeb Asnake of Ethiopian Electric Power Corporation (EEPCo), Ethiopia for facilitating our field visit to Gibe III site and providing all the useful data for our study. We also thank Dr. John Malala of Kenya Marine and Fisheries Research Institute for sharing Lake Turkana water level data and facilitating our field visit to Lake Turkana.

Executive Summary

Introduction
1. The Gibe III hydroelectric project is the third in a series of five dams the Ethiopian government is constructing
on the Omo River to meet the demands of the region’s power industry. Once completed, it will be the largest
hydroelectric plant in Africa, with a power output of about 1 870 Megawatts (MW), more than doubling Ethiopia’s total installed capacity from its 2007 level of 814 MW (EEPCo 2009b).

2. Lake Turkana is the largest permanent desert lake in the world. It lies in a low, closed basin in northwestern
Kenya and southwestern Ethiopia. As a closed lake, the influx from rivers and evaporation from the lake’s surface determine water-level fluctuations. Of the three rivers that contribute to Lake Turkana—Omo, Turkwel and Kerio—the Omo River, which flows from the Ethiopian highlands, contributes more than 80 per cent (Ricketts and Johnson 1996).

Historical Lake Levels and Previous Studies
3. Lake Turkana experiences seasonal variations in its water level of 1–1.5 m over the year, and it has a longterm natural variability of 5–10 m. Although Lake Turkana’s water level declined considerably over the last
century—by more than 10 m—the trend shows a slight increase over 1992–2010.

4. Analysis of Lake Turkana’s water levels from 1880 to 2008 obtained from Kenya’s Marine and Fisheries
Research Institute (KMFRI) shows that in the late 19th century, Lake Turkana’s water level was about 20 m
higher than the level in 2011 (362.5 m above sea level), followed by a general decline during the first half of the
20th century. The water level decreased to a minimum in the 1950s, and during the early 1900s, it was about 10 m
higher than the 2011 level. There was a rapid increase in the 1960s through to the 1970s, with the level reaching
a height of 366 m during the late 1970s and 1980s.

5. The most recent water-level fluctuations captured by the TOPEX/Poseidon satellite (between 1993 and 2009) show that the lake’s level gradually increased to reach a maximum of about 365 m by the end of the 20th century. However, between 2000 and 2006, the level gradually declined to about 361 m. Satellite altimetry data show that by the end of 2011, the lake’s water level was about 362.5 m.

6. There has been no considerable change in rainfall patterns in East Africa over the last few decades (Cheung
and others, 2008). In our study, analysis of satellite rainfall estimates over the Turkana basin for 1998–2009 reveals a nearly constant overall trend in mean basin rainfall.

7. Avery (2010) determined that the dam would cause the lake’s level to drop by up to 2 m. Salini Constructions
(2010) reported that the potential hydrological impact of Gibe III on Lake Turkana’s water levels would be a loss
of up to 1.5 m during the initial impoundment period. These reports, however, limited their results to the
reservoir’s first impoundment period.

The Present Study
8. This assessment study uses multiple sources of satellite data from 1998 to 2009 and a hydrologic modeling
approach (Velpuri and others, 2012) to study the potential hydrological impact of Gibe III on Lake Turkana
water levels.

9. This study incorporates the operational strategies for the Gibe III dam published by the Ehtiopian Electric Power
Corporation (EEPCo) (2009b), which include discharge from the reservoir: (a) all-time environmental flow of 25 m3
/s, (b) artificial flood release of 1 000 m3/s for 10 days in September each year, and (c) scheduled releases from
the reservoir after power production.

10. This study also considers the potential impact of the dam beyond the reservoir’s first impoundment period
and analyzes the dam’s impact on lake levels under different rainfall scenarios.

Hydrologic impact Assessment: Results
11. Results indicate that the Gibe III dam would moderate the releases into the lake, with inflows decreasing to
around 1 000 m3/s in wet seasons, but with the dry-season base flow increasing to an all-time average of 500 m3/s. Furthermore, the dam would have a greater impact when the basin receives above-normal rainfall and a smaller impact when the basin receives below-normal rainfall compared to conditions without the dam.

12. Three different approaches that use existing satellite data and various future rainfall scenarios were used
to assess the potential impact of the Gibe III dam on the lake’s water levels. The first approach is based
on the simple assumption that the Gibe III dam was commissioned (start of reservoir filling) in the past (on
1 January 1998). Thus, we use the observed historical climatic data for the period 1998–2009 to assess the
dam’s impact had it actually been built at that time. The model’s results show that the Gibe III reservoir would
have reached a minimum operation level (MOL) of 201 m depth of reservoir by August 1998 (in around 220 days). The results further indicate that during the initial period of dam/reservoir filling, the lake level would have dropped by up to 2 m (95 per cent confidence interval). This result is similar to the one reported by Avery (2010).

13. It is impossible to accurately predict the Turkana basin’s future climate. In the second approach, however, we
built future rainfall scenarios based on a knowledge and understanding of the frequency and distribution of rainfall over the Turkana basin region. The climate’s past variability provides the context in which different belownormal (drier) and above-normal (wetter) rainfall years were combined to generate 20 likely rainfall scenarios
and to assess the potential impacts of the dam on lake water levels. The results of this approach show that the
Gibe III reservoir would reach MOL in 8–16 months, depending on the rainfall under different scenarios. When compared to lake levels modeled without the dam, there will be either no change (with above-normal rainfall) or a decline of up to 4.3 m in the below-normal rainfall scenario after the dam is commissioned. Lakelevel variability due to regulated inflows after the dam’s completion was found to be within the lake’s natural variability (4.8 m).

14. The nonparametric resampling technique using the most recent 12 years of satellite data was used to generate several future scenarios of climate data and to evaluate the potential impact of Gibe III. The results indicate that in the median scenario, it would take about 10 months for the Gibe III reservoir to reach MOL. Results also indicate that the average decline in the lake’s level because of the dam would range from 1.5 to 2.3 m (UCI), 1.2 to 2.2 m (Median), and 0.6 to 1.8 m (LCI) under the three rainfall scenarios, respectively. Due to the regulation of lake inflows, the dam would have a greater impact when the basin receives above-normal rainfall and a smaller one when the basin receives below-normal rainfall.

15. Changes in the shoreline or surface area are associated with the lake-level variations. This study identified hot
spots of shoreline change, such as the Omo River delta, Ferguson’s gulf, and the Turkwel-Kerio River deltas,
which all show periodic shrinking and expansion. Further analysis is required to assess the impact of change in seasonal variations on the Omo River flows and the consequent impact on the ecology and fisheries
in the lake.

16. The use of satellite-based data in this study, to estimate runoff and evapotranspiration, makes the modeling approach consistent and robust, especially for a basin where long-term historical runoff and climate data are
scarce. The results obtained under different scenarios will be of great use to planners and others assessing the
hydrological and environmental impacts of the dam under future climatic uncertainty.

Click here to read the full report in PDF format

Here is the conclusion of the study:

Conclusion

Africa’s biggest dam (by height) is under construction on the Omo River. The river contributes more than 80 per
cent of inflows to Lake Turkana. However, the impact of regulated inflows to the lake and potential interactions
between the Gibe III dam and Lake Turkana are not well understood due to its remote location and the dearth of
reliable ground-truth datasets. The overarching goal of this study was to assess the potential hydrological impact
of the Gibe III dam on Lake Turkana water levels using a calibrated water-balance model driven by satellite data for
Lake Turkana. In this study, we analyzed existing data and presented a hydrologic modeling approach that uses multisource satellite data to study the potential hydrological interactions between the Gibe III dam and Lake Turkana
water levels using topo-bathymetry data, satellite-based rainfall estimates, modelled ET, runoff, and satellite altimetry data using different approaches.

Historical data indicate that there were wide fluctuations in Lake Turkana’s water levels in the past. During the late
19th century, the lake’s water level was about 20 m higher than the present day lake level (363 m), followed by a
general decline during the first half of the 20th century.

Lake levels declined to a minimum in the 1950s. There was a rapid increase in the 1960s through the 1970s and it reached a height of 366 m during late 1970s and 1980s. Most recent fluctuations show the lake has been expanding.
Lake Turkana’s level has seasonal variations of 1–1.5 m over a year, with a long-term natural variability of 5–10 m.
The rainfall patterns over East Africa during the last few decades suggest there has been no considerable change
(Cheung and others 2008). MODIS satellite land-cover data for 2001–2009 indicate that the percentage area under each
class has remained almost unchanged, except for areas under grasslands, shrublands and woody savannah, which have undergone minor changes of less than 5 per cent.

We assessed the impact of the Gibe III dam on the lake’s water levels with three different approaches that use
existing satellite data and various future rainfall scenarios. The first approach uses the past climatic data for the period 1998–2009 and the assumption that the rainfall pattern would be same after the dam is commissioned. The results
indicate that during the initial period of reservoir filling, the lake level would drop up to two metres (95 per cent
confidence interval). This result is similar to that reported by Avery (2010). We further found that the Gibe III dam would moderate water releases into the lake: inflows decrease from over 1 500 m3/s to around 1 000 m3/s in wet seasons, but base flow would increase in dry seasons with an all-time average flow rising to nearly 500 m3/s.

It is not possible to predict the future climate in the lake’s basin. To understand the potential climate, we used
a second approach to build future rainfall scenarios based on knowledge and understanding of the frequency and
distribution of rainfall over the Lake Turkana basin. The past variability in the climate provides the context under
which we combined different below-normal (drier) and above-normal (wetter) rainfall years to generate 20 likely
rainfall scenarios and to then assess the impacts of the dam on the lake’s water levels. Based on the results from
this approach, we found that the Gibe III reservoir would reach MOL in 8–16 months, depending on the occurrence
of rainfall under different scenarios. When compared to the lake level modeled without the dam, lake levels will
remain unchanged or decline by up to 4.3 m in the below normal rainfall scenario after the dam is commissioned. The
variability in the lake levels due to regulated inflows after the dam is commissioned is found to be within the lake’s
natural variability.

We employed the nonparametric resampling technique using the most recent 12 years of satellite data to generate
several future scenarios of climate data and to evaluate the impact of Gibe III. Results indicate that the time required for the Gibe III reservoir to reach MOL is about 10 months for the median scenario. Furthermore, results indicate that the average loss in lake levels as a result of the dam would range from 1.5 to 2.3 m (UCI), 1.2 to 2.2 m (Median); and 0.6 to 1.8 m (LCI) under the three respective rainfall scenarios.

The impact of the dam would be greater when the basin receives above-normal rainfall, and the impact would be
smaller when the basin receives below-normal rainfall. In this study, we also identified hot spots along the
Lake Turkana shoreline due to fluctuations in lake levels. Our shoreline-change analysis revealed that regions of
the Omo River delta in the north, Ferguson’s gulf and the Turkwel-Kerio River delta regions in the west and Allia Bay
and the Koobi Fora regions in the east are more susceptible to change. Furthermore, we found that under the below normal rainfall scenario, the lake would shrink up to 5 m and the lake shoreline would show periodic wetting and
drying in up to 20 per cent of the lake-surface area. Under the near-normal scenario, up to 9 per cent of the total lake surface area would experience shrinking and expansion.

Under the above-normal rainfall scenario, the lake would not shrink at all but would expand by up to 10 per cent of the surface area. Further analysis is required to assess the impact of regulated Omo River flows due to the dam on the lake’s ecology and fisheries.

The use of satellite-based data for estimating runoff and evapotranspiration modeling makes the approaches used
in the study consistent and robust, especially for a basin in which there is a dearth of long-term historical runoff and climate data. Results obtained from this study are thus based on observed remotely sensed data and results under
different scenarios will be of great use to planners and others involved in the hydrological and environmental assessment of the dam’s impacts under future climatic uncertainty.

Click here to read the full report in PDF format