Hydrological processes under changing climate and land use scenarios in the Baro– Akobo River Basin, Ethiopia
No Thumbnail Available
Date
2023-11-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Water is an essential component of agricultural productivity and is crucial for food security. It is
also a vital component of the environment. Water security is becoming a global issue, but the issue
is chronic in most developing countries, including sub-Saharan Africa. Therefore, sustainable
water resource management is essential to achieve food and water security in Ethiopia. The Baro–
Akobo River basin, which is found in the southwest part of Ethiopia, has witnessed a substantial
change in population, climate, and land use during the last four decades. The climate of the region
is warming at an alarming rate, and it is expected that this tendency will persist in the coming
century. On the other hand, the impacts of climate and land use change on the hydrology of this
basin are not well understood. Understanding the impacts of climate and land use change on basin
hydrology is critical for developing effective water management practices. Therefore, this study
examined the individual and combined impacts of climate and land use changes on the basin
hydrology.
This study combines a statistical, geospatial, and hydrological model to investigate the hydrologic
impacts of climate and land use change. First, seven raw and bias-corrected RCMs (RCA4
(CNRM), RCA4 (ICHEC), RCA4 (MPI), CCLM4 (CNRM), CCLM4 (MPI), REMO (MPI)) and
the ensemble mean were evaluated for model skill in reproducing the observed baseline climate
for the period 1975–2005 at several weather stations in the basin. A pixel-to-point approach was
used to compare RCMs against weather stations. Monthly distribution patterns and several
statistical metrics were used for evaluating the performance of RCMs in capturing the historical
observed climate of the basin. The Mann-Kendall test and Sen's slope estimate were used to
examine the decreasing and increasing trend of the climatic variables, as well as to estimate the
magnitudes of the significant decreasing and increasing trends. Furthermore, the ensemble mean
and five bias corrected RCMs were used to examine climate and hydrological changes in the
baseline and future 2021–2050 (2030s) and 2071-2100 (2080s) periods under the representative
concentration pathway (RCP4.5 and RCP8.5) scenarios. On the other hand, the geospatial and
synergistic techniques of cellular automata (CA) and artificial neural networks (ANN) were used
to develop historical and future land use change scenarios. The maximum likelihood classifier
(MLC) in the Earth Resource Data Analysis System (ERDAS) was used to conduct land use
change classification for Landsat imagery from 1985, 2002, and 2019. Three land use maps with
seven classes were identified, and then a change detection process was conducted. The classified (1985–2002) and (2002–2019) land use maps, along with a transition matrix and spatial drivers,
were put into the Module for Land-use Change Evaluation (MOLUSCE) model to predict land use
maps for 2019 (current) and 2040 (a business-as-usual scenario) land use scenarios, respectively,
using the CA-ANN multilayer perceptron methods (MLP). Besides, this study considered a further
increase in altitudinal forest expansion and watershed management practices (conservation) in a
land use scenario. Then the best-performing climate model (ensemble mean) under RCP4.5 and
RCP8.5 for the 2023–2040 time frame and plausible land use scenarios considering the current,
business as usual, and conservation was used as inputs to the calibrated models to examine the
individual and combined impacts of climate and land use changes on the hydrology.
Three independent climate datasets, including observed, Climate Hazards Group InfraRed
Precipitation with Stations (CHIRPS), and Climate Forecast System Reanalysis (CFSR), were
used to calibrate and validate the Soil and Water Assessment Tool (SWAT) model for the periods
1990–1998 and 1999–2002, respectively. The Sequential Uncertainty Fitting 2 (SUfI2) method in
the SWAT-CUP program was used for model calibration and sensitivity. The Nash–Sutcliffe
Efficiency (NSE), Coefficient of Determination (R2), and Percent Bias (PBIAS) as well as two
uncertainty measurements (r-factor and p-factor), were used to assess the model's performance.
Then the calibrated model using the water balance equation was used to examine the individual
and combined impacts of climate and land use change scenarios on the basin hydrology. The
climate and land use change impacts on hydrology were analyzed on monthly, seasonal, and annual
scales with respect to the baseline period. Statistical tests such as the t-test and Levene test were
also used to determine the change in mean and standard deviation between the baseline and
different climate and land use scenarios. Besides, the Indicators of Hydrologic Alterations (IHA)
method was used to assess the hydrologic changes between the baseline conditions and future
climate change scenarios.
Results from the global climate models (GCMs) downscaled through the CCLM4, RCA4, and
REMO modeling schemes are characterized by several biases, such as shifting the rainy season
and under- and overestimation of the observed climate. However, the skill of the models was
substantially enhanced after bias correction. All models best capture the annual cycle with less
bias. Therefore, it was beneficial to account for such biases using a robust statistical bias correction
method before utilizing RCM simulations to generate climatic scenarios and climate impact
scenarios. Results from future bias-corrected RCMs show a consistent increase in monthly Tmax and Tmin under RCP4.5 and RCP8.5 in the 2030s and 2080s relative to the baseline climate, while
rainfall does not show consistency. The future climate change projections from the ensemble mean
show an increase in the R20mm, CDD, R95p, RX1, and RX5 indices, but the R10 indices show a
decreasing value under both RCPs.
Observed climate and CHIRPS rainfall combined with the CFSR dataset yielded reasonable and
comparable streamflow simulation performance in terms of statistical metrics at both Baro and
Sore hydrological stations. All climate impact scenarios from the ensemble mean demonstrated a
decline in surface runoff and water yield and an increase in evapotranspiration. Except for the
extreme flow segment (i.e., 0–3% exceedance probability), the projection for simulation under
climate change scenarios shows a decrease in flow. The increase in temperature and the decrease
in rainfall is attributed to a relatively higher impact than the combined and land-use change-alone
scenarios. This will have an impact on future agricultural production and water availability.
Moreover, the projected increase in rainfall extremes, the expansion of agricultural land, and
urbanization all lead to increased surface runoff and flooding. Therefore, to implement adaptation
and mitigation strategies, the inclusion of predicted climate and land use change in hydrological
impact studies is useful.
Description
Keywords
CA-ANN, climate model, evapotranspirtaion, surface runoff,Baro basin