Cogent Food & Agriculture

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/oafa20

Modeling hydrological characteristics based on

land use/land cover and climate changes in Muga
watershed, Abay River Basin, Ethiopia

Tatek Belay & Daniel Ayalew Mengistu

To cite this article: Tatek Belay & Daniel Ayalew Mengistu (2024) Modeling hydrological
characteristics based on land use/land cover and climate changes in Muga
watershed, Abay River Basin, Ethiopia, Cogent Food & Agriculture, 10:1, 2319935, DOI:
10.1080/23311932.2024.2319935

To link to this article: https://doi.org/10.1080/23311932.2024.2319935

© 2024 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 23 Feb 2024.

Submit your article to this journal

Article views: 1061

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=oafa20
Cogent Food & Agriculture
2024, VOL. 10, NO. 1, 2319935
https://doi.org/10.1080/23311932.2024.2319935

Food Science & Technology | Research Article

Modeling hydrological characteristics based on land use/land cover and
climate changes in Muga watershed, Abay River Basin, Ethiopia
Tatek Belaya and Daniel Ayalew Mengistub,c
a
Department of Geography and Environmental Studies, Debre Tabor University, Debre Tabor, Ethiopia; bDepartment of Geography
and Environmental Studies, Bahir Dar University, Bahir Dar, Ethiopia; cGeospatial Data and Technology Center, Bahir Dar University,
Bahir Dar, Ethiopia

ABSTRACT ARTICLE HISTORY

Land use/land cover changes and climate change can have profound effects on hydrology Received 16 June 2023
within watersheds. The Soil and Water Assessment Tool model was used to evaluate the Revised 8 January 2024
individual and combined impacts of land use/land cover and climate on the hydrology of the Accepted 13 February
2024
Muga watershed. The model was calibrated and validated based on observed streamflow
data. The calibrated and validated model demonstrated good performance, as indicated by KEYWORDS
Nash Sutcliff Efficiency values of 0.79 for calibration and 0.82 for validation. Similarly, the Climate; hydrology; land
Percentage of Bias values of 8.2% for calibration and 9.3% for validation indicate a small use/land cover; soil and
overall bias in the model’s predictions. According to the findings of this study, there is an water assessment tool
overall increase in surface runoff (15.78% in 2022 and 15.28% in 2038) and water yield (1.22% REVIEWING EDITOR
in 2022; 0.83% in 2038), along with a reduction in lateral flow (8.55% in 2022; 7.0% in 2038), Manuel Tejada,
groundwater flow (9.21% in 2022; 10.0% in 2028), streamflow (5.8% in 2022; 3.3% in 2038), Universidad De Sevilla,
and evapotranspiration (1.54% in 2022; 0.95% in 2038), compared to the baseline year (1994). Spain
The land use/land cover changes in 2022 and projected for 2038 contributed to these
alterations in hydrological parameters. The combined effect of land use/land cover and climate SUBJECTS
change on water balance is relatively higher than the impact of land use/land cover change Hydrology; Surface
scenario alone. The intensified rainfall contributes to the observed increase in surface runoff Hydrology; Climate
Change
and water yield. Hence, the study is crucial to implementing sustainable land use practices
and appropriate watershed management strategies.

IMPACT STATEMENT
Climate change is a global concern affecting many landscape characteristics. Land use changes
as an important variable in soil erosion are expected to change under projecting climatic
scenarios. Evaluation of future soil erosion is crucial for devising appropriate interventions.
Moreover, research results from such studies contribute to planning, decision-making, and
policy development which could make possible sustainability of natural resources.

1. Introduction has multiple effects on the hydrological cycle, affect-

ing water resources (Swain et al., 2020). Population
Land use/land cover (LULC, hereafter we call it) and
growth, climate change, and land use policy are
climate change are two major factors that affect some of the variables that have caused substantial
water resources (Safriel, 2007). Recently, these threats changes in land cover and land use, which have an
have become among the most important ecological impact on the hydrological system at the watershed,
problems worldwide (Nkonya et al., 2016). The poten- sub-basin, and regional levels (Birhanu et al., 2019).
tial impacts of LULC and climate change could lead Water resource is influenced by diverse land uses,
to unforeseen future water resource crises (Talib & including cropland, built-up expansion and forest,
Randhir, 2017). Hence, spatiotemporal LULC and cli- and grassland decline. Additionally, climate change
mate change analysis are essential in water resources has made the effects of population growth, poverty,
management (Afzal & Ragab, 2020). Climate change and rapid urbanization on the depletion of natural

CONTACT Tatek Belay [email protected] Department of Geography and Environmental Studies, Debre Tabor University, Debre Tabor,
Ethiopia
© 2024 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unre-
stricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the
posting of the Accepted Manuscript in a repository by the author(s) or with their consent.
2 T. BELAY AND D. A. MENGISTU

resources worse (Gebre & Ludwig, 2015). Climate focused on the impact of either LULC change (Belihu
change (i.e. temperature and precipitation) will likely et al., 2020; Tekleab et al., 2014; Woldesenbet et al.,
continue to change in the 21st century (Gizaw et al., 2017) or climate change (Abdo et al., 2009; Ayele
2017), leading to water scarcity. A change in LULC et al., 2016) on hydrology. Although the distinct and
also changes surface runoff characteristics, ground- combined impacts of LULC and climate changes on
water recharge, and evapotranspiration. hydrology have been studied, there are limited stud-
In Ethiopia, temperatures have exhibited a notice- ies on their combined effects on hydrology, particu-
able upward trend over the last three to five decades, larly in the Abay River Basin. Some studies in Ethiopia
especially the minimum temperatures (Abebe, 2017; showed that climate change affects hydrological pro-
Berhane et al., 2020; NMA, 2007). In contrast, most cesses more than changes in LULC (Mekonnen et al.,
studies in Ethiopia have indicated that rainfall shows 2018). The impacts of LULC change, however, have
no distinct trend, either positive or negative (Abebe, been found in some studies to be more important
2017; Berhane et al., 2020; Kiros et al., 2016). than the influences of climate change (Woldesenbet
LULC change, like climate change, is the main et al., 2018).
component of environmental change in many parts The results of the previous studies did not provide
of the world. In Ethiopia, the conversion of LULC a conclusive impact of LULC and climate change on
from natural vegetation to farmlands, grassland hydrological processes. For instance, the increase in
lands, human settlements, and built-up centers is surface runoff and total water yield attributed to
common (Demissie et al., 2017). However, LULC LULC changes partially offset the decrease in surface
changes in trend, pattern, type, and magnitude were runoff and total water yield caused by the effects of
heterogeneous across the country (Demissie et al., climate change. On the contrary, moist climate con-
2017). LULC change, combined with climate change, ditions exacerbated the increase in streamflow
affects the water resources of Ethiopia’s highland caused by LULC change (Woldesenbet et al., 2018).
areas (Setegn et al., 2014). Changes in precipitation, The combined impact of future LULC and climate
temperature, and LULC affect water resources through change on hydrology was also the subject of a few
interception, evapotranspiration, runoff, evaporation, studies (Teklay et al., 2021; Woldesenbet et al., 2018).
and surface infiltration, thereby affecting the process However, the context is still not well understood at
of watershed hydrology (Bewket & Sterk, 2005). the watershed level. The influence of LULC and cli-
Rapid population growth adds pressure to the mate change on natural resources such as soil and
effects of climate and LULC changes on water avail- hydrology remains a contentious issue and requires
ability, which negatively affects nations’ agricultural further research (Demessie, 2015; Simane et al.,
sector and food supplies (Kurukulasuriya & 2013). Hence, examining the impact of LULC and cli-
Rosenthal, 2013). Therefore, to mitigate their impacts mate change on hydrology provides important infor-
and improve overall management practices, it is mation for managing and conserving water resources
crucial to understand how changes in land use and at a watershed level.
climate affect the availability and quality of water Large-scale irrigation and hydroelectric projects in
resources. LULC change affects the hydrological pro- the Abay River basin, including the Grand Ethiopian
cess (Birhanu et al., 2019). Climate change also Renaissance Dam (GERD), are being constructed. This
affects the hydrological cycle (Bewket & Sterk, 2005; study was conducted in Muga watershed, one of the
Demessie, 2015; Taye et al., 2015) and the manage- Abay River Basin tributaries from Choke Mountain.
ment of water resources (Jury & Funk, 2013; Kim & The study area and its environs are severely affected
Kaluarachchi, 2009). Therefore, studying the impacts by land degradation due to the severity of LULC and
of LULC and climate changes on water resource climate changes. Moreover, small-scale irrigation
dynamics is vital to providing helpful information projects along the Muga River are being initiated to
on sustainable watershed management and land improve agricultural productivity and the communi-
use policies. ty’s livelihood.
In Ethiopia, several studies were conducted in the Assessing the impact of climate and LULC change
Abay River basin to examine the impact of LULC and hydrology is essential to implement effective water-
climate change on hydrology (Birhanu et al., 2019; shed management interventions and local climate
Mengistu et al., 2021; Tekleab et al., 2013). In the adaptation strategies. Furthermore, the implementa-
Abay River basin, water resources are also highly vul- tion of context-specific interventions is crucial, which
nerable to LULC and climate change (Berihun et al., necessitates conducting specific studies in areas of
2019; Teklay et al., 2021). However, most studies concern. However, there has been no study
 Cogent Food & Agriculture 3

conducted to date that provides scientific evidence categorized into three agro-ecological zones:
on the impact of past and future climate and LULC Weynadega (1500–2300 m), dega (2300–3200 m), and
change on the hydrological process in Muga water- Wurch (above 3200 m). The area experiences an aver-
shed. This knowledge gap is crucial as it allows for a age annual rainfall of 1058.8 mm, with over 85%
more comprehensive understanding of the environ- occurring during the wet season (May–October). The
mental changes occurring in the watershed. Therefore, mean monthly minimum and maximum tempera-
this study is crucial and necessary for activities such tures are about 9.3 and 23.7 °c, respectively (Belay &
as soil and water conservation, land use planning, Mengistu, 2019).
climate change adaptation, and mitigation interven- As per the FAO soil map unit classification system,
tions. Considering all the aforementioned issues, the the major soils in the study watershed are eutric
objective of this study was to quantify the individual cambisols, eutric vertisols, haplic luvisols, and haplic
and combined effects of LULC and climate changes nitisols. The organic content of the soils in the study
on hydrological characteristics in Muga watershed, watershed ranges from 2.09 to 3.22, and the soil pH
located in the Abbay River basin of Ethiopia. values range from 5.5 to 6.0 (BCEOM, 1998).
The main land use types found in the study water-
2. Materials and methods shed include cultivation, grassland, forest, and bush-
land. Due to the agroecological variation of the
2.1. Study area
watershed, various plant and tree species flourish in
The study area is located between 10° 05’ N to 10°44’ the area. The livelihood of the local community pri-
N and 37° 49’ E to 38° 9’ E (Figure 1). The total area marily engaged in agriculture with a mixed farming
coverage of the study watershed is about 424 km2. system, constituting about 95% of the population’s
According to Hurni (Hurni, 1998) classification sys- sources of income. The location map of the study
tem, the agro-climatic condition of the study area is area is shown in Figure 1.

Figure 1. Location map of the study area.

4 T. BELAY AND D. A. MENGISTU

2.2. Datasets 1998). The HS method requires the daily minimum

and maximum air temperatures as input.
The study used both spatial and temporal, with main
For model calibration and validation, it is essential
datasets including a Digital Elevation Model (DEM),
to have observed streamflow data covering the
satellite images, Ground Control Points (GCPs), soil
entire simulation period. The daily streamflow data
map, land use data, and hydro-meteorological data.
for twenty-five years (1995–2019) at the outlet of the
The study used an integrated method that includes
watershed (Muga near Dejen) was obtained from the
the CA-Markov chain model, and hydrological and
Ethiopian Ministry of Water and Energy Hydrology
climate models to evaluate the effects of LULC and
Department. The streamflow data gaps were filled
climate change on the water balance of Muga
with statistically similar day averages (long-term
watershed.
averages) from earlier years with available data, as
suggested by Perry and Hollis (Perry & Hollis, 2005).
2.2.1. Hydro-meteorological data The geographical location of meteorological sta-
Based on coupled hydro-climatic data and physical tions and their mean annual rainfall, minimum and
characteristics of the watershed, the hydrological maximum temperature for the reference (1994–2022)
processes of the watershed were simulated. and future (2023–2055) periods of Muga watershed
Long-term (1994–2022) daily rainfall, maximum and are given in Table 1.
minimum temperature, solar radiation, relative Climate data from Six Regional Climate Models
humidity, and wind speed were required for SWAT’s (RCMs) forced by two Representative Concentration
daily calculation of hydrological processes. The grid- pathways (RCPs), RCP4.5 and RCP8.5 were used. The
ded rainfall and temperature data with the spatial projected future climate change in the study area
resolution of 4 km * 4 km was obtained from the from 2023 to 2055 was based on data that was
National Meteorological Services Agency (NMSA) of acquired from the Coordinated Regional Downscaling
Ethiopia. Experiment (CORDEX)-Africa. The future climate data
Ground-observed wind, relative humidity, and for the SWAT model were daily rainfall and maximum
solar radiation data are not available. Therefore, and minimum temperatures.
Climate Forecasting System Reanalysis (CFSR) data The stochastic weather generator (WXGEN) is used
were used for these parameters. The global weather to generate climate variables for 2023–2055, encom-
data is accessible on the SWAT website as well passing solar radiation, relative humidity, and wind
(http://globalweather.tamu.edu/). It is worth noting speed (Sharpley & Williams, 1990). The stochastic
that the CFSR data is widely used for SWAT model weather generator is commonly used in climate
applications, particularly for variables such as wind change studies (Cousino et al., 2015; Fan & Shibata,
speed, relative humidity, and solar radiation, which 2015; Ficklin et al., 2009). Hence, a stochastic weather
are often unavailable at desired spatiotemporal generator is used in this study to predict specific
resolution. future meteorological variables (solar radiation, rela-
The Hargreaves-Sami (HS) method was used to tive humidity, and wind speed) based on historical
compute potential evapotranspiration (PET), which is statistics of the observed data and climate informa-
used as input for the SWAT model. FAO recommends tion. These data were used as input for the SWAT
the HS as an alternative method when observed model to predict the future impacts of climate
meteorological data are unavailable (Allen et al., change on hydrology.

Table 1. Mean annual minimum and maximum temperature (oC) and precipitation (mm) for 1994–2022 and 2023–2055
periods.
Future climate (2023–2055)
Coordinate location Baseline period (1994–2022) RCP 4.5 RCP 8.5
Rainfall Rainfall Rainfall
Station Latitude Longitude Tmin (oC) Tmax (oC) (oC) Tmin (oC) Tmax (oC) (mm) Tmin(oC) TmaxoC) (mm)
Debre Work 10.7 38.2 10.4 24.3 956.3 10.9 24.0 1329.7 11.2 24.0 1430.2
Felege Birhan 10.7 38.1 10.3 23.8 1193.3 11.7 25.3 1263.8 11.8 25.4 1309.1
Kuy 10.5 38.0 9.3 23.7 1058.7 9.4 22.2 1275.5 9.8 22.2 1277.1
Rob Gebeya 10.6 37.8 8.3 23.2 1253.8 9.4 22.2 1275.5 9.8 22.2 1277.1
Yetemn 10.3 38.2 10.2 24.0 1096.6 13.8 27.1 1307.0 14.1 27.2 1392.4
Yetnora 10.3 38.1 10.2 23.4 1034.8 12.5 22.2 1073.5 12.8 25.3 1109.0
 Cogent Food & Agriculture 5

2.2.2. Land use land cover data images of the research area. The satellite images of
Landsat thematic mapper (TM), enhanced thematic the study site were divided into five land uses (Table
mapper (ETM+), and operational land imager (OLI) 2). The GEE platform (Gorelick et al., 2017), a highly
time series images were obtained. The Landsat effective free cloud platform for processing and
images for the years 1994, 2008, and 2022 were uti- analyzing satellite image data (Xiong et al., 2017),
was used for the majority of tasks in this investigation.
lized in the Google Earth Engine (GEE) image classi-
Google Earth images were utilized to extract refer-
fication process to produce LULC maps of the study
ence data and 430 ground control points were col-
area. To ensure cloud-free images, all images were
lected for each year using a random sampling
taken during the dry months of December and
technique. The mean pixel values of the spectral sig-
January. In addition to historical LULC data, topo-
natures were used in GEE to extract these sample
graphic maps and SPOT images were used to gather
training data. The datasets served as a training test
ground control points for the two historical periods
for machine learning algorithms used to categorize
(1994 and 2008).
the types of land use and cover. The data used to
The Ethiopian Geospatial Information Institute
train the classifier must be kept apart from the data
(EGII) provided topographic maps from 1994 with a
used to assess its accuracy to ensure a ‘fair’ evalua-
scale of 1:250,000 and SPOT images for 2008 with a
tion of the classifier’s generalization. As a result,
spatial resolution of 5 m, respectively. Five LULC
training and test sets of labeled data are often cre-
classes, i.e. forest, shrubland, grassland, cropland, and
ated. For each LULC type, the sample size was deter-
built-up area, were identified based on field visits,
mined by the percentage of its area.
information from local informants, researchers’ prior
The researchers experimented with different num-
knowledge, image classification results, and visual
bers of training samples (a trial-and-error process)
interpretation using Google Earth historical function
until the optimal amount that could achieve high
and supportive supplementary data. Based on the
accuracy in classification. The training (70%) and test-
principles of ‘complete consistency’ and ‘temporal
ing (30%) datasets were separated from the total data.
stability’ (Hu & Hu, 2019); sample points were filtered
Finally, test data were used to evaluate how well the
from five land cover types.
modified RF model performed. The test datasets’ LULC
classes were subsequently predicted using the trained
2.2.2.1. LULC classification. Machine learning-based
image classifiers are very helpful for recognizing model. The user accuracy values and kappa coeffi-
patterns in complex functional spatial characteristics cients were used to examine the performance of RF
(Richards & Richards, 1999). Using remote sensing (Congalton & Green, 2019). The accuracy of the map
data, one of the most widely used algorithms for based on the data was evaluated by error matrices,
classifying land cover is random forest (Breiman, which were used to compare the areal size of the
2001). Several researchers carried out a crucial study classes in the classification results to the reference
on LULC classification on GEE using random forest data set. Using the training datasets for each study
methods, and the results were outstanding (e.g. year, we then used the trained model to predict five
Foody & Mathur, 2004; Luo et al., 2021; Magidi et al., LULC classifications. The validated classes were used to
2021). Thus, the ee.smileRandomForest function in assess producer and user accuracy (Santos et al., 2022).
the GEE was used to execute the LULC class. Thus,
five land uses were identified from the satellite
2.3. Markov chain, cellular automata (CA), and
Table 2. Land-use types of the study area. CA-Markov models
S. N Class name Description
1 Cropland Areas that are used for agricultural activities In the study, the CA-Markov chain model was used
and rural farmsteads closely associated to predict the LULC in a study area for the year 2038
with the croplands.
2 Grassland Land primarily covered with grasses and based on LULC maps of 1994, 2008, and 2022. The
allocated as a source of animal feed model used transition suitability images and Markov
3 Forest Areas covered with nearly closed canopies
of forest are dominated by natural and transition parameters to determine the likelihood of
plantation trees, which are relatively tall, change. The CA-Markov chain model is a powerful
dense trees.
4 Shrubland Shrubland dominated area with small
integration of the Markov chain model and the CA
isolated trees always with a lower range model for the prediction of spatial LULC change
of grass. (Halmy et al., 2015). The CA Markov model incorpo-
5 Built-up Areas characterized by artificial surfaces,
such as residential areas occupied by rates both non-spatial and spatial data as input to
living houses and constructions simulate LULC changes. Datasets, including land use
6 T. BELAY AND D. A. MENGISTU

land cover maps (1994, 2008, and 2022), topography, changes in land use on the hydrological response of
land suitability maps, proximity to infrastructure, the watershed. The study used the historical (1994,
transportation networks, proximity to rivers, and 2008, and 2022) and predicted LULC maps of 2038 to
other spatially explicit datasets, were used as the show the study watershed’s hydrological response to
basis for defining the rules and constraints governing LULC changes. The SWAT model requires the conver-
the transitions between LULC types in the CA model. sion of the LULC types into four-digit SWAT codes.
Through the examination of relevant literature The SWAT database assigned the land use codes
(Bewket & Teferi, 2009; Gidey et al., 2017; Hishe et al., such as AGRL, FRST, SHRB, GRAS, and URBN, to rep-
2020; Teferi et al., 2013) and engagement with stake- resent cropland, forest, shrubland, grassland, and
holders, such as farmers, local experts, and officials built-up areas, respectively.
from regional land bureaus, and taking spatial cor-
relations into account, we identified potential drivers
of LULC change. These drivers were further analyzed 2.4.1. Soil and geological information
to understand their specific impacts on land-use The soil physical and chemical properties parameters
dynamics in the study area. In our investigation, a required by SWAT determine runoff factors. The study
digital elevation model with a resolution of 30 watershed has five types of soil, the details of these
meters, derived from ASTER data, was used to create soil types can be found in (Belay & Mengistu, 2021).
slope, elevation, and stream network data. To incor- To create a soil map for the SWAT (Soil and Water
porate road network data, researchers obtained it Assessment Tool) model, a scale of 1:250,000 was
from OpenStreetMap (https://www.openstreetmap. used. The soil map was obtained from the Ministry
org/), a collaborative mapping platform that provides of Water & energy of Ethiopia. A soil database was
freely available geospatial data. Population data, a prepared that includes information on the physical
crucial factor in LULC analysis, was sourced from the and chemical properties of each soil layer.
Ethiopian Central Statistical Agency (CSA). The data-
sets, which include the DEM, road network, and pop-
ulation, were processed and analyzed using ArcGIS 2.4.2. Digital elevation model (DEM)
software. After processing and analyzing the data in The study used DEM data from ASTER to derive
ArcGIS, the resultant datasets and potential transition hydrologic parameters and hydrological response
maps were exported to Idrisi Selva software for fur- units (HRUs) set up in the ArcSWAT interface. In addi-
ther analysis. Iterations were performed within the tion to the soil data, topographic parameters were
CA-Markov model to refine and improve the accu- used to drive the SWAT model for the watershed. The
racy of the predicted LULC maps (Arsanjani stream network is a critical component of watershed
et al., 2011). hydrology and is used to simulate the flow of water
through the channels. In this study, the stream net-
work data was validated through a process called
2.4. CA-Markov model validation ground truthing, which involves physically verifying
the presence and location of streams through field-
Model validation is a crucial step in assessing
work. Based on the topographic features, the SWAT
the performance and accuracy of a predictive model.
model identified 13 sub-watersheds within the main
The kappa coefficient Index was used to evaluate the
watershed under study.
performance and accuracy of the CA-Markov model
in predicting LULC changes in the study area. Thus,
the accuracy of the simulated 2022 LULC map was 2.4.3. Climate change scenarios
evaluated by valuing the degree of agreement In the study, bias-corrected Regional Climate Models
between a simulated and actual map of 2022. Kappa (RCMs) were used as inputs for the SWAT model.
statistics are used to test accuracy using various These RCMs are driven by Global Climate Models
measures, including traditional Kappa (Kstandard), (GCMs) and provide regional-scale climate projec-
Kappa for no information/ability (Kno), Kappa for loca- tions. The ensemble mean of six RCMs is used as the
tion (Klocation), and quantity of correct cells (Kappa for input for the SWAT model. The climate data used in
quantity). Variations of kappa have been strongly rec- the study has a resolution of 50 × 50 km. The climate
ommended and widely used to validate LULC change data for the study was obtained from the CORDEX
predictions (Pontius, 2002; Singh et al., 2015). (Coordinated Regional Climate Downscaling
The use of the LULC map is a crucial part of the Experiment) -Africa data portal (https://esgf-data.dkrz.
SWAT model, as it helps to understand the impact of de/search/cordex-dkrz/). CORDEX is an international
 Cogent Food & Agriculture 7

program that focuses on producing downscaled cli- (Teutschbein & Seibert, 2012). These methods are
mate information for regional studies. The commonly employed to adjust climate model data
CORDEX-Africa data portal specifically provides cli- and improve their accuracy for impact assessments.
mate data for the African continent. The details concerning the types and sources of
The selection of the six RCMs in our study was datasets used in this study are listed in Table 3
based on a comprehensive review of existing studies (Table 4).
on climate change conducted by different scholars in The result indicates that the bias correction tech-
Ethiopia (Alemseged & Tom, 2015; Teklesadik et al., nique had a positive effect on the simulated data.
2017; Worku et al., 2018). These studies found that This is demonstrated by the decrease in RMSE values
the chosen RCMs are suitable for impact assessment and the fact that the standard deviation values
studies. Additionally, we considered their applicabil- became closer to the observed data. Moreover, there
ity to our specific study site. Our prioritization was was a significant improvement in the correlation
based on the success demonstrated by these RCMs between the simulation and observation data. The
in similar contexts, aligning with the objectives and bias correction performed in the study was deemed
scope of our research. The study incorporates the satisfactory and consistent with previous research
two representative concentration pathways (RCPs) findings, as reported by Teklesadik et al. (Betrie
recommended by the Intergovernmental Panel on et al., 2011).
Climate Change (IPCC) for climate change impact
and adaptation studies (Stocker et al., 2013). Two
pathways, namely RCP4.5 and RCP8.5, are consid- 2.5. The soil water assessment Tool (SWAT)
ered. RCP4.5 represents an intermediate stabilization The Soil Water Assessment Tool (SWAT) is a physically
scenario, assuming the implementation of appropri- based and semi-distributed hydrological model
ate adaptation measures and efforts to reduce green- developed at the USDA-ARS (Arnold et al., 2012).
house gas emissions. On the other hand, RCP8.5 SWAT used to study large-scale and complex water-
represents a high-emission scenario, assuming mini- sheds for each hydrological response unit (HRU),
mal to no effort to reduce greenhouse gas emissions units with similar soil types, land use, and slope
(van Vuuren et al., 2011). classes (Arnold et al., 2012). HRUs depict the essen-
The study utilized power transformation bias cor- tial soil water content, nutrient cycles, surface runoff,
rection for precipitation and linear scaling bias cor- crop growth, sediment yield, and simulated manage-
rection methods for temperature minimum and ment practices.
maximum, as suggested by Teutschbein & Seibert Some researchers have used the SWAT model
across different watersheds in Ethiopia [e.g. (Betrie
Table 3. Dataset sources and types (1994, 2008, and 2022) et al., 2011; Gessesse et al., 2015)]. They proved that
and predicted LULC (2038). the model helps to examine erosion and streamflow
No Data Sources of watershed responses to LULC and climate change.
1 DEM (30*30 m) https://earthexplorer.usgs.gov/
2 Soil (Scale 1:250,000) Ministry of Water & Energy of
Ethiopia Table 4. List of RCMs used in the CORDEX-Africa dataset for
3 Land use and land cover https://earthexplorer.usgs.gov/ this study.
(1994, 2008 & 2022)
4 Daily meteorological data of National Meteorological Agency Institution RCM name Country Resolution
the 7 weather stations in of Ethiopia http:// NOAA GFDL: GFDL-ESM2M USA 50 km*50km
and around the watershed globalweather.tamu.edu/). Geophysical Fluid
from 1994 to 2022 Dynamics
Precipitation Laboratory
Maximum & minimum CSIRO-Commonwealth CSIRO-Mk3.6.0 Australia 50 km*50km
temperature Scientific and
Relative humidity Industrial Research
Wind speed Organization
Solar radiation CCCma: Canadian CanESM2 Canada 50 km*50km
5 Daily river discharge data of Ministry of water and energy of Centre for Climate
Muga from 1994 to 2022 Ethiopia Modelling and
6 Road https://www.openstreetmap.org Analysis
7 Population dataset Central Statistical Agency of MOHC: Met Office HadGEM2-ES United 50 km*50km
Ethiopia Hadley Centre Kingdom
8 Ground control points Collected from field using GPS MIROC: Developed by MIROCS Japan 50 km*50km
9 CORDEX data (50*50 km) Earth System Grid Federation the Japanese
(ESGF) research community
https://esgf-node.llnl.gov/ NCC: The Norwegian NorESM1-M Norway 50 km*50km
projects/esgfllnl/. Climate Centre
8 T. BELAY AND D. A. MENGISTU

In this study, the runoff response for the rainfall Where NSE is the Nash-Sutcliffe coefficient, PBIAS
events was simulated using the Soil Conservation is the percentage of deviation, Qobs is the observed
Service Curve-Number (SCS-CN) method (SCS, 1972). data, Qsim is simulated data, and Qmean is mean
To obtain details about SCS-CN, one can refer to ear- streamflow.
lier studies (Betrie et al., 2011; Schneiderman et al.,
2007; SCS, 1972). The methodology used in this
study to assess the impacts of LULC and climate 3. Results and discussion
change on hydrology using the SWAT model is 3.1. Accuracy assessment
depicted in Figure 2.
Before using image classification results for further
analysis, it is compulsory to assess their accuracy to
2.5.1. Performance measures evaluate the percentage of pixels classified correctly
The model’s performance can be evaluated through and incorrectly per LULC cluster. Overall, the accu-
various measures of efficiency. Nash-Sutcliffe racy of the 1994, 2008, and 2022 LULC maps was
Efficiency (NSE) and the Percentage of Bias (PBIAS) assessed by independent datasets using the kappa
are most commonly used (Moriasi et al., 2007). We coefficient, overall accuracy, and producer and user
also used the NSE and PBIAS to determine the model accuracy. This assessment helps to determine the
performance. quality and reliability of the LULC maps. The results
showed that the overall accuracy for 1994, 2008, and
 ∑ n ( Q − Q )2  2022 images was 86.4, 87.6, and 89.5%, respectively,
NSE = 1−  ni =1 (1)
obsi simi

 ∑ ( Qobsi − Qmean )2  with kappa statistics values of 0.86, 0.88, and 0.91
 i =1  (Table 5). This indicates a high level of confidence in
the accuracy of the LULC data used in the subse-
quent analysis (Congalton & Green, 2019).
 ∑ n ( Qobsi − Qsimi )  The validation of simulated and actual LULC maps
PBIAS = 1−  i =1 n  × 100(2) for 2022 also reveals consistency between the actual
 ∑ ( Qobsi ) 
 i =1  and simulated LULC Maps of 2022. The Kappa

Figure 2. Flowchart representing the overall framework SWAT based modeling .

 Cogent Food & Agriculture 9

statistics show that the values for kno (0.89), kloca- period. During the study period from 1994 to 2008,
tion (0.83), klocationstata (0.85), and kstandard (0.82) forest areas exhibited a growth rate of approximately
are all greater than 0.8, indicating the model’s accu- 0.98% per year. However, in the second period, there
racy to predict future LULC changes (Mondal was a higher rate of forest reduction. The increase in
et al., 2016). forest cover during the first period can be attributed
to improved land use and management practices
implemented by the local community, including
3.2. Land use land cover changes analysis activities such as rehabilitation, area closure, and
The result of the classification identified five domi- reforestation.
nant LULC classes, namely cropland, forest, shru- The results showed that the growth of cropland
bland, grassland, and built-up areas. The spatial led to a decline in forest and grassland, which aligns
pattern and extent of LULC types extracted from with previous studies in Ethiopia (Berihun et al.,
classified images showed that cropland was the most 2019). Furthermore, a significant decline in grassland
dominant in the study watershed during the study was observed in some parts of the catchment, which
periods. To assess the trend and evolution of histori- aligns with other studies in Ethiopia (Abdulahi et al.,
cal land use changes over time, land use data for the 2016). The increase in forest land from 1994 to 2008
years 1994, 2008, and 2022 were used. was attributed to the establishment of Eucalyptus
The results of the study indicated that between plantations in the study area. However, the continu-
1994 and 2022, there was an increase in cropland ous expansion of built-up areas at the expense of
and built-up areas by 13.3 and 55.3%, respectively. grassland and shrubland resulted from infrastructural
However, during the same period, there was a forest development in the catchment. In general, the results
decline, grassland, and shrubland by 2.6, 62.4, and of this study highlight the need for sustainable land
52.7%, respectively (Table 6). These changes in land management practices in the Muga watershed to
use are important for understanding the hydrologic ensure the conservation of natural resources and to
processes in the study watershed, as they can signifi- mitigate the negative impacts of LULC changes on
cantly impact the water balance and affect the quan- the environment.
tity and quality of water resources in the area. The According to the LULC map of 2038, cultivated
study watershed is primarily dominated by cropland land remains the dominant land use type within the
as the main land use type, as indicated in Table 7. study watershed, accounting for about 83.1% of the
For detailed information regarding the area coverage study area. During the period from 2022 to 2038,
and trends of each land use land cover type, anyone there has been a decrease in the area covered by
can refer to Table 6. forests by 18.7%. However, built-up areas and grass-
The study revealed a consistent increase in crop- lands have experienced an increase of 25.7 and
land at a rate of 48.4 ha per year during the first 24.5%, respectively. It is worth noting that the
period (1994–2008) and a higher rate of 254.4 ha per increase in grassland area is likely attributed to the
year during the second period (2008–2022). Similarly, alteration of shrubland into grasslands, as indicated
built-up areas demonstrated an increase of 8.6 hect- in Table 6.
ares per year during the first period and a higher Figure 3 and Table 7 depict the expansion of urban
rate of 34.6 hectares per year during the second areas, loss of forested areas, and increase in degraded
or barren lands resulting from overgrazing or other
Table 5. Accuracy assessment for the years 1994, 2008, and factors. Based on these findings, one can conclude
2022 classification. that if the community persists with unsustainable
1994 2008 2022 practices, the watershed’s condition is likely to deteri-
Overall accuracy (%) 86.4 87.6 89.5 orate further. This deterioration may result in adverse
Kappa statistics 0.86 0.88 0.91
effects including diminished water quality,

Table 6. Spatial pattern of LULC change in the study watershed.

1994–2008 2008–2022 1994–2022 2022–2038
LULC Types Area (ha) % Area (ha) % Area (ha) % Area (ha) %
Cropland 678.0 2.2 4471.0 13.9 3513.0 13.3 −435.5 −1.2
Grassland −350.0 −9.4 −1986.1 −58.5 −1986.1 −62.4 343.9 24.5
Shrub land −433.0 −9.4 −1987.7 −47.8 −1987.7 −52.7 −73.0 −3.4
Forest 178.0 12.0 −933.0 −65.4 −216.0 −2.6 −270.7 −18.7
Built up 168.0 15.4 435.8 34.6 435.8 55.3 435.3 25.7
10 T. BELAY AND D. A. MENGISTU

Table 7. Area coverage and percentage of LULC classes in Muga watershed.

1994 2008 2022 2038
LULC class Area (ha) % Area (ha) % Area (ha) % Area (ha) %
Cropland 31396.2 74.2 32074.2 75.4 35587.2 84.1 35151.7 83.1
Grassland 3742.7 8.8 3392.7 8 1406.5 3.3 1750.4 4.1
Shrub land 4594.7 10.9 4161.7 9.8 2174 5.1 2101 5
Forest 1488.9 3.5 1666.9 3.9 1450.9 3.4 1180.2 2.8
Built up 1091.2 2.6 1259.2 3 1695 4 2130.3 5
Total 42313.7 100 42554.7 100.1 42313.6 99.9 42313.6 100

Figure 3. LULC maps of 1994, 2008, 2022 and 2038.

exacerbating soil erosion, loss of biodiversity, and a The results showed that the SCS runoff curve number
reduction in the ecosystem services provided by the (CN2) was more sensitive to streamflow, which is con-
watershed. sistent with several other studies (Gebremicael et al.,
2013; Melaku et al., 2018; Worqlul et al., 2018).
The absolute values of T-stat and p-Value were
3.3. SWAT calibration and validation
used to rank the sensitive parameters. The T-stat
Successful hydrological models rely heavily on the value measures the size of the difference in terms of
parameters’ calibration and sensitivity analysis variability in the sample data. At the same time, the
(Abbaspour et al., 2018). The model simulations were p-value indicates the probability of observing a test
divided into a warm-up (1995–1997), calibration statistic at least as large as the one calculated with
(1998–2014), and validation (2015–2019) phases. By the assumption that a hypothesis is correct (Figure
following this three-phase approach, the SWAT model 4). A parameter with the highest absolute T-stat
was tuned and tested to simulate the behavior of the value and minimum p-value compared to the other
watershed. Initially, thirteen parameters were chosen parameters was considered highly sensitive
for this study. Finally, seven sensitive parameters were (Abbaspour et al., 2018).
chosen based on their sensitivity values. The cali- The NSE values of 0.79 and 0.82 for the calibration
brated sensitivity parameters of the SWAT model and and validation periods, respectively, indicate a sub-
the adjusted values obtained are shown in Table 8. stantial agreement between the simulated and
 Cogent Food & Agriculture 11

Table 8. List of parameters with fitted values and global sensitivity results for daily flow.
Parameter name Description Parameter range Best sim Rank t-stat P-value
A_CN2.mgt SCS runoff curve number 11.23–14.53 11.56 1 −23.62 0.00
V_ESCO.hru Soil evaporation compensation factor −0.09–0.12 0.02 2 10.46 0.00
V_ALPHA-BF.gw Base flow alpha factor or recession constant (days) 0.32–0.51 0.42 3 8.01 0.00
R__RCHRG_DP.gw Deep aquifer and percolation fraction −0.25–0.12 −0.02 4 −3.21 0.01
A__GWQMN.gw Threshold depth of water in the shallow aquifer 3621.29–3810.43 3693.16 5 2.63 0.02
required for return flow to occur (mm)
R__GW_DELAY.gw Groundwater delay (days) 0.05–0.64 0.07 6 −2.19 0.04
V__SURLAG.bsn Surface runoff lag coefficient −0.07–0.29 0.01 7 2.18 0.05

Figure 4. Sensitivity analysis of parameters for river discharge calibration, p-value shows the measure of sensitivity, the larger
t-values are more sensitive. t-Test represents the significance of sensitivity, the smaller the p-value, the less chance of a param-
eter being by chance assigned as sensitive.

observed streamflow (Figure 5). Moreover, the PBIAS findings are summarized in Table 10, which presents
values are 8.2 and 9.3% for the calibration and vali- the mean annual values of various water balance
dation periods, respectively (Table 9). Thus, the SWAT components for different LULC scenarios. During the
model demonstrates ‘very good’ performance during dry season, the mean monthly streamflow exhibited
both the calibration and validation periods, meeting a decline from 71.1 mm in 1994 to 43.5 mm in 2008
the criteria of NSE >0.5 and PBIAS < ± 25% (Moriasi and 48.6 mm in 2022, with a projected decrease to
et al., 2007). 49.0 mm in 2038. This decline in streamflow is pri-
marily attributed to anthropogenic activities, partic-
ularly changes in LULC. Conversely, the mean
3.4. Impact of LULC and climate changes on the
monthly streamflow during the wet season showed
hydrology of Muga watershed
an increase from 71.1 mm in 1994 to 43.5 mm in
3.4.1. Individual impacts of land use land 2008, further to 48.6 mm in 2022, and is projected
cover changes to reach 49.0 mm in 2038. Additionally, the mean
LULC change due to human activities such as defor- monthly streamflow during the rainy season
estation, loss of vegetation, and increased settle- increased from 133.2 mm in 1994 to 147.0 mm in
ment are the most significant contributors to 2008 and subsequently to 146.8 mm in 2022. In
changes in the hydrological response of a watershed summary, the average monthly streamflow during
(Berihun et al., 2019; Gashaw et al., 2018; Woldesenbet the dry season has experienced a decline from
et al., 2017). The four different periods of land use 71.1 mm in 1994 to 43.5 mm in 2008, 48.6 mm in
considered in the study are 1994, 2008, 2022, and 2022, and is predicted to reach 49.0 mm in 2038. On
2038. The simulations carried out using the SWAT the contrary, the average monthly streamflow during
model revealed several important findings concern- the wet season has observed an increase from
ing the impact of LULC changes on the water bal- 71.1 mm in 1994 to 43.5 mm in 2008, and 48.6 mm
ance components of the Muga watershed. These in 2022, and is projected to reach 49.0 mm in 2038.
12 T. BELAY AND D. A. MENGISTU

Figure 5. Simulated and observed streamflow for calibration and validation periods.

Table 9. SWAT model evaluation parameters for streamflow Groundwater flow decreased by −4.36 and −9.21%
in the Muga watershed. in 2008 and 2022, respectively, and is expected to
Evaluation reduce by 10% in 2038 compared to 1994. Similarly,
Component parameters Calibration Validation the average annual ET was decreased from 592.4 mm
Streamflow NSE 0.79 0.82 in 1994 to 580.1 mm in 2008, 583.3 mm in 2022, and
PBIAS (%) 8.2 9.3
expected to be 586.8 mm in 2038, which is 12.3 mm
lower in 2008, 9.1 mm lower in 2022, and 5.6 mm
The average annual streamflow for the LULC 1994, lower in 2038 compared to 1994. A possible reason
2008, 2022, and 2038 scenarios is 203.3, 190.5, 195.4, for the decline in the average annual ET is the
and 198.6 mm, respectively, which revealed that the decline in forests and shrubland. Similarly, there was
average annual streamflow decreased significantly a negative trend in the average annual lateral flow
during the study period. The decline in mean yearly over the Muga watershed, declining by 3.9% in 2008,
streamflow is due to the expansion of cropland and 8.5% in 2022, and 7.0% in 2038 compared to 1994.
built-up areas and the loss of forests, shrublands, and According to Ayivi and Jha (Ayivi & Jha, 2018), a
grasslands. In contrast, surface runoff increased from slight decrease in groundwater and lateral flow can
190.4 mm in 1994 to 204.6 mm and 220.4 mm in 2008 be attributed to low soil infiltration and high sur-
and 2022, respectively, and is expected to rise to face runoff.
219.5 mm in 2038. These indicate that an increase in The mean annual water yield increased slightly in
cropland and built-up areas at the expense of forests 2008 and 2022 in the study watershed compared to
and shrub-bush lands may be the reason for the 1994. In 2008 and 2022, water yield increased by 2.7
increased rate of surface runoff. The results of this and 5.6 mm, respectively, while water yield in the
study are similar to the previous studies conducted study watershed is expected to increase by 3.8 mm
by (Berihun et al., 2019; Gashaw et al., 2018). They in 2038. The decline of vegetation cover, predomi-
reported that surface runoff increased while stream- nantly forest, increases surface runoff and water
flow declined due to the rapid deforestation of nat- yield. According to Woldesenbet et al. (Woldesenbet
ural forests and expansion of farmland and built-up et al., 2017), the reduction of forest cover leads to
area at the expense of shrubland and grassland. higher water yield. As reported by Aylward (Aylward,
 Cogent Food & Agriculture 13

2005), water yield was increased with a decline in The mean annual water yield increased by 0.6% in
forest cover, and water yield would increase if forests 2008 and 1.2% in 2022 compared to the baseline
were cleared. year, likely attributed to the gradual expansion of
The findings indicate that from 1994 to 2022, cropland. The increase in surface runoff resulted from
some parts of the forest, grassland, and shrubland the decline of forests and shrubland, contributing to
were converted to cropland and built-up areas. This a higher water yield. The anticipated average annual
trend is anticipated to persist in the future, resulting water yield in the future, under the LULC change
in a significant increase in surface runoff and water scenario, is expected to be closely similar to that of
yield of Muga watershed. The results of this investi- the baseline period, with a projected increase of
gation align with earlier studies (Bewket & Sterk, only 0.8%.
2005; Shrestha et al., 2018), confirming that the con- Expansion of residential areas with increased agri-
version of forested areas to cropland is a contribut- cultural activities, reduction in vegetation cover, and
ing factor to the increase of surface runoff. The forest areas leads to more significant areas of imper-
decrease in vegetation cover diminishes canopy vious surfaces and soil compaction, reducing the
interception, resulting in an increased amount of potential for soil infiltration and possibly increasing
rainfall directly reaching the soil surface. Consequently, surface runoff (Woldesenbet et al., 2018). Table 11
this heightened direct rainfall raises the likelihood of shows water balance components under different
increased infiltration and runoff from the land LULC scenarios and historical climate data (1994–
(Rogger et al., 2017). 2023) in the Muga watershed. In general, if the
The increase in surface runoff is linked to a decline incoming water through overland flow is more than
in infiltration (Benegas et al., 2014). Furthermore, the infiltration water, the excess water can cause an
areas affected by deforestation and increased settle- increase in surface runoff. As a result, water yield, a
ment experience a reduction in soil soil infiltration combination of surface runoff, groundwater flow,
capacity. The reduction of soil water infiltration sub- and lateral flow, has also increased in cropland and
sequently leads to a decline in groundwater flow. built-up areas.
Forests play a crucial role in absorbing water through The results of this study showed agreement with
leaves and roots, promoting rainwater infiltration other studies (Birhanu et al., 2019; Gashaw et al.,
into groundwater. The growing demand for ground- 2018; Shawul et al., 2019; Woldesenbet et al., 2018)
water due to a growing population can further exac- which reported an increase in surface runoff and
erbate the reduction in groundwater levels through water yield. For example, Gashaw (Gashaw et al.,
increased abstraction. 2018) showed that a decrease in groundwater flow

Table 10. Average annual evapotranspiration (ET), surface runoff (SURQ), groundwater (GWQ), water yield (WYLD), and lateral
flow (LQ) change under four LULC scenarios and historical climate data (1994–2022).
P STQ (m3/s) Season ET SURQ GWQ LQ WYLD
Year mm mm % Wet Dry Mm % mm % mm % mm % mm %
1994 1087 203.3 133.2 71.1 592.4 190.4 208.5 60.8 459.7
2008 1087 190.5 −7.2 147.0 43.5 580.1 −2.08 204.6 7.46 199.4 −4.36 58.4 −3.9 462.4 0.59
2022 1087 195.4 −5.8 146.8 48.6 583.3 −1.54 220.4 15.76 189.3 −9.21 55.6 −8.55 465.3 1.22
2038 1087 198.6 −3.3 149.6 49.0 586.8 −0.95 219.5 15.28 187.5 −10.0 56.5 −7.07 463.5 0.83
Note: Reference LULC is 1994, P: rainfall; STQ: streamflow.

Table 11. Predicted mean annual water balance components for different combinations of climate change and LULC
scenarios.
STQ ET SURQ GW LQ WYLQ
Scenarios mm Δ% mm Δ% mm Δ% mm Δ% mm Δ% mm Δ%
LULC 2022 & CC 193.4 586.3 217.4 190.3 56.8 463.3
1994–2022
LULC 2038 & CC 220.1 13.8 647.3 10.4 230.2 5.9 185.6 −2.5 55.2 −2.8 517 1.7
1994–2022
LULC 2022 & RCP 4.5 232.2 21.3 652.6 11.3 231.5 6.5 231 21.4 60.8 7 523.3 12.9
LULC 2022 & RCP 8.5 236 22 652.3 11.3 240 10.4 234.2 23.1 65.8 15.8 540 16.6
LULC 2038 & RCP 4.5 229.8 18.8 648.4 10.6 238.7 9.8 229.8 20.8 58.8 3.5 527.3 13.8
LULC 2038 & RCP 8.5 234 21 650.8 11 248.3 14.2 232.5 22.2 62.2 9.5 543 17.2
Note: CC: Climate change; RCP 4.5 & RCP 8.5 (2023–2055); Reference period for change in water balance components is LULC 2022 and CC
1994–2022.
14 T. BELAY AND D. A. MENGISTU

and an increase in surface runoff in Andasa water- under future climate scenarios, an increase in surface
shed in the Abay basin between 1994–2015 were runoff is projected. The projected surface runoff for
associated with changes in LULC of the watershed. the future period is estimated to be around 230.2 mm
According to Shawul [85], between 1974 and 2014, and 231.5 mm per year under RCP4.5 and RCP8.5,
increased surface runoff and decreased groundwater respectively. The findings of this study align with the
flow in the upper Awash River basin were associated results of the study conducted by Chanapathi and
with LULC change. Contrary, when agricultural land is Thatikonda (Chanapathi & Thatikonda, 2020). This
plowed, compaction of the lower soil horizons occurs, suggests that the Muga watershed is likely to expe-
causing the reduction of infiltration capacity and rience an increase in surface runoff in the future due
more runoff (Jin et al., 2008). In study watershed, a to the effects of climate change. Increased surface
rapid change in LULC was observed between 1994 runoff can have implications for water availability,
and 2022 due to different proximate and underlying streamflow, erosion, and other hydrological processes
drivers. As a result, the scarcity of land and the need within the watershed. These changes could impact
to cultivate more land encouraged the community to water resources management, flood risk, and ecosys-
cultivate on steep slopes. Cropland of steep slopes tem dynamics in the region.
increases, in turn, increases surface runoff and reduces An increase in surface runoff can contribute to
infiltration, affecting water resource availability. erosion and sedimentation processes within a water-
The surface runoff trend and water yield were shed. The transport capacity of the surface runoff,
inconsistent during the study periods. However, which depends on factors such as flow velocity and
these showed an increasing trend in 2008, 2022, and volume, plays a crucial role in determining the
2038 compared to 1994. On the other hand, a decline amount of sediment that can be transported by the
in mean annual streamflow, evapotranspiration, and runoff. The study conducted by Abebe & Gebremariam
lateral flow was observed compared to the reference (Abebe & Gebremariam, 2019), reported that a reduc-
year. According to the study, forest cover had tion in surface runoff can lead to a significant
declined historically. As a result, compared to 1994, decrease in sediment yield. This finding suggests that
forest loss will accelerate surface runoff and water when surface runoff is reduced, there is less water
yield while decreasing evapotranspiration in the available to transport sediments, resulting in lower
Muga watershed in 2008, 2022, and 2038. The find- sediment yields.
ings of this study were similar to those of previous The findings of this study indicate that average
studies [e.g. (Chemura et al., 2020; Ridwansyah et al., annual rainfall is projected to increase in the 2055s,
2020)], which showed the shift of high evapotranspi- leading to an increase in streamflow discharge.
ration LULC to low evapotranspiration leads to an Additionally, it is expected that future temperatures
increment of surface runoff. The reduction in evapo- will also rise under the RCP4.5 and RCP8.5 emission
transpiration in 2008, 2022, and 2038 compared to scenarios. This increase in temperature tends to result
1994 may be due to the loss of vegetation cover in higher evapotranspiration (ET) rates when simu-
(forests, grasses, and shrublands). As a result, surface lated with LULC data from 2022 and under the
runoff and water yields are supported. RCP4.5 and RCP8.5 scenarios, compared to historical
climate datasets. Specifically, the study reveals that
ET is expected to rise by 10.4% (61.0 mm) under the
3.4.2. Individual impacts of climate change simulation using LULC data from 2022 and the
The water balance components, simulated by the RCP4.5 scenario. Similarly, under the simulation with
SWAT model for the baseline period (1994–2022), LULC data from 2022 and the RCP8.5 scenario, ET is
were compared with the results for the subsequent projected to increase by 11.3% (66.3 mm) compared
period (2023–2055) under RCP4.5 and RCP8.5. This to historical climate datasets. These results indicate a
comparison aimed to investigate the potential distinct increase in the predicted ET rates when con-
impacts of climate change on the hydrological com- sidering the climate simulation scenarios. The combi-
ponents of the study watershed. The findings indi- nation of increased rainfall and higher temperatures
cate anticipated changes in water balance can lead to enhanced evapotranspiration, suggesting
components, particularly surface runoff, for Muga an intensification of water loss from the study area.
watershed in the future (2023–2055) compared to Under both emission scenarios (RCP 4.5 and 8.5),
historical climate datasets (1994–2022) under the the intensity of climate change impacts on various
RCP4.5 and RCP8.5 scenarios. In the baseline period, hydrological components will increase in the future
surface runoff was approximately 217.4 mm. However, under both the RCP4.5 and RCP8.5 emission
 Cogent Food & Agriculture 15

scenarios within the study watershed. The projected Compared to the LULC 2022 & CC 1994–2022 sim-
increase in precipitation and temperature in the ulation, the combined impact of LULC and climate
2055s is considered as the primary driver of these changes is projected to cause the mean annual ET of
changes. The study highlights that the magnitude of the study area to increase. The finding of this study
change in most hydrological components is more indicated that the smallest and highest increase in
significant under RCP8.5 simulation scenario com- ET is expected to be under simulation LULC 2038
pared to the RCP4.5 scenario. For instance, the aver- and RCP 4.5 scenario (10.6%) and LULC 2038 and
age annual streamflow is projected to increase by RCP 8.5 scenario (11.3%), respectively.
13.8% under RCP4.5 and by 21.3% under RCP8.5 The comparison of the combined impact of future
compared to the historical baseline period. Moreover, LULC and climate change on ET with the individual
the study indicates that groundwater recharge, sur- effects of LULC change scenarios and climate change
face runoff, lateral flow, and water yield are also scenarios reveals that temperature and rainfall have
expected to undergo notable changes. a greater influence on future ET compared to LULC
Under the RCP4.5 scenario with the 2022 LULC changes. Additionally, the direction of change in
map, the average changes in groundwater recharge, average annual ET under the combined LULC and cli-
surface runoff, lateral flow, and water yield are pro- mate change scenario aligns with the changes
jected to be 20.1, 5.9, 10.4, and 11.6%, respectively, observed when LULC and climate change scenarios
compared to the reference period of 1994–2022. are considered individually. This implies that the
Similarly, under the RCP8.5 scenario with the same combined effect of LULC and climate change on ET
LULC map, the corresponding changes are estimated does not lead to significant deviations from the indi-
to be 21.4, 6.5, 11.3, and 12.9%. These findings align vidual impacts.
with the study conducted by Dile (Dile et al., 2013), According to the results of this study, the com-
which suggests that climate change can impact hydro- bined effect of changes in LULC and climate is gen-
logical components and alter the behavior of the erally similar to the distinct impacts of LULC change
hydrological cycle. The consistent results between and climate change on evapotranspiration (ET). The
these studies strengthen the understanding of the impacts of these drivers indicate the same direction
potential effects of climate change on the hydrological in terms of their influence on ET. However, it is noted
system, emphasizing the need for adaptive measures that in future climate change scenarios, ET is slightly
to manage water resources effectively in response to higher compared to simulations under the combined
projected changes. Table 11, indicates that the hydro- impacts of LULC and climate change. This implies
logical components show an increasing trend, and this that climate change alone, represented by the future
increment is associated with a significant rise in rain- climate scenarios, has a slightly stronger effect on
fall for both the RCP4.5 and RCP8.5 scenarios. increasing ET than when LULC changes and climate
change are considered together. Nevertheless, it is
highlighted that both LULC changes and climate
3.4.3. Combined impacts of land use land cover change scenarios lead to a significant increase in ET.
and climate changes The individual effects of LULC change and climate
According to the findings presented in Table 11, change, as well as their combined impacts, are
both LULC changes and climate change have notice- expected to have nearly equal influences on ET com-
able effects on the hydrological components of the pared to the simulation under the LULC 2022 and
study area. However, the individual effect of climate climate data from 1994 to 2022.
played a more significant role than LULC. This sug- Table 11 shows that the average annual surface
gests that changes in climate variables such as runoff under simulation LULC 2038 and RCP 8.5 sce-
precipitation patterns, temperature, and evapotrans- nario is likely to increase compared to simulation
piration have a stronger influence on the water bal- under LULC 2022 and CC 1994–2022; it is anticipated
ance components than modifications in LULC. to increase by 14.2%. Surface runoff is expected to
Climate change can directly alter the overall water increase by 10.4 and 9.8% under simulations LULC
availability, streamflow, and evapotranspiration rates 2022 and RCP 8.5 and LULC 2038 and RCP 4.5 scenar-
within the watershed. While LULC changes can affect ios, respectively. Although surface runoff is anticipated
local hydrological processes, such as surface runoff to increase in the future, this may happen mainly due
and infiltration, their influence seems to be compar- to future climate rather than future LULC. The finding
atively smaller when compared to the impact of cli- showed that future LULC changes would increase sur-
mate change face runoff and water yield, followed by less
16 T. BELAY AND D. A. MENGISTU

vegetation cover and settlement expansion. Hence, flow increase as the infiltration capacity is
the increment of surface runoff and water yield in the exceeded, resulting in higher water yields.
study watershed require special attention, as they Additionally, the study evaluates the response of
may cause massive erosion and sedimentation. lateral flow to the combined future land use and cli-
The results indicate that the combined effects of mate change scenario. The simulation results indi-
LULC changes and climate change are anticipated to cate that under the LULC 2038 and RCP 4.5 scenario,
increase groundwater flow in the study area. the average lateral flow is expected to increase by
Specifically, under the LULC 2038 and RCP 4.5 sce- 3.5% compared to the simulation under the LULC
nario, groundwater flow is projected to increase by 2022 and climate data from 1994 to 2022. Similarly,
20.8% compared to simulations under LULC 2022 under the LULC 2038 and RCP 8.5 scenario, the aver-
and climate data from 1994 to 2023. Similarly, under age lateral flow is projected to increase by 9.5%.
the LULC 2038 and RCP 8.5 scenario, groundwater These findings suggest that both land use changes
flow is projected to increase by 22.2% compared to and climate change influence lateral flow, with the
the same reference simulation. Furthermore, the combined effects resulting in increased lateral flow
study indicates that climate change has a higher compared to the reference simulation.
impact on groundwater flow than the LULC change The study compared streamflow values under
scenario alone. This implies that the changes in cli- future climate and land-use scenarios with simula-
mate variables, such as temperature and precipita- tions based on LULC 2022 and historical climate
tion, have a more significant influence on groundwater data (CC 1994–2022). The results demonstrate that
flow compared to modifications in LULC. the combination of future LULC and climate change
Although the study finds a slight effect of LULC scenarios has a significant impact on streamflow
alone on groundwater flow, with increased ground- discharge compared to the simulation based on
water flow observed in certain simulations, the mag- LULC 2022 and historical climate datasets. It indi-
nitude of groundwater flow change under different cates that streamflow is likely to be highest under
LULC scenarios is reported to be lower than the simulation of LULC 2038 and RCP 8.5 com-
expected. This suggests that LULC changes have a pared to the simulation of LULC 2022 and CC
relatively smaller impact on groundwater flow com- 1994–2022.
pared to climate change. The findings of other stud- The increase in future rainfall, particularly under
ies (Marhaento et al., 2018; Pan et al., 2017; Yan RCP 8.5, is reported to be approximately 20.2%, lead-
et al., 2016) align with the findings of this study, ing to a corresponding increase of 40.6 mm (21.0%)
indicating that the impact of climate change on in streamflow discharge. Similarly, under the LULC
groundwater flow is expected to be higher than that 2038 and RCP 4.5 scenario, which showed a 17.2%
of LULC change. increase in precipitation compared to the historical
The findings of this study show that the com- period, streamflow is expected to increase by
bined effects of climate change and LULC change 36.4 mm (18.8%). These changes in streamflow are
show an increase in water yields in the future com- approximately 4.4 and 6.3% higher than the impact
pared to simulation under LULC 2022 and CC of the LULC scenario alone. The study concludes that
1994–2022. The study highlights that there are dif- the combined effect of future LULC and climate
ferences among the scenarios in terms of varia- change has a greater influence on streamflow dis-
tions in mean annual total water yield. The impact charge than the LULC scenario alone. The intensifica-
of climate change is particularly pronounced, with tion of rainfall projected under RCP 4.5 and RCP 8.5
the average yearly water yield expected to increase contributes to this increase in streamflow. However,
by approximately 13.8% under the LULC 2022 and the study also emphasizes that the contribution of
RCP 4.5 scenario and 17.2% under the LULC 2022 LULC change to streamflow change is still
and RCP 8.5 scenario. These increases are about 2.0 substantial.
and 5.0% higher, respectively, compared to the Overall, the projected rise in future rainfall due to
impact of the LULC change scenario alone. The rea- climate change scenarios is expected to enhance
son for this difference can be attributed to the average annual streamflow discharge and water
increased precipitation projected under the RCP resource availability in the Muga watershed. This has
4.5 and RCP 8.5 climate change scenarios. The implications for water resources planning and opera-
excess rainfall leads to an abundance of water on tion in the study area and downstream areas of the
the soil surface, reducing the likelihood of water Abay basin of Ethiopia, providing benefits for local
infiltration. Consequently, surface runoff and base and regional water resources infrastructure.
 Cogent Food & Agriculture 17

3.5. Limitations study watershed than simulations-based LULC 2022

and historical climate datasets. Notably, the simula-
The resolution of the digital elevation model can sig-
tion under LULC 2038 and RCP 8.5 exhibits the
nificantly influence hydrological components, soil
highest impact.
parameters, LULC, and hydrologic response unit
The results of our investigation suggest a signifi-
delineation. Higher-resolution DEMs provide more
cant increase in streamflow under the combined
detailed information and potentially improve model
scenarios of LULC 2038 and RCP 8.5, as well as LULC
accuracy. Furthermore, changes in future LULC and
2038 and RCP 4.5 when compared to the baseline
climate conditions may affect model parameters,
period. Specifically, the increase in streamflow is
which could introduce uncertainties that could affect
40.6 mm (21.0%) and 36.4 mm (18.8%) for these
the accuracy of the model results.
respective scenarios. These alterations in streamflow
In this study, model parameters were calibrated
discharge demonstrate the influence of both LULC
and bias-corrected to simulate the future water bal-
and climate change on the hydrological compo-
ance components of the Muga watershed. However,
nents of the watershed. Furthermore, the study
it is acknowledged that future changes in LULC and
emphasizes that the increase in streamflow result-
climate might alter these parameters (Glavan &
ing from the combined impact of LULC and climate
Pintar, 2012). Thus, it is essential to recognize exist-
change exceeds the impact of LULC change alone.
ing limitations and consider potential implications
This indicates that climate change, characterized by
for the accuracy of the model results.
intensified rainfall under the RCP 4.5 and RCP 8.5
Despite these limitations and uncertainties, the
scenarios, plays a significant role in the projected
findings of this study are considered relevant, reli-
increase in streamflow. However, the contribution of
able, and valid within the current contexts of LULC
LULC change to the changes in streamflow remains
and climate change, following the standards outlined
substantial.
by Moriasi et al. (Moriasi et al., 2007). While it may
Certainly, the results of this research provide valu-
be challenging to precisely compare the effects of
able insights for the identification and implementa-
LULC and climate change, this study successfully
tion of effective climate adaptation strategies and
identifies and assesses the individual and combined
soil and water conservation methods within the
impacts of these factors on watershed hydrology. It
study watershed. A comprehensive understanding of
is crucial to approach the results of any study with
the individual and combined impacts of LULC and
caution and interpret them within the context of its
climate change on hydrology empowers policymak-
limitations and uncertainties. Ongoing research and
ers, land managers, and other stakeholders to make
advancements in modeling techniques have the
well-informed decisions regarding the management
potential to further enhance the accuracy and reli-
and sustainable use of water resources.
ability of future hydrological assessments.
The study can also contribute valuable insights into
the planning and operation of local and regional
4. Conclusions water resource infrastructure within the study water-
shed. Decision-makers can design and optimize water
The assessment of the individual and combined management systems by taking into account the pro-
impacts of LULC changes and climate change on jected changes in hydrological components. Overall,
watershed hydrology is crucial for the effective this study enhances understanding of the potential
management of water resources and the develop- impacts of LULC and climate change on hydrology,
ment of climate adaptation strategies. The results of establishing a foundation for decision-making pro-
this study suggest that LULC change significantly cesses related to water resource management, climate
affects streamflow, surface runoff, groundwater flow, adaptation, and soil and water conservation in the
and lateral flow, with relatively minor effects on
study watershed.
evapotranspiration and water yields. Moreover, the
study predicts an increase in surface runoff, ground-
water flow, streamflow, lateral flow, and water yield Acknowledgments
in the Muga watershed by the 2055s, under both
The National Meteorological Agency of Ethiopia and the
RCP 4.5 and RCP 8.5 scenarios, when compared to
Ethiopian Ministry of Water, Irrigation, and Electricity
reference climate datasets. It emphasizes that the deserve gratitude for providing the meteorological and
combined effects of future LULC and climate streamflow data, respectively. The Geological Survey of the
changes have a greater impact on hydrology in the United States (USGS) was very helpful.
18 T. BELAY AND D. A. MENGISTU

Disclosure statement Ethiopia. SN Applied Sciences, 1(5), 446. https://doi.

org/10.1007/s42452-019-0347-1
No potential conflict of interest was reported by the author(s). Afzal, M., & Ragab, R. (2020). Impact of the future climate
and land use changes on the hydrology and water
resources in South East England, UK. American Journal of
Funding Water Resources, 20, 218–231. https://doi.org/10.12691/
ajwr-8-5-2
The study was made possible due to the financial support
Alemseged, T. H., & Tom, R. (2015). Evaluation of regional
provided to the first author by the International Foundation
climate model simulations of rainfall over the Upper
for Science (IFS), Sweden [Grant No. W_6250-1,
Blue Nile basin. Atmospheric Research, 161–162, 57–64.
January 2019).
https://doi.org/10.1016/j.atmosres.2015.03.013
Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop
evapotranspiration-guidelines for computing crop water
About the authors requirements-FAO irrigation and drainage paper 56 (vol.
Tatek Belay (PhD), is an Assistant Professor 300, p. D05109). FAO.
in the Department of Geography and Arnold, J. G., Kiniry, J. R., Srinivasan, R., Williams, J. R.,
Environmental Studies, Debre Tabor Haney, E. B., & Neitsch, S. L. (2012). Soil and water
University, Ethiopia. His area of interests are assessment tool input/output file documentation: version
modeling soil erosion, watershed manage- 2012 (vol. 436). Texas Water Resources Institute.
ment, land use change, GIS and Remote Arsanjani, J. J., Kainz, W., & Mousivand, A. J. (2011).
sensing, climate change, and hydrology. Tracking dynamic land-use change using spatially
explicit Markov Chain based on cellular automata: The
Daniel Ayalew Mengistu (PhD), is an Associate Professor in case of Tehran. International Journal of Image and Data
the Department of Geography and Environmental Studies, Fusion, 2(4), 329–345. https://doi.org/10.1080/19479832
Bahir Dar University, Ethiopia. He has been teaching for .2011.605397
several years in the fields of Geoinformatics, Climatology, Ayele, H. S., Li, M.-H., & Tung, C.-P. (2016). Assessing climate
and Hydrology. He has published numerous articles on change impact on Gilgel Abbay and Gumara watershed
topics related to climate change, hydrology, and soil and hydrology, the Upper Blue Nile Basin, Ethiopia. Terrestrial,
water conservation. Atmospheric & Oceanic Sciences, 27,1005–1018. https://
doi.org/10.3319/TAO.2016.07.30.01
Ayivi, F., & Jha, M. K. (2018). Estimation of water balance
ORCID and water yield in the Reedy Fork-Buffalo Creek water-
shed in North Carolina using SWAT. International Soil and
Tatek Belay http://orcid.org/0000-0003-0039-0240
Water Conservation Research, 6(3), 203–213. https://doi.
org/10.1016/j.iswcr.2018.03.007
Aylward, B. (2005). Land use, hydrological function and
Data availability statement economic valuation, forests, water and people in the
Data is available based on reasonable request from the humid tropics: Past, present and future hydrological
authors. research for integrated land and water management
. International Hydrology Series. Cambridge University
Press, pp. 99–120.
References BCEOM. (1998). Abbay river basin integrated development
master plan project, phase 2, data collection and
Abbaspour, K., Vaghefi, S. A., & Srinivasan, R. (2018). A site-investigation survey and analysis, section II, sectoral
guideline for successful calibration and uncertainty analysis studies.
for soil and water assessment: A review of papers from the Belay, T., & Mengistu, D. A. (2019). Land use and land cover
2016 International SWAT Conference. Multidisciplinary dynamics and drivers in the Muga watershed, Upper
Digital Publishing Institute. Blue Nile basin, Ethiopia. Remote Sensing Applications:
Abdo, K. S., Fiseha, B. M., Rientjes, T. H. M., Gieske, A. S. M., Society and Environment, 15, 100249. https://doi.
& Haile, A. T. (2009). Assessment of climate change org/10.1016/j.rsase.2019.100249
impacts on the hydrology of Gilgel Abay catchment in Belay, T., & Mengistu, D. A. (2021). Impacts of land use/land
Lake Tana Basin, Ethiopia. Hydrological Processes, 23(26), cover and climate changes on soil erosion in Muga water-
1–22. https://doi.org/10.1002/hyp.7363 shed, Upper Blue Nile basin (Abay), Ethiopia. Ecological
Abdulahi, M. M., Hashim, H., & Teha, M. (2016). Rangeland Processes, 10(1), 1–23. https://doi.org/10.1186/s13717-021-
degradation: Extent, impacts, and alternative restoration 00339-9
techniques in the rangelands of Ethiopia. Tropical and Belihu, M., Tekleab, S., Abate, B., & Bewket, W. (2020).
Subtropical Agroecosystems, 19(3), 305–3018. https://doi. Hydrologic response to land use land cover change in
org/10.56369/tsaes.2234 the upper Gidabo watershed, Rift Valley Lakes basin,
Abebe, G. (2017). Long-term climate data description in Ethiopia. HydroResearch, 3, 85–94. https://doi.
Ethiopia. Data in Brief, 14, 371–392. https://doi. org/10.1016/j.hydres.2020.07.001
org/10.1016/j.dib.2017.07.052 Benegas, L., Ilstedt, U., Roupsard, O., Jones, J., & Malmer, A.
Abebe, T., & Gebremariam, B. (2019). Modeling runoff and (2014). Effects of trees on infiltrability and preferential
sediment yield of Kesem dam watershed, Awash Basin, flow in two contrasting agroecosystems in Central
 Cogent Food & Agriculture 19

America. Agriculture, Ecosystems & Environment, 183, 185– Sensing Applications: Society and Environment, 8, 224–
196. https://doi.org/10.1016/j.agee.2013.10.027 230. https://doi.org/10.1016/j.rsase.2017.10.001
Berhane, A., Hadgu, G., Worku, W., & Abrha, B. (2020). Dile, Y. T., Berndtsson, R., & Setegn, S. (2013). Hydrological
Trends in extreme temperature and rainfall indices in response to climate change for Gilgel Abay River, in the
the semi-arid areas of Western Tigray, Ethiopia. Lake Tana basin-upper Blue Nile basin of Ethiopia. PLOS
Environmental Systems Research, 9(1), 1–20. https://doi. One. 8(10), e79296. https://doi.org/10.1371/journal.
org/10.1186/s40068-020-00165-6 pone.0079296
Berihun, M. L., Tsunekawa, A., Haregeweyn, N., Meshesha, Fan, M., & Shibata, H. (2015). Simulation of watershed hy-
D. T., Adgo, E., Tsubo, M., Masunaga, T., Fenta, A. A., drology and stream water quality under land use and
Sultan, D., Yibeltal, M., & Ebabu, K. (2019). Hydrological climate change scenarios in Teshio River watershed,
responses to land use/land cover change and climate northern Japan. Ecological Indicators, 50, 79–89. https://
variability in contrasting agro-ecological environments doi.org/10.1016/j.ecolind.2014.11.003
of the Upper Blue Nile basin, Ethiopia. The Science of the Ficklin, D. L., Luo, Y., Luedeling, E., & Zhang, M. (2009).
Total Environment, 689, 347–365. https://doi.org/10.1016/j. Climate change sensitivity assessment of a highly agri-
scitotenv.2019.06.338 cultural watershed using SWAT. Journal of Hydrology,
Betrie, G. D., Mohamed, Y. A., van Griensven, A., & Srinivasan, 374(1–2), 16–29. https://doi.org/10.1016/j.jhydrol.2009.
R. (2011). Sediment management modelling in the Blue 05.016
Nile Basin using SWAT model. Hydrology and Earth Foody, G. M., & Mathur, A. (2004). Toward intelligent train-
System Sciences, 15(3), 807–818. https://doi.org/10.5194/ ing of supervised image classifications: Directing train-
hess-15-807-2011 ing data acquisition for SVM classification. Remote
Bewket, W., & Sterk, G. (2005). Dynamics in land cover and Sensing of Environment, 93(1–2), 107–117. https://doi.
its effect on stream flow in the Chemoga watershed, org/10.1016/j.rse.2004.06.017
Blue Nile basin, Ethiopia. Hydrological Processes, 19(2), Gashaw, T., Tulu, T., Argaw, M., & Worqlul, A. W. (2018).
445–458. https://doi.org/10.1002/hyp.5542 Modeling the hydrological impacts of land use/land cov-
Bewket, W., & Teferi, E. (2009). Assessment of soil erosion er changes in the Andassa watershed, Blue Nile Basin,
hazard and prioritization for treatment at the watershed Ethiopia. The Science of the Total Environment, 619-620,
level: Case study in the Chemoga watershed, Blue Nile 1394–1408. https://doi.org/10.1016/j.scitotenv.2017.11.191
basin, Ethiopia. Land Degradation & Development, 20(6), Gebre, S. L., & Ludwig, F. (2015). Hydrological response to
609–622. https://doi.org/10.1002/ldr.944 climate change of the upper Blue Nile River Basin: Based
Birhanu, A., Masih, I., van der Zaag, P., Nyssen, J., & Cai, X. on IPCC fifth assessment report (AR5). Journal of
(2019). Impacts of land use and land cover changes on Climatology & Weather Forecasting, 3, 1–15.
hydrology of the Gumara catchment, Ethiopia. Physics Gebremicael, T. G., Mohamed, Y. A., Betrie, G. D., Van der
and Chemistry of the Earth, Parts A/B/C, 112, 165–174. Zaag, P., & Teferi, E. (2013). Trend analysis of runoff and
https://doi.org/10.1016/j.pce.2019.01.006 sediment fluxes in the Upper Blue Nile basin: A com-
Breiman, L. (2001). Random forests. Machine Learning, 45(1), bined analysis of statistical tests, physically-based mod-
5–32. https://doi.org/10.1023/A:1010933404324 els and landuse maps. Journal of Hydrology, 482, 57–68.
Chanapathi, T., & Thatikonda, S. (2020). Investigating the https://doi.org/10.1016/j.jhydrol.2012.12.023
impact of climate and land-use land cover changes on Gessesse, B., Bewket, W., & Bräuning, A. (2015). Model-based
hydrological predictions over the Krishna river basin un- characterization and monitoring of runoff and soil ero-
der present and future scenarios. The Science of the Total sion in response to land use/land cover changes in the
Environment, 721, 137736. https://doi.org/10.1016/j.scito- Modjo watershed, Ethiopia. Land Degradation &
tenv.2020.137736 Development, 26(7), 711–724. https://doi.org/10.1002/
Chemura, A., Rwasoka, D., Mutanga, O., Dube, T., & Mushore, ldr.2276
T. (2020). The impact of land-use/land cover changes on Gidey, E., Dikinya, O., Sebego, R., Segosebe, E., & Zenebe, A.
water balance of the heterogeneous Buzi sub-catchment, (2017). Cellular automata and Markov Chain (CA_Markov)
Zimbabwe. Remote Sensing Applications: Society and model-based predictions of future land use and land
Environment, 18, 100292. https://doi.org/10.1016/j.rsase. cover scenarios (2015–2033) in Raya, northern Ethiopia.
2020.100292 Modeling Earth Systems and Environment, 3(4), 1245–
Congalton, R. G., & Green, K. (2019). Assessing the accuracy 1262. https://doi.org/10.1007/s40808-017-0397-6
of remotely sensed data: principles and practices. CRC Gizaw, M. S., Biftu, G. F., Gan, T. Y., Moges, S. A., & Koivusalo,
press. H. (2017). Potential impact of climate change on stream-
Cousino, L. K., Becker, R. H., & Zmijewski, K. A. (2015). flow of major Ethiopian rivers. Climatic Change, 143(3–4),
Modeling the effects of climate change on water, sedi- 371–383. https://doi.org/10.1007/s10584-017-2021-1
ment, and nutrient yields from the Maumee River water- Glavan, M., & Pintar, M. (2012). Strengths, weaknesses, op-
shed. Journal of Hydrology: Regional Studies, 4, 762–775. portunities and threats of catchment modelling with Soil
https://doi.org/10.1016/j.ejrh.2015.06.017 and Water Assessment Tool (SWAT) model. Water
Demessie, E. T. (2015). Soil hydrological impacts and climatic Resources Management and Modeling, 39–64.
controls of land use and land cover changes in the Upper Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau,
Blue Nile (Abay) basin. CRC Press. D., & Moore, R. (2017). Google Earth Engine:
Demissie, F., Yeshitila, K., Kindu, M., & Schneider, T. (2017). Planetary-scale geospatial analysis for everyone. Remote
Land use/Land cover changes and their causes in Sensing of Environment, 202, 18–27. https://doi.
Libokemkem District of South Gonder, Ethiopia. Remote org/10.1016/j.rse.2017.06.031
20 T. BELAY AND D. A. MENGISTU

Halmy, M. W. A., Gessler, P. E., Hicke, J. A., & Salem, B. B. Sediments, 18(4), 1743–1755. https://doi.org/10.1007/
(2015). Land use/land cover change detection and pre- s11368-017-1901-3
diction in the north-western coastal desert of Egypt us- Mengistu, D., Bewket, W., Dosio, A., & Panitz, H.-J. (2021).
ing Markov-CA. Applied Geography, 63, 101–112. https:// Climate change impacts on water resources in the Upper
doi.org/10.1016/j.apgeog.2015.06.015 Blue Nile (Abay) River Basin, Ethiopia. Journal of
Hishe, S., Bewket, W., Nyssen, J., & Lyimo, J. (2020). Hydrology, 592, 125614. https://doi.org/10.1016/j.jhy-
Analysing past land use land cover change and drol.2020.125614
CA-Markov-based future modelling in the Middle Suluh Mondal, M., Sharma, N., Garg, P. K., & Kappas, M. (2016).
Valley, Northern Ethiopia. Geocarto International, 35(3), Statistical independence test and validation of CA
225–255. https://doi.org/10.1080/10106049.2018.1516241 Markov land use land cover (LULC) prediction results.
Hu, Y., & Hu, Y. (2019). Land cover changes and their driv- The Egyptian Journal of Remote Sensing and Space Science,
ing mechanisms in Central Asia from 2001 to 2017 sup- 19(2), 259–272. https://doi.org/10.1016/j.ejrs.2016.08.001
ported by Google Earth Engine. Remote Sensing, 11(5), Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner, R. L.,
554. https://doi.org/10.3390/rs11050554 Harmel, R. D., & Veith, T. L. (2007). Model evaluation guide-
Hurni, H. (1998). Agroecologial belts of Ethiopia: Explanatory lines for systematic quantification of accuracy in water-
notes on three maps at a scale of 1:1,000,000. Soil conser- shed simulations. Transactions of the ASABE, 50, 885–900.
vation research program of Ethiopia. Addis Abeba. Nkonya, E., Mirzabaev, A., & Braun, J. V. (2016). Economics
Jin, K., Cornelis, W. M., Gabriels, D., Schiettecatte, W., De of land degradation and improvement–a global assess-
Neve, S., Lu, J., Buysse, T., Wu, H., Cai, D., Jin, J., & ment for sustainable development. Springer Nature.
Harmann, R. (2008). Soil management effects on runoff NMA. (2007). Climate change national adaptation pro-
and soil loss from field rainfall simulation. CATENA, 75(2), gramme of action (NAPA) of Ethiopia., Addis Ababa.
191–199. https://doi.org/10.1016/j.catena.2008.06.002 Pan, S., Liu, D., Wang, Z., Zhao, Q., Zou, H., Hou, Y., Liu, P.,
Jury, M. R., & Funk, C. (2013). Climatic trends over Ethiopia: & Xiong, L. (2017). Runoff Responses to Climate and
regional signals and drivers. International Journal of Land Use/Cover Changes under Future Scenarios. Water,
Climatology, 33(8), 1924–1935. https://doi.org/10.1002/ 9(7), 475. https://doi.org/10.3390/w9070475
joc.3560 Perry, M., & Hollis, D. (2005). The development of a new set
Kim, U., & Kaluarachchi, J. J. (2009). Climate change im- of long-term climate averages for the UK. International
pacts on water resources in the Upper Blue Nile River Journal of Climatology, 25(8), 1023–1039. https://doi.
Basin, Ethiopia. Journal of the American Water Resources org/10.1002/joc.1160
Association, 45(6), 1361–1378. https://doi.org/10.1111/j. Pontius, R. G. (2002). Statistical methods to partition effects
1752-1688.2009.00369.x of quantity and location during comparison of categori-
Kiros, G., Shetty, A., & Nandagiri, L. (2016). Analysis of vari- cal maps at multiple resolutions. Photogrammetric
ability and trends in rainfall over northern Ethiopia. Engineering and Remote Sensing, 68, 1041.
Arabian Journal of Geosciences, 9(6), 451. https://doi. Richards, J. A., & Richards, J. A. (1999). Remote sensing digi-
org/10.1007/s12517-016-2471-1 tal image analysis. Springer.
Kurukulasuriya, P., & Rosenthal, S. (2013). Climate change Ridwansyah, I., Yulianti, M., Onodera, S.-I., Shimizu, Y., Wibowo,
and agriculture : A review of impacts and adaptations. H., Fakhrudin, M., & Apip, A. (2020). The impact of land use
World Bank. and climate change on surface runoff and groundwater in
Luo, J., Liu, Y., Zhang, s., & Liang, J. (2021). Extreme random Cimanuk watershed, Indonesia. Limnology, 21(3), 487–498.
forest method for machine fault classification. Measurement https://doi.org/10.1007/s10201-020-00629-9
Science and Technology, 32(11), 114006. https://doi. Rogger, M., Agnoletti, M., Alaoui, A., Bathurst, J. C., Bodner,
org/10.1088/1361-6501/ac14f5 G., Borga, M., Chaplot, V., Gallart, F., Glatzel, G., Hall, J.,
Magidi, J., Nhamo, L., Mpandeli, S., & Mabhaudhi, T. (2021). Holden, J., Holko, L., Horn, R., Kiss, A., Kohnová, S.,
Application of the random forest classifier to map irri- Leitinger, G., Lennartz, B., Parajka, J., Perdigão, R., …
gated areas using Google Earth Engine. Remote Sensing, Blöschl, G. (2017). Land use change impacts on floods at
13(5), 876. https://doi.org/10.3390/rs13050876 the catchment scale: Challenges and opportunities for
Marhaento, H., Booij, M. J., & Hoekstra, A. Y. (2018). future research. Water Resources Research, 53(7), 5209–
Hydrological response to future land-use change and cli- 5219. https://doi.org/10.1002/2017WR020723
mate change in a tropical catchment. Hydrological Safriel, U. N. (2007). The Assessment of Global Trends in
Sciences Journal, 63(9), 1368–1385. https://doi.org/10.108 Land Degradation. In M. V. K. Sivakumar, N. Ndiang’ui
0/02626667.2018.1511054 (Eds.), Climate and Land Degradation. Environmental
Mekonnen, D. F., Duan, Z., Rientjes, T., & Disse, M. (2018). Science and Engineering. Springer, Berlin, Heidelberg.
Analysis of combined and isolated effects of land-use https://doi.org/10.1007/978-3-540-72438-4_1
and land-cover changes and climate change on the up- Santos, D., Cardoso-Fernandes, J., Lima, A., Müller, A.,
per Blue Nile River basin’s streamflow. Hydrology and Brönner, M., & Teodoro, A. C. (2022). Spectral analysis to
Earth System Sciences, 22(12), 6187–6207. https://doi. improve inputs to random forest and other boosted en-
org/10.5194/hess-22-6187-2018 semble tree-based algorithms for detecting NYF
Melaku, N. D., Renschler, C. S., Holzmann, H., Strohmeier, S., Pegmatites in Tysfjord, Norway. Remote Sensing, 14(15),
Bayu, W., Zucca, C., Ziadat, F., & Klik, A. (2018). Prediction 3532. https://doi.org/10.3390/rs14153532
of soil and water conservation structure impacts on run- Schneiderman, E. M., Steenhuis, T. S., Thongs, D. J., Easton,
off and erosion processes using SWAT model in the Z. M., Zion, M. S., Neal, A. L., Mendoza, G. F., & Todd
northern Ethiopian highlands. Journal of Soils and Walter, M. (2007). Incorporating variable source area hy-
 Cogent Food & Agriculture 21

drology into a curve-number-based watershed model. Ecohydrology & Hydrobiology, 21(2), 315–332. https://doi.
Hydrological Processes, 21(25), 3420–3430. https://doi. org/10.1016/j.ecohyd.2020.12.001
org/10.1002/hyp.6556 Tekleab, S., Mohamed, Y., & Uhlenbrook, S. (2013).
SCS. (1972). National engineering handbook: Section 4: Hydro-climatic trends in the Abay/Upper Blue Nile basin,
Hydrology. USDA Washington. Ethiopia. Physics and Chemistry of the Earth, Parts A/B/C,
Setegn, S., Melesse, A. M., Rayner, D., & Dargahi, B. (2014). 61–62, 32–42. https://doi.org/10.1016/j.pce.2013.04.017
Climate change impact on water resources and adaptation Tekleab, S., Mohamed, Y., Uhlenbrook, S., & Wenninger, J.
strategies in the Blue Nile River Basin (pp. 389–404). Nile (2014). Hydrologic responses to land cover change: The
River Basin, Springer. case of Jedeb mesoscale catchment, Abay/Upper Blue
Sharpley, A. N., & Williams, J. R. (1990). EPIC-erosion/productiv- Nile basin, Ethiopia. Hydrological Processes, 28(20), 5149–
ity impact calculator. I: Model documentation. II: User manual. 5161. https://doi.org/10.1002/hyp.9998
Technical Bulletin-United States Department of Agriculture. Teklesadik, A. D., Alemayehu, T., Van Griensven, A., Kumar,
Shawul, A. A., Chakma, S., & Melesse, A. M. (2019). The re- R., Liersch, S., Eisner, S., Tecklenburg, J., Ewunte, S., &
sponse of water balance components to land cover Wang, X. (2017). Inter-model comparison of hydrological
change based on hydrologic modeling and partial least impacts of climate change on the Upper Blue Nile basin
squares regression (PLSR) analysis in the Upper Awash using ensemble of hydrological models and global cli-
Basin. Journal of Hydrology: Regional Studies, 26, 100640. mate models. Climatic Change, 141(3), 517–532. https://
https://doi.org/10.1016/j.ejrh.2019.100640 doi.org/10.1007/s10584-017-1913-4
Shrestha, B., Cochrane, T. A., Caruso, B. S., & Arias, M. E. Teutschbein, C., & Seibert, J. (2012). Bias correction of re-
(2018). Land use change uncertainty impacts on stream- gional climate model simulations for hydrological
flow and sediment projections in areas undergoing rap- climate-change impact studies: Review and evaluation
id development: A case study in the Mekong Basin. Land of different methods. Journal of Hydrology, 456–457, 12–
Degradation & Development, 29(3), 835–848. https://doi. 29. https://doi.org/10.1016/j.jhydrol.2012.05.052
org/10.1002/ldr.2831 van Vuuren, D. P., Edmonds, J., Kainuma, M., Riahi, K.,
Simane, B., Zaitchik, B. F., & Ozdogan, M. (2013). Thomson, A., Hibbard, K., Hurtt, G. C., Kram, T., Krey, V.,
Agroecosystem analysis of the Choke Mountain water- Lamarque, J.-F., Masui, T., Meinshausen, M., Nakicenovic,
sheds, Ethiopia. Sustainability, 5(2), 592–616. https://doi. N., Smith, S. J., & Rose, S. K. (2011). The representative
org/10.3390/su5020592 concentration pathways: An overview. Climatic Change,
Singh, S. K., Mustak, S., Srivastava, P. K., Szabó, S., & Islam, 109(1–2), 5–31. https://doi.org/10.1007/s10584-011-0148-z
T. (2015). Predicting spatial and decadal LULC changes Woldesenbet, T. A., Elagib, N. A., Ribbe, L., & Heinrich, J.
through cellular automata Markov chain models using (2017). Hydrological responses to land use/cover chang-
earth observation datasets and geo-information. es in the source region of the Upper Blue Nile Basin,
Environmental Processes, 2(1), 61–78. https://doi.org/10. Ethiopia. The Science of the Total Environment, 575, 724–
1007/s40710-015-0062-x 741. https://doi.org/10.1016/j.scitotenv.2016.09.124
Stocker, T., Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Woldesenbet, T. A., Elagib, N. A., Ribbe, L., & Heinrich, J.
Boschung, J., Nauels, A., Xia, Y., Bex, V., & Midgley, P. M. (2018). Catchment response to climate and land use
(2013). Climate change 2013: The physical science basis. changes in the Upper Blue Nile sub-basins, Ethiopia. The
Contribution of working group I to the fifth assessment re- Science of the Total Environment, 644, 193–206. https://
port of the intergovernmental panel on climate change. doi.org/10.1016/j.scitotenv.2018.06.198
Cambridge university press Cambridge. Worku, G., Teferi, E., Bantider, A., Dile, Y. T., & Taye, M. T.
Swain, S. S., Mishra, A., Sahoo, B., & Chatterjee, C. (2020). (2018). Evaluation of regional climate models perfor-
Water scarcity-risk assessment in data-scarce river basins mance in simulating rainfall climatology of Jemma
under decadal climate change using a hydrological sub-basin, Upper Blue Nile Basin, Ethiopia. Dynamics of
modelling approach. Journal of Hydrology, 590, 125260. Atmospheres and Oceans, 83, 53–63. https://doi.
https://doi.org/10.1016/j.jhydrol.2020.125260 org/10.1016/j.dynatmoce.2018.06.002
Talib, A., & Randhir, T. O. (2017). Climate change and land Worqlul, A. W., Ayana, E. K., Yen, H., Jeong, J., MacAlister, C.,
use impacts on hydrologic processes of watershed sys- Taylor, R., Gerik, T. J., & Steenhuis, T. S. (2018). Evaluating
tems. Journal of Water and Climate Change, 8(3), 363– hydrologic responses to soil characteristics using SWAT
374. https://doi.org/10.2166/wcc.2017.064 model in a paired-watersheds in the Upper Blue Nile
Taye, M. T., Willems, P., & Block, P. (2015). Implications of cli- Basin. Catena, 163, 332–341. https://doi.org/10.1016/j.cat-
mate change on hydrological extremes in the Blue Nile ena.2017.12.040
basin: A review. Journal of Hydrology: Regional Studies, 4, Xiong, J., Thenkabail, P. S., Tilton, J. C., Gumma, M. K.,
280–293. https://doi.org/10.1016/j.ejrh.2015.07.001 Teluguntla, P., Oliphant, A., Congalton, R. G., Yadav, K., &
Teferi, E., Bewket, W., Uhlenbrook, S., & Wenninger, J. (2013). Gorelick, N. (2017). Nominal 30-m cropland extent map
Understanding recent land use and land cover dynamics of continental Africa by integrating pixel-based and
in the source region of the Upper Blue Nile, Ethiopia: object-based algorithms using Sentinel-2 and Landsat-8
Spatially explicit statistical modeling of systematic tran- data on Google Earth Engine. Remote Sensing, 9(10),
sitions. Agriculture, Ecosystems & Environment, 165, 98– 1065. https://doi.org/10.3390/rs9101065
117. https://doi.org/10.1016/j.agee.2012.11.007 Yan, R., Huang, J., Wang, Y., Gao, J., & Qi, L. (2016). Modeling the
Teklay, A., Dile, Y. T., Asfaw, D. H., Bayabil, H. K., & Sisay, K. combined impact of future climate and land use changes on
(2021). Impacts of climate and land use change on hy- streamflow of Xinjiang Basin, China. Hydrology Research,
drological response in Gumara Watershed, Ethiopia. 47(2), 356–372. https://doi.org/10.2166/nh.2015.206