Assessment of Hydrological Impacts of Land Cover Changes and Climate Variability in the Geba Catchment, Ethiopia

Abstract

Land use/land cover (LU/LC) and climate are the two main factors directly influencing catchment hydrological processes and consequently changes in these factors will result in significant hydrological impacts. Quantifying the magnitude and direction of these impacts is of great importance for land use planning and sustainable water resources management. The Geba catchment (5137 km2) located in the highlands of Northern Ethiopia; Africa contributes a significant portion of flow in the river Nile and forms an important source of water to a large population. In the past few decades, the catchment has experienced significant changes in LU/LC in the form of degradation due to anthropogenic activities and subsequent restoration brought about by conservation measures. Also, trend analysis of hydro-meteorological data carried out as part of this study provided evidence of changes in rainfall and temperature regimes in the catchment. Therefore, the present study was taken up to characterize the hydrology of the Geba catchment using available hydro-meteorological data and to apply and evaluate the potential of the Soil and Water Assessment Tool (SWAT) model to simulate major hydrological processes and sediment dynamics in the catchment. The objective was to use SWAT to simulate changes in hydrological processes brought about by changes in LU/LC and climate variability within the catchment. Accordingly, the research methodology adopted involved the following tasks: (1) Using historical (1971-2013) ground-based observations of rainfall, air temperature (7 climate stations) and streamflow (1 gauging station) statistical and trend analyses were carried out for monthly and seasonal time steps. Also, trends in several extreme climatic indices related to rainfall and temperature were analysed (2) LANDSAT satellite imagery acquired for multiple dates during the period 1971-2013 were subject to standard image processing and supervised classification procedures to derive LU/LC maps for the Geba catchment. These classified maps were used to detect changes in different LU/LC classesii during the periods 1973 – 1987, 1987 – 2000 and 2000 – 2013 (3) Using a variety of inputs (ground and satellite-based) related to topography, soils, LU/LC, rainfall and climatic variables, the ArcGIS version of the SWAT model (ArcSWAT) was applied to the Geba catchment. Given that the catchment experienced significant changes in LU/LC over the 40 year period considered, a novel model calibration/validation approach was adopted involving the use of different LU/LC maps for different time periods (4) Using observed streamflow records at the outlet of the Geba catchment, the SWAT model was subject to sensitivity analysis following which calibration and validation was carried out using both monthly and daily time steps. Model performance in simulating streamflow and sediment concentration at the outlet was evaluated using different statistical criteria (5) Using the validated SWAT model, a novel method to evaluate the separate and combined impacts of LU/LC changes and climate changes on major water balance components in the Geba catchment was implemented. Results of trend analysis revealed that during the study period (1971-2013), rainfall and streamflow exhibited a decreasing trend, while maximum daily air temperature had an increasing trend and minimum daily air temperature showed decreasing trend at 95 % confidence level. As regards LU/LC changes, during 1973–1987 and 1987–2000 time periods about 10.83 % and 9.13 % of the catchment area was transformed largely from shrub, forest and rangeland mainly to agriculture and barren land. During 2000–2013, about 18.37 % of the total catchment area was transformed from barren land and range to agriculture, shrub, forest and urban area. SWAT model validation using observed streamflow records yielded values of coefficient of determination (R2) between 0.86 and 0.96 and Nash-Sutcliffe efficiencies (ENS) between 0.73 and 0.83 for different simulation periods with a monthly time step. For daily streamflow predictions, R2 values ranged between 0.77 and 0.91 and ENS values were between 0.7 and 0.79. SWAT also provided reasonably accurate predictions of daily sediment concentrations during validation (R2: 0.81-0.895, ENS: 0.79-0.80). These results prove that the SWAT model is a reasonably accurate tool for simulation of hydrological processes in the Geba catchment, whereas R2iii and ENS for daily and monthly flow were very less (satisfactory) for the single static LU/LC (2000) map, mostly followed in many studies. Impacts of LU/LC changes and climate variability were evaluated by dividing the study period (1973-2013) into three phases based on LU/LC and climatic conditions: Phase (I) - LU/LC maps of 1973 and 1987, climate of 1974-1983 and 1984-1993 Phase (II) - LU/LC maps of 1987 and 2000, climate of 1984-1993 and 1994-2003 Phase (III) - LU/LC maps of 2000 and 2013, climate of 1994-2013 and 2004-2013. The SWAT model was run separately for four scenarios in each phase involving combinations of LU/LC and climate. Results indicated that the combined impacts of the LU/LC changes and climate variability increased streamflow and potential evapotranspiration in both Phases I and II, while available soil water contents decreased. Positive impacts in the form of reduced streamflow and increased soil moisture resulted in Phase III due to extensive conservation measures implemented after 2000. Overall, changes in LU/LC seemed to have a higher impact on hydrological processes than changes in climate. The present study has demonstrated the applicability and efficacy of a convenient methodology integrating satellite remote sensing and modelling to characterize hydrological processes and simulate hydrological changes in a heterogeneous tropical catchment. The proposed strategy may be adopted to formulate strategies for sustainable land and water resources management in the region, and in similar hydro-climatic settings elsewhere in Africa.