Abstract:
Many river catchments, including those in the upper Jemma sub-basin, lack continuous
streamflow monitoring, and existing hydrologic models for ungauged catchments are
data-intensive, requiring calibration and uncertainty analysis. The study evaluates the
regionalization of SWAT (Soil and Water Assessment Tool) model parameters for use in
ungauged catchments of the upper Jemma sub basin targeted on Ajima ungauged
catchment. The SWAT model is calibrated and validated using monthly streamflow data
from five selected nearby gauged catchments over the period 1996-2002 and 2003-2009
respectively using SWAT-CUP (SIFU-2). Model performance is satisfactory, with a
coefficient of determination (R²) ranging from 0.61 to 0.90 and Nash-Sutcliffe Efficiency
(NSE) from 0.62 to 0.93 for both calibration and validations. Parameters like CN2,
GW_DELAY, ESCO and GW_REVAP are identified the most influential model
parameter for streamflow prediction in the study area. Selected regionalization methods
are tested by applying a leave-one-out cross-validation technique on the four gauged
catchments, treating each as pseudo-ungauged catchment. All regionalization methods
performed reasonably well (R² > 0.6) in estimating streamflow for ungauged catchments.
However, spatial proximity outperformed the other methods, providing the most accurate
streamflow estimates. The stream flow of the ungauged Ajima catchment is estimated to
44.55 million cubic meters (MMC).base flow is a key water balance components which
influence the stream flow. The estimated stream flow and other water balance
components used for different water application in the catchment.
