Abstract:
Land use land cover changes of an ecosystem should be significantly studied, assessed
and controlled by any means to cooperate with sequential progress of mankind. This
study examined land use land cover changes in Toyboy river catchment located at Mo gibe river basin, Southern Ethiopia with the aim of LULC classification, streamflow
modelling in relation with LULC changes of different decades and identifying sub basins
of maximum and minimum contribution of surface runoff using ArcGIS 10.3 with its
extension ArcSWAT and ERDAS Imagine 2015 with maximum likelihood algorism of
supervised classification. Change detection analysis and accuracy assessment was done
after Image pre-processing and classification. Significantly; from year 1992 to 2013,
both agricultural land and settlements increased from 24.5% to 41% and from 1.02% to
32% respectively while forest and grass lands decreased from 25.4% to 6.7% and from
15.1% to 5.5% respectively. Accuracy assessment of land use classification by confusion
matrix and kappa index for 1992, 2000 and 2013 maps indicated an overall accuracy of
87.8%, 80.8% and 90.5% and kappa coefficient of 0.88, 0.8 and 0.9 respectively and
image classification accuracy was almost in strong agreement. Model calibration (1999-
2009) and validation (2010-2015) were performed for monthly flow data at Woybo
gauging station. The model performance was found very well according to the three
objective function results of R2
, NSE and PBIAS with different LULC. Hence it is 0.76,
0.75 and 1.8% for calibration and 0.75, 0.74 and -1.5% for validation. Considerable
consequences were obtained in the surface runoff of the watershed which was increased
from 1992 to 2000 by 7.4% and from 2000 to 2013 by 10.5% due to change of LULCwhereas soil water content, groundwater flow and interflow were decreased when
surface runoff was maximum. Due these results, there is some sort of recommended
works for the future studie
