Abstract:
Evapotranspiration is one of the key variables in the hydrological cycle. The objective in this
study is to evaluate the spatial-temporal pattern of actual evapotranspiration in the Kilti
watershed using SEBAL for different land uses and land covers. Landsat 8 satellite image was
used for land use classification and to extract inputs to the Surface Energy Balance Algorithm
for Land (SEBAL) model. About 1235 ground control points were collected in Kilti catchment
during field visit. The image classification was done for the recent year (2017) with the
maximum likelihood method for 9 land cover classes (such as rain-fed crops, irrigated crops,
water and marshy, built-up, forests, plantations, grassland, shrubs, and bare lands). Cropping
calendar and NDVI provides key information for refining irrigated land cover. Accuracy of the
classification was determined by means of the confusion matrix (sometimes called error
matrix). An overall classification accuracy of 94%, with a producer's accuracy of 92% and
user's accuracy of 96%, Kappa statistics of 89 %, which are within the acceptable range as
suggested by previous studies. SEBAL was used to estimate the actual evapotranspiration of
Kilti for the period from January to December, 2017. Cloud free images were used, however,
data gaps in certain images, which were affected by the cloud, was filled using advanced linear
interpolation technique, using neighboring pixels from next available and previous image.
Daily, monthly, and annual AET was estimated for this study area by SEBAL algorithm using
satellite data from Landsat 8. The estimated evapotranspiration has comparable magnitude and
temporal pattern as of the potential evapotranspiration in the study area in previous studies.
These estimates are also within reported ranges in literature for other catchments. The highest
and minimum AET value was estimated in September and April as 7.5 and 4.2 mm/d
respectively. The temporal pattern the AET is similar to that of the PET while the magnitude
of the former is relatively smaller as expected. The plantations, dominated by Eucalyptus tree
abstract the largest amount of water in the study area compared to other land cover classes.
Similar studies should be undertaken in other catchments in Ethiopia to better our
understanding of catchment water abstractions.
