Evaluating the Land Use and Land Cover Change Impact on Streamflow Kulfo Catchment
No Thumbnail Available
Date
2021-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Stream flow is one of the major hydrological cycle components which can be altered and /or affected by different factors. From these, the Land Use Land Cover change is one of the main factors to alter and change the stream flow characteristics. Therefore, in this study, the LULC change impact on the stream flow was evaluated using the SWAT model for the case of Kulfo catchment which is situated in the Southern part of Ethiopia. Sensitivity, calibration, and validation were also conducted using sequential uncertainty fitting–version 2 (SUFI-2) in SWAT-CUP (Calibration and Uncertainty program) using historical streamflow data for each of the land use land cover years. Four different algorithms (SVM, CART, RF, and Navia's) which are existed in the Google Earth Engine were compared and the best performed was selected to generate the time-series LULC maps of, (1986, 2000, 2016, and 2020), the study area. The accuracy of these algorithms was evaluated corresponding to GCPs collected from the field. High resolution (30m) Landsat images which are thematic mapper (TM), Enhanced thematic mapper plus (ETM+), and Operational land imager (OLI) were used with the aid of historical trends and ground-based data used to train and validate the LULC maps generated from Google Earth Engine platform. As a result, the SVM algorithm was performed better in LULC classification than other algorithms which are compared in this study. In the analysis period of this study, the Vegetation land cover area has decreased from 18.81% to 3.1%, the agricultural land was increased from 19.44% to 57.12%, whereas the shrub land area has been decreased from 34.18% to 14.73%. Therefore, the effect of these LULC changes on stream flow was evaluated using Soil & water assessment tool (SWAT) model and high mean monthly and seasonal streamflow variability was observed in the analysis period of this study. These variabilities were increased from 6.72% to 7.85% monthly. In seasonal variability, the stream flow has decreased trend for all of the seasons (Kiremt, Belg, and Bega) from the year 2016 to 2020 whereas increased trend for the period of 2000 to 2016 (Kiremt, Belg and Bega). The Results from the calibration resulted in an acceptable range (0.6, 0.8, 0.75, and 0.75 for NSE and 0.75, 0.76, 0.79, and 0.81 for R2) between observed and simulated streamflow respectively. The results of validation were also fallen in the acceptable range (0.72, 0.6, 0.74, and 0.75 for NSE and 0.8, 0.75, 0.73, and 0.8 for R2) observed and simulated streamflow respectively.
Description
Keywords
Streamflow, Machine Learning, Kulfo, GEE, LULC, Kulfo, SWAT-CUP