‘Comparison and Evaluation of Satellite Rainfall products for Hydrological Modeling (case of Wabe watershed, Ethiopia
No Thumbnail Available
Date
2019-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Satellite-based Rainfall products have been playing an immense role and used as an alternative
source of data in regions where conventional rainfall measurements are not readily available or
inadequate.
The objective of this study is to evaluate and compare high-resolution satellite rainfall products such
as CHIRPS, PERSIANN-CCS, RFE and TAMSAT with the ground-based observed rainfall data
over the Omo-Gibe River Basin. For spatial assessment, a point to pixel approach at different
temporal scale over the time window of 2003-2017 is used. Moreover, the capabilities, applicability,
and limitations of satellite rainfall products were also evaluated by forcing the hydrological model
(HBV-light) in Wabe Watershed from Omo-Gibe River Basin. Continuous statistics was used to
assess their performance in estimating and reproducing rainfall amounts and categorical statistics was
used to evaluate rain detection capabilities while Kling-Gupta Efficiency (KGE) is used for model
performance evaluation.
At mean daily scale, a good correlation agreement was observed by Wolkite with TAMSAT, RFE,
PERSSIAN-CCS, CHRIPS (r=0.769, r=0.686, r=0.627, r=0.543) respectively. While the remaining
station with the products show a low correlation and the time step has an important influence. This
showed when the time step increases the accuracy of satellite rainfall products to predict the rainfall
event relative to the station value will increase.
The HBV model was simulated using datasets from 2003-2012 considering both satellite rainfall
products and ground-based observed dataset. The model has shown good performance when
calibrated with the gauged observed rainfall data. During calibration, the objective function showed
KGE= 0.42 for daily, and KGE=0.61 for monthly time scale; whereas, the validation period showed
KGE=0.54 and 0.71 for daily and monthly time scales respectively. The adjusted satellite rainfall
estimates TAMSAT and RFE showed relatively good performance while adjusted PERSIANN-CCS
and RFE relatively showed less performance. The adjusted TAMSAT data set perform the best result
than that of the other dataset at daily and monthly time scale with KGE=0.39 and KGE=0.54 for
calibration and KGE=0.6 KGE=0.72 for validation period respectively. Finally, this study reveals
that besides the gauged observed rainfall data the bias-adjusted TAMSAT and RFE dataset can be
used as an alternative dataset for hydrological modeling for the study area.
Description
Keywords
Omo –Gibe, Wabe, HBV-light, CHIRPS, PERSIANN-CCS, RFE, TAMSAT, KGE