Performance Evaluation of Satellite Rainfall Estimates for Flood Monitoring in Gumera Watershed, Amhara Region, Ethiopia
No Thumbnail Available
Date
2022-12-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Since the 1980s, several satellites have been providing rainfall data for the tropical
parts of the world, including Ethiopia. These data could be used for different
applications in various fields of study in areas where ground-based rain gauge stations
are unavailable and sparsely distributed. The purpose of this study is to evaluate the
performances of satellite rainfall estimates (CHIRPS-V2, TAMSAT-V3.1, and
PERSIANN-CDR) for flood monitoring in the most flood-prone area of Amhara Region,
Ethiopia. Daily rainfall data from 2004 to 2019 were collected from the National
Meteorological Agency. In Addition, satellite rainfall estimates that cover the same
periods, and historical flood events in the study area were collected from the internet
database and district agriculture office respectively. Continuous statistical indices and
categorical statistical indices were employed for data analysis. The performances of
the three-satellite rainfall estimates were evaluated on four major temporal scales;
annual, seasonal, monthly, and daily temporal scale. In terms of annual and seasonal
temporal scales, TAMSAT-V3.1 and CHIRPS-V2 performed well and outperformed
PERSIANN-CDR on the major statistical indices. In the Monthly temporal scale,
CHIRPS-V2 performed well on the major continuous statistical indices. The application
of satellite rainfall products for flood monitoring is also evaluated by using the daily
rainfall estimates in two ways; using the overall daily observed rainfall from 2004 to
2019 and using the daily rainfall data during the three most flood events (2006, 2017
and 2019) in the study area. In both cases, all the three satellite datasets underestimate
the highest amount of observed rainfall and overestimate the lowest amount of daily
rainfall conditions. Relatively, CHIRPS-V2 had the best skill in estimating the highest
daily rainfall observed in the study area. This research indicated that performance
evaluation of satellite rainfall datasets must be conducted before applying for flood
monitoring. The study could not address the performance of satellite rainfall estimates
in predicting when and where floods will occur in the future. Therefore, future research
should be focused on the flood prediction performance of satellite-derived rainfall
data.
Description
Keywords
Remote Sensing; Flood; Satellite Rainfall Estimate, Evaluation; Dataset