Real-Time Flood Forecasting Using Artificial Neural Networks (ANN) and Flood Inundation Mapping
No Thumbnail Available
Date
2019-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Flooding is one of the most destructive and harmful natural disasters occurring in many parts of
the world and there is increasing evidence that losses are rising largely because more people are
settling in flood prone areas. In many regions of the world, flood forecasting is one among the
few feasible options to manage floods and Ethiopia has no exceptional. Flooding in the country
is mainly linked with heavy rainfall and the topography of the highland mountains and lowland
plains with river banks system formed by the major river basins such as Baro River.
This study presents Real Time Flood Forecasting system using Artificial Neural Network (ANN)
and HEC-RAS integrated modeling in Baro River. ANN hydrological flood forecasting model
set up using both deterministic and stochastic approach with Rainfall, Temperature and
Topographic Wetness Index (TWI) as parametric inputs and trained random neurons weights as
stochastic variable. The hydrological model trained and validated using 7 years (1999-2005) and
three years (2006-2008) observed stream flow data respectively. And its performance also
evaluated with 0.84 and 0.87 NSE values at calibration and validation period respectively.
Similarly, for hydraulics modeling, using Normal Difference water Index (NDWI) revealed that
both recorded flood events and flood extent area obtained from HEC-RAS are overlapped up to
96% during calibration and validation.
The Real time forecasting of flood and its inundation area also evaluated using forecasted daily
rainfall and temperature for 3, 7 and 10-days during (May 27, 2019-June 05, 2019) rainy period
and these results further compared with the real time condition after 3, 7 and 10 days and showed
very good performance. In addition to these, three decades future flood affected areas with
different climate change scenarios identified to warn the inhabitants and development
investments.
Description
Keywords
Artificial Neural Networks (ANN), HEC-RAS, Inundation Mapping and Flood, prone Areas, Normal Difference water index, Real-time flood forecasting