Author(s): Gwangseob Kim; Kun-Yeun Han; Hyuk-Joon Cho I.; Jong Tae Lee
Linked Author(s):
Keywords: Rainfall forecast; Satellite data; Neural networks; Flood predict ion
Abstract: A quant itative rainfall forecasting model was developed using satellite data and neura l networks and improved rainfall forecasts were applied to flood analysis. To overcome the geographical limitat ion of Korean peninsula and to get the long forecast lead time of 3 to 6 hour, the developed rainfall forecast model took use of satellite imageries and wide range AWS data. The architecture of neural network model is a mult i-layer neural network which consists of one input layer, one hidden layer, and one output layer. Neural network is trained using a momentum back propagation algorithm. We developed a dynamic flood inundat ion model which is associated with 1-dimensional flood routing model. The developed rainfall forecast model and flood analys is system were applied to the Nakdong River Basin for the heavy storm period between 6th and 16th of August, 2002. The results demonstrated that the rainfall forecasts of 3 hours lead time showed good agreement with observed data. The inundat ion aspect of simulation depends on actual levee failure in the same basin. Rainfall forecasts were used for flood amount computation in the target watershed. Also the flood amount in the target watershed was used on boundary condit ion for flood inundation simulat ion in protected lowland and a river. The results of simulat ion are consistent with actuality inundation traces and flood level data of the target watershed. This study provides practical applicability o f satellite data in rainfall forecast of extreme events such as heavy rainfall or typhoon. Also this study presented an advanced integrated model of rainfall, runo ff, and inundat ion analys is which can be applicable for flood disaster prevent ion and mit igat ion.
Year: 2007