Development a Flood Forecasting Model Using the Multi-Sensor Data Fusion Approach: Case Study of Karoun Basin Iran (PAP014728)
Reza bazaz, Shahab Araghinejad
Flood forecasting and early warning systems
One of the main concerns of governmental agencies is forecasting and warning of floods to mitigate the huge loss of this phenomenon in the basins. Application of multi-sensor data fusion in the context of hydrology has raised accuracy and precision of hydrologic estimations. In this study, the technique of multi-sensor data fusion is applied to combine information received from different sources including CMORPH (Climate prediction center MORPHing technique), and TRMM (Tropical Rainfall Measurement Mission) sensors as well as the rainfall recording gauges to calibrate a real time flood forecasting model for the great Karoun basin in the western part of Iran. In order to forecast rainfall with an acceptable lead time, a customized Artificial Neural Network (ANN) model is used. By simulating runoff from the forecast rainfall, the contribution of each data source to maximize the accuracy and precision of flood forecasting model is investigated. In the other hand, The criteria for determining the contribution of each data source comes from its role in estimating the flood hydrograph of the whole basin. The precipitation maps obtained from CMORPH and TRMM are combined by the recorded field data to create a modified rain map using a proposed spatial version of K-nearest neighbor technique. A multi-sensor data fusion scheme is used for data sources to determine the contributing weight of each data source at different time and space situations. The results demonstrate the supremacy of the proposed dynamic approach of data fusion over the conventional approaches.