Flash flood forecasting using estimated precipitation by global satellite mapping in the North-East region of Bangladesh

Abstract

.Flash flood in the northeast region of Bangladesh damages. agricultural products of large areas causes deathand. destruction. to roads and bridges. Pre-monsoon flash floods damage the main crop Boro rice at the time of or just before the time of harvesting. About 60% of the total runoff in the Northeast Region is produced; mostly in the form of flash flood generated by the three Indian catchments- the Meghalaya River catehments~ the Barak River catchments~ and the Tripura River catchments. As . ..' rain gaugedata in the Indian catchments are not so avai1l:1ble~satellite based rainfall could be a good resource for estimation of flash flood inflows towards the northeast. region. The objectives of the study are to simulate Rainfall Runoff Model of the Northeast Bangladesh~ the Meghalaya River catchments~ the Barak River catchments~ and the Tripura River catchment with ..GSMap precipitation data and generate Flash Flood Forecast using Hydrodynamic Model incorporating WRF predicted precipitation. The result shows underestima~ion of runoff.As a result bias correction in GSMap rainfall is needed prior to application into operational flood prediction. With six years of data from 2009 to 2014; the 7-day moving average bias-adjustment was derived comparing the gauge observe'rainfall. The bias-adjustments were applied to every catchment. These bias-adjusted rainfalls when applied to the NAM model resulted in improvement in runoff for all catchments. The calibrated hydrodynamic model shows good -result in flood forecasting. Overall~ findings from this study indicate that the. GSMap underestimates rainfall signific~nt1y over Barak:, trans boundary and north-east catchments. The accuracy of GSMap can be improved by applying a. bias-adjustment. Prediction of water level using bias-adjusted rainfall estimates can improve the accuracy of water level prediction with considerable increase in the predictive capability of flood prediction for which the hydrological model needs to be calibrated.