Evaluating Dry and Wet Season Precipitation from Remotely Sensed Data Using Artificial Neural Networks for Floodplain Mapping in an Ungauged Watershed

Abstract

This research creates a framework for modelling the rainfall-runoff process using satellite precipitation data and a floodplain map in ungauged urban watersheds. The combined effects of urbanisation and climate change over the past few decades have increased the number of flooding incidents. Accurate prediction of the flood-prone zone is crucial for policymakers and system managers to build the watershed's resilience during catastrophic flooding events. Precipitation and runoff data are crucial for hydraulic and hydrologic analysis and for identifying flood-prone areas. However, it is difficult to obtain precipitation and discharge data for hydrologic analysis in data-scarce regions. In this context, this research employs satellite precipitation products for rainfall-runoff analysis, which is subsequently utilised in a hydraulic model to delineate a flood-prone zone in an ungauged watershed. The Hydrologic Engineering Centre-River Analysis System (HEC-RAS) and Hydrologic Modelling System (HEC-HMS) models were utilised in the study region to simulate and analyse interactions between rainfall, runoff, and the extent of the flood zone. Setting up and calibrating the HEC-HMS model using a satellite precipitation product is required for the dry and wet seasons. For the wet and dry seasons, HEC-HMS gets validated with an R-square value of 0.72 and 0.85, respectively. Three types of simulations were conducted in the calibrated HEC-HMS model to create the hydrograph with 25-, 50-, and 100-year of rainfall return periods. Finally, the one dimensional HEC-RAS model generates a flood inundation map for the pertinent flooding occurrences from the acquired peak hydrograph. By comparing the values of the specified return periods, the produced flood map depicts the affected area during various return periods of flooding events and provides a quantifiable display of inundation extent percentage (IE%).