This paper presents a new data-driven modelling framework for forecasting probabilistic flood inundation maps for real-time applications. The proposed end-to-end (rainfall-inundation) method combines a suite of Machine Learning (ML) algorithms to forecast discharge and deliver probabilistic flood inundation maps with three-hour lead time. The study applies Random Forest (RF)-based rainfall-discharge models on top of Multi-layer Perceptron (MLP)-based classifiers to classify wet/dry cells. The concept of an ML-based hybrid modelling framework is tested using two subsets of data created from an observed fluvial flood event in a small flood prone town in the UK. The results show that the model can effectively emulate the outcomes of a hydrodynamic model (i.e. Flood Modeller) with considerably high accuracy measured in terms of flood arrival time error and classification accuracy. The mean arrival time difference between the proposed model and the hydrodynamic model is of order 1 hour 53 minutes. The classification accuracy is measured against a radar image and the accuracies read 88.22% and 86.58% for the proposed data-driven and Flood Modeller (FM) model, respectively. The key features of the proposed modelling framework are that it is simple to implement, detects flooded cells effectively and substantially reduces computational time.
|Journal||Proceedings of the Institution of Civil Engineers: Water Management|
|Early online date||26 May 2020|
|Publication status||E-pub ahead of print - 26 May 2020|