Multi-model fusion-based framework for daily flood forecasting in multiple-step-ahead and near future under climate change scenarios

Abstract This study proposes a novel fusion framework for flood forecasting based on machine learning, statistical, and geostatistical models for daily multiple-step-ahead and near future under climate change scenarios. To do this, remote sensing precipitation data of ERA5, CHIRPS, and PERSIANN-CDR were utilized to fill the gap data of meteorological stations. Four Individual Machine Learning (IML) models, including Random Forest, Multiple-Layer Perceptron, Support Vector Machine, and Extreme Learning Machine were developed for twelve days ahead of streamflow modeling. Then, three fusion models, including Random Forest, Bayesian Model Averaging, and Bayesian Maximum Entropy were applied to combine the outputs of IML models. The proposed framework also was implemented to downscale the precipitation variable of three general climate models (GCMs) under SSP5-8.5 and SSP1-2.6 scenarios. The results indicated that individual models illustrated weak performance, especially in far steps flood forecasting, so it is necessary to utilize a fusion technique to improve the results. In the fusion step, the RF model indicated high efficiency compared to other fusion models. This technique also demonstrated an effective proficiency in downscaling precipitation data of GCMs on a daily scale. Finally, flood forecasting model was developed based on the fusion framework in the near future (2020–2040) by using the precipitation data of two scenarios. We conclude that flood events based on both SSP5-8.5 and SSP1-2.6 will increase in the future in our case study. Also, the frequency evaluation shows that floods under SSP1-2.6 will occur about 10 percent more than SSP5-8.5 in the Kan river basin from 2020 to 2040.