Distributed hydrological model using machine learning algorithm for assessing climate change impact

Abstract

Rapid population growth, economic development, land-use modifications, and climate change are the major driving forces of growing hydrological disasters like floods and water stress. Reliable flood modelling is challenging due to the spatio-temporal changes in precipitation intensity, duration and frequency, heterogeneity in temperature rise and land-use changes. Reliable high-resolution precipitation data and distributed hydrological model can solve the problem. This study aims to develop a distributed hydrological model using Machine Learning (ML) algorithms to simulate streamflow extremes from satellite-based high-resolution climate data. An integrated statistical index coupled with a classification optimisation algorithm was used to select coupled model intercomparison project (CMIP6) global climate model (GCMs). Several bias-correction methods were evaluated to identify the best method for downscaling GCM simulations. The study also evaluated the performance of different Satellite-Based Products (SBPs) in replicating observed rainfall to select the best product. A novel two-stage bias correction method were used to correct the bias of the selected SBP. Besides, four widely used bias correction methods were compared to select the best method for downscaling GCM simulations at SBP grid locations. A novel ML-based distributed hydrological model was developed for modelling runoff from the corrected satellite rainfall data. Finally, the model was used to project future changes in runoff, and streamflow extremes from the downscaled GCM projected climate. The Johor River Basin (JRB) located at the south of Peninsular Malaysia was considered as the case study area. The results showed that three GCMs, namely EC-Earth, EC-Earth-Veg and MRI-ESM-2, were the best in replicating the precipitation climatology in mainland Southeast Asia. IMERG was the best among five SBPs with an R2 of 0.56 compared to SM2RAIN-ASCAT (0.15), GSMap (0.18), PERSIANN-CDR (0.14), PERSIANN-CSS (0.10) and CHIRPS (0.13). The two-step bias correction approach improved the performance of IMERG, which reduced the mean bias up to 140 % compared to the other conventional bias correction methods. The method also successfully simulates the historical high rainfall events that caused floods in Peninsular Malaysia. The distributed hydrological model developed using ML showed NSE values of 0.96 and 0.78 and RMSE of 4.01 and 5.64 during calibration and validation. The simulated flow analysis using the model showed that the river discharge would increase in the near future (2020 - 2059) and the far future (2060 - 2099) for different SSPs. The largest change in river discharge would be for SSP-585. The extreme rainfall indices, such as R95TOT, R99TOT, Rx1day, Rx5day and RI, were projected to increase from 5% for SSP-119 to 37% for SSP-585 in the future compared to the base period. The ML based distributed hydrological model developed using the novel two-step bias corrected SBP showed sufficient capability to simulate runoff from satellite rainfall. Application of the ML-based distributed model in JRB indicated that climate change and socio-economic development would cause an increase in the frequency streamflow extremes, causing larger flood events. The modelling framework developed in this study can be used for near-real time monitoring of flood through bias correction near-real time satellite rainfall.