Deep Learning Approaches for Multi Scale Flood Modelling of Ganga River

Código

I-EBHE0135

Autores

Rahul Neware

Tema

WG 1.12: Development & application of river basin simulators

Resumo

Floods are the leading cause of natural disasters worldwide. In recent times the frequency and intensity of floods in all forms (flash, riverine, and storm surge) have dramatically increased due to anthropogenic impact and climate variability. Flood forecasting and mitigation have become increasingly crucial in recent years, with the Ganga River basin being a region of particular concern due to its susceptibility to devastating floods. Machine learning and deep learning techniques have emerged as promising solutions to address the challenges of accurate, scalable, and timely flood prediction in this region. Multi-scale flood modelling, which accounts for the complex interplay of factors at different spatial and temporal scales, is essential for effectively managing flood risks along the Ganga River. In order to reduce flood damage, researchers, policymakers, and decision makers must do hydrological modelling within a catchment. Especially in urban areas where flood modelling is challenging and requires dynamic downscaling of hydrological and hydrodynamic processes across multiple scales. Our study introduces a deep learning method for multi-scale flood modelling of the Ganga river and its challenges. The study covers coupled hydro meteorological and fine-scale (~1?10 m) hydrodynamic models at the mesoscale (~1?40 km), as well as the use of remote sensing methods to validate the simulations. Thus, the work will help policy makers and the hydrological research community better comprehend multi-scale coupled models for flood modelling.