Assessment of Stormwater Drainage Network of the Southwest Zone in the Surat City

Authors : Arpit Sharma* and Ganesh D. Kale**

Artificial Intelligence and Water Resource Development(Ad)

*PhD Student, Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat, Gujarat-395007, India

**Associate Professor, Department of Civil Engineering, Sardar Vallabhbhai National Institute of Technology, Surat, Gujarat-395007, India

The World Meteorological Organization described the year 2020 as one of the most extensively flooded years in its State of the Global Climate 2020 Report. Due to its impacts on hydrologic components, climate change is a crucial concern that influences the effectiveness of drainage networks in urban areas. Urbanization, unlawful settlement along the riverbanks, bank erosion, dense population, heavy rainfall, industry, and high tide all contribute to flooding in shoreline urban floodplains such as Surat City. Increases in precipitation inside the city, its surrounding areas, and in the Tapi basin are predicted to be caused by climate change, which will likely worsen the risk of flooding. The population of the Southwest Zone of Surat City according to Census 2001 was 242466 and according to Census 2011, it was 348423. This indicates rapid urbanization of the Southwest Zone of Surat City. As a result, the current study is being carried out to assess the suitability of the storm drainage system in the Southwest Zone of the city of Surat in the context of urbanization that occurred in the past few decades and climate change. In the current investigation, an effort has been made to analyze the performance of several multi-model ensembles corresponding to CMIP6 and generated by using four distinct machine learning regression techniques namely Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbour Algorithm (KNN) and Artificial neural network-multilayer perceptron neural network (ANN-MLP). The study also tried to identify the optimum number of GCMs to be included in the multi-model ensemble. The storm drainage network of the Southwest Zone of Surat City was evaluated for the observed rainfall and rainfall generated by using a multi-model ensemble of 18 top-ranked CMIP6 GCMs corresponding to SSPs namely ssp585 and ssp245 for different return periods viz. 2, 5, 10, and 20 year. The investigation revealed that the existing storm drainage network contains 5 hotspots and 14 hotspots for rainfall events corresponding to two-year return periods for observed and SSP585 (ANN-MLP) CMIP6 datasets, respectively, and these hotspots are increasing for the rainfall events corresponding to five, ten and twenty year return periods.