Future Sea Level Rise at Indian Ports Using a Combined Numerical and Data-Driven Approach

Abstract

The estimation of future sea level rise (SLR) under the changing climate at India’s major ports is necessary considering India’s large port infrastructure, spread across its ~7000 km of coastline. In this study, the future SLR for next 30 years is predicted using a combination of general circulation models (GCMs) and artificial neural networks (ANNs) to take advantage of both physics-based as well as data-driven approaches. For every port, monthly means of six climatic causal variables, namely sea surface temperature, precipitation, sea level pressure, surface salinity, wind speed, and surface height above geoid were used as input to the ANN to obtain the output of monthly mean sea levels. In the training, past sea levels recorded by tide gauges were used as output. The climatic input variables pertained to eleven different CMIP6 GCMs with SSP2-4.5 as the future climate scenario. Using the trained network and for every GCM, monthly sea levels were predicted for next 30 years, where the input was the future causal parameters. The rate of SLR was determined by fitting regression to the variation of the sea levels against time. This process was repeated for all GCMs. The median of all such GCM-yielded SLR, derived for each port, was compared with the same based on a software protocol called: SimCLIM which is based on the use of GCMs alone. It was found that such ANN-based SLR rate at major Indian ports was lower than the SimCLIM-based one, and it varied from 1.94 mm/year (Chennai) to 4.11 mm/year (Mumbai). The proposed approach is site-specific and hence more appropriate to use than the spatially averaged projections from SimCLIM or satellite data.