A Quantitative Tropical Cyclone Risk Assessment Study over the Eastern Coast of India

Abstract

Tropical Cyclones (TCs) are regarded as major devastating phenomena that generate torrential rainfall, extreme winds, and severe storm surges, causing loss of lives and infrastructure and ecological-environmental damages annually. Globally, the tropical warm North Indian Ocean (NIO) basin is considered as an active breeding zone for TCs, and a higher population density over the coastal regions exacerbates the risk. The eastern coastal areas of India, which are more prone to frequent TC strikes, are considered for the study. A statistical approach is adopted in this study to project the future probabilities of risk induced by TC rainfall and extreme wind speed events. The extreme value theory technique via frequentist and Bayesian paradigms is evaluated. The frequentist statistics include generalized extreme value (GEV) and gamma, whereas the Bayesian approach is configured using a non-informative prior (NIP) method. Also, various parameter estimation schemes are incorporated with previous models, viz., maximum likelihood estimation (MLE), generalized MLE, and linear moments (L-mom), to capture the heterogeneous characteristics of the rainfall and wind extremes. The results revealed that the frequentist paradigm in GEV with L-mom configuration is robust approach to capture the reasonable outputs over most locations with lower error and the least uncertainty for wind extremes. However, the Gamma with MLE and Bayesian paradigm in NIP configuration exhibits better and optimal fitting for rainfall having heavier tails. The estimated recurrence interval shows that the coastal districts of West Bengal and Odisha, followed by Andhra Pradesh, are highly susceptible to severe cyclonic wind hazards in the near future. However, the coastal and near-coastal parts of Andhra Pradesh and Odisha, followed by the southern coastal districts of West Bengal, are exposed to TC-induced heavy rainfall events in the near future.