Influence of atmospheric rivers in the occurrence of devastating flood associated with extreme precipitation events over Chennai using different reanalysis data sets

Abstract

The main focus of the present study is to examine the presence of Atmospheric Rivers (ARs) in the occurrence of Chennai (13.5° N, 80.5° E) flood event over Tamil Nadu which is south-eastern part of Indian peninsular region of India during November to December 2015. The vertically integrated horizontal water vapor transport (IVT) algorithm was used for the detection of ARs using three different reanalysis products, namely, the NCEP-NCAR (National Center for Atmospheric Research), NASA's Modern-Era Retrospective Analysis for Research and Applications (MERRA) and ECMWF Interim Re-Analysis (ERA-Interim). It is noticed that the occurrence of heavy precipitation events (HPEs) prior to the flood in the Chennai city, has been in good agreement with the presence of ARs during that time. The distribution from TRMM-3B42RT, TRMM-3B42 are validated with the with in situ rain gauge observations over the Chennai city. We have conducted quantitative analysis of IVT and their impact on heavy precipitation events and flood occurred over Chennai city. It is seen that MERRA reanalysis could be able to provide better presence of ARs. The study mainly reveals that persistent ARs of >18 h resulted in extremely heavy precipitation and lead to associated flood over Chennai. Also, the atmospheric conditions, ARs and rainfall simulated from Advanced Research Weather Research and Forecasting model (ARW) show good agreement with reanalysis and observations. Over the given reference period 1979–2015, correlation between IVT and HPEs is statistically significant at 99.5% confidence level. But the correlation between IVT during persistent ARs and corresponding HPEs is statistically significant at 90% confidence level. A large fraction of HPEs occur after ARs, with a small portion of ARs would lead to HPEs. The Chennai gird historical rainfall data is suitably fitted by Extreme Value Distribution (EVD) models. Inference model using the Maximum Likelihood Estimation (MLE) was applied on EVD considering their impact on the shape parameter and the confidence interval width. A Non-stationary generalized extreme value (GEV) analysis, was used to determine the strength of association between the precipitation extremes and IVT. The influence of the location and the scale parameter in the GEV model on different return levels (2-year, 20-year, and 100-year) are explicitly described. The return-level threshold values are found to change with IVT considerably. This study advocates that the detection of ARs can be used as important proxy in identifying the likelihood of occurrence of heavy precipitation/flash flood events for a given location using reanalysis data sets.