Assessment of Summer Monsoon Precipitation Predictability Skill Over Homogeneous Regions of India Using Monsoon Mission Coupled Models

Due to intrinsic predictability limits, predicting precipitation on a regional scale at a seasonal time scale using a coupled model is challenging. Despite this, rainfall variability information at the regional scale is crucial for various applications, including agricultural practices, water resource management, disaster management, etc. In this study, we compare the skills of the Monsoon Mission Coupled Forecast version 2 (MMCFSv2) model in reproducing the precipitation over five homogenous regions of India between 1998 and 2022 and compare it with the previous version of the model (MMCFSv1). We also identify why a particular model would exhibit better interannual variability and skill (anomaly correlation coefficient) over the other. Analysis reveals that both models have a dry bias in precipitation over the Indian landmass. However, the MMCFSv2 model has a lesser bias and better simulates interannual rainfall variability compared to MMCFSv1. The MMCFSv2 model has better monthly and seasonal mean precipitation skills over India compared to MMCFSv1. The MMCFSv1 and MMCFSv2 models have a skill of 0.54 and 0.72, respectively, in reproducing interannual variability of mean rainfall over India. MMCFSv2 improves prediction skills over four homogeneous regions viz. CNEI, NEI, SPI, and WCI. The skill over NWI is lesser in MMCFv2 compared to MMCFSv1. The MMCFSv2 model reasonably simulates the relationship between precipitation in each homogeneous region of India and tropical Indo-Pacific Sea surface temperature. Additionally, it exhibits significantly better spatial skill in capturing the correlation between rainfall and SST in these regions compared to MMCFSv1.