Impact of the Ensemble Kalman Filter Based Coupled Data Assimilation System on Seasonal Prediction of Indian Summer Monsoon Rainfall

• Published in: Geophysical research letters 2022-08, Vol.49 (15), p.n/a
• Main Authors: Gade, Sagar V., Sreenivas, Pentakota, Rao, Suryachandra A., Srivastava, Ankur, Pradhan, Maheswar
• Format: Article
• Language: English
• Subjects: Adaptation; Agriculture; Analysis; Annual rainfall; Annual variations; Atmosphere; CFSv2; Climate system; Computer applications; Convection; coupled data assimilation; Data assimilation; Data collection; ensemble methods; Indian summer monsoon; Kalman filters; LETKF; Meteorology; Methods; Monsoon forecasting; Monsoon rainfall; Monsoons; Oceans; Precipitation; Prediction models; Rainfall; Rainfall forecasting; Satellite observation; seasonal prediction; Summer; Summer monsoon; Tropical meteorology; Wind
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2021GL097184

The sensitivity of seasonal prediction (June to September) of Indian monsoon to initial state from two variants of coupled data assimilation (CDA) products, viz. the Climate Forecast System (CFS) Reanalysis (CFSR) and Indian Institute of Tropical Meteorology, University of Maryland‐ Weakly Coupled Analysis (IWCA) is explored in this study. The IWCA implements the local ensemble transform Kalman filter, and incorporates theoretically advanced features of flow‐dependency and ensemble‐based analysis compared to CFSR. The CFS version‐2 predictions using IWCA simulate the large‐scale monsoon features, and convection centers well, and improve prediction skills compared to CFSR predictions. The enhanced analysis quality and Ocean‐Atmospheric cross‐domain equilibrium in IWCA reduce initial shocks in springtime predictions. Further, the sustained ensemble consistency aided to simulate the variability better and improved the seasonal predictions. The study strongly advocates the adaptation of advanced CDA methods for seasonal monsoon and probable seamless predictions. Plain Language Summary Early prediction of sub‐seasonal to inter‐annual variations in Indian summer monsoon rainfall has multifaceted benefits (e.g., agriculture, economy, etc.). Hence any significant improvement in the prediction skill could be highly appreciated. The Ocean and Atmosphere observations are increased tremendously during the recent decades due to satellites and improved observational networks. Since monsoon prediction partly depends on the state of the Ocean and Atmosphere (i.e., model starting point or “analysis”), these observations can be used to improve the analysis, thereby the predictions. Even though sophisticated data assimilation techniques have been demonstrated to strengthen the analysis quality, their operational utilization in the context of monsoon prediction is still far from the reality due to the difficulty in adaptation and computational limitations. The recent improvements in high‐performance computing and data assimilation research under Monsoon Mission have aided us in implementing an advanced data assimilation method to the operational monsoon prediction model. Using this new analysis, the seasonal monsoon predictions improved. The present study reports the enhancements and attempts to explore probable mechanisms responsible for the improvement. The study is vital to operational agencies in adopting advanced data assimilation methods, particularly to boost monsoon p