A Comprehensive Review of Cloudburst Prediction Techniques Using Modern Computing: An IMD-Based Perspective

Abstract

Cloudbursts are severe destructive events characterised by extreme rainfall in short duration particularly in hilly and mountainous regions. Cloudbursts pose a significant threat in India, especially during the South-West Monsoon season in the duration from June to September. India has many different climate regions, such as the Himalayas, the Indo-Gangetic Plain, the southern peninsula, and coastal areas, and cloudbursts occur occasionally across these regions. However, only 31 cloudburst events have been officially recorded, most of them in Himachal Pradesh, Uttarakhand, and Jammu & Kashmir. Even with advancement in numerical weather prediction, accurate forecasting of cloudbursts events remains a major challenge due to limited observation strategies. Moreover, mini-cloudbursts which are more frequent but occur over small spatial area in very short duration also go undetected by the existing methods that are designed for cloudburst-scale event. In recent years, modern computing techniques such as Machine Learning (ML), Deep Learning (DL), and big data analytics are being used as promising tools for extreme rainfall prediction. This paper presents a comprehensive review of existing cloudburst prediction techniques while focusing on IMD based meteorological datasets. The critical analysis of Traditional Statistical Methods, ML models, DL architectures, and hybrid method approaches is performed. This paper proposes a conceptual framework for cloudburst prediction using modern computing. This review not only identifies key research gaps but also outlines future directions for developing more reliable, region-specific cloudburst early warning systems.