Projections of Active and Break Spells of the Indian Summer Monsoon using Original and Statistical Downscaled CMIP5 GCMs

Thursday, 18 December 2014
Swati Singh, Kaustubh Salvi, Subimal Ghosh and Subhankar Karmakar, Indian Institute of Technology Bombay, Mumbai, India
The dual nature of Indian monsoon system viz. south-west and the north-east is a potent advantage for Indian economy, which is mainly supported by agriculture industry. Out of these, the south-west monsoon contributes around 70% of the total annual rainfall over India and hence, forms the major source of water for agriculture. However, the south-west monsoon possesses variability on different temporal scales such as daily, intra-seasonal, annual and decadal time-scales. Among these, understanding of the intra-seasonal variability of south-west monsoon, which is characterized by spells of good/excess rainfall (active spells) and spells of less/no rainfall (break spells) over core monsoon zone in the peak monsoon months (July and August), is the most important for crop cycle planning. Long and intense active (break) spells may lead to flood (drought) and both these situations are crucial for the critical (initial) growth period of the crops and these situations may lead to reduced yield. So, accurate prediction of these events is essential for better agricultural planning. Therefore, we evaluate the ability of, General Circulation Models (GCMs) and statistically downscaled rainfall simulations at 0.5˚ resolution, to capture the intra-seasonal variability (in terms of active and break spells) revealed by observed rainfall data. We also project the duration and frequency of active and break spells with original and downscaled GCM simulations. The analysis is performed over spatially averaged rainfall time series (over core monsoon zone) for July and August, simulated with five GCMs simulations (both with original and downscaled) over historic period (1951-2005) and extreme future (RCP45, 2071-2100). Comparison results revealed that the GCMs simulations (original and downscaled) lack the skills to reproduce active/break spells shown by observed data. Both original and downscaled GCMs showed increase/decrease in active/break spells in future and changes are more in downscaled simulations. This is not consistent with observed data and hence needs critical evaluation.