Efficiency of monsoon mission climate forecast system (CFSv2-T382) to predict the large-scale and regional short-term drought over India during boreal summer and its possible controlling factors

The present study assesses the ability of Monsoon Mission CFSv2-T382 initialized at 4-month and 2-month leads (Feb IC and Apr IC hindcast) to simulate the short-term drought during the Indian Summer Monsoon (June to September). The rainfall-based standardized precipitation index (SPI) averaged over the Indian land region showed significant skills for the 1981–2017 hindcast period for both the ICs. However, the spatial distribution of the skill is moderate and is limited to some regions, and varies within the season. Category-wise forecasts indicate that even with strong rainfall bias and moderate false alarm ratio, model hindcasts have a better detection ratio, hit rate, etc., demonstrating the model’s usefulness for predicting the all-India drought. The observed SPI over the entire India during the month of June has a strong positive (negative) relationship with soil moisture (surface temperature) over India, along with warm (cold) SST anomalies in the northern Indian Ocean (north tropical Atlantic Ocean). The SST anomalies associated with El Nino Southern Oscillation (ENSO) dominate the drought towards the end of the monsoon season. However, SPI simulated by model hindcasts has a stronger relationship with ENSO throughout the season. The homogenous region-wise skill analysis and teleconnections of SPI explain its scattered skill over Indian land regions. The observed regional surface temperature and soil moisture have a strong 1–2-month lead relationship with SPI in all homogenous regions, and forcing from ENSO is evident for WCI and SPInd only. The model overestimates the simultaneous ENSO relationships for all the homogenous regions, while the lead role of moisture, surface temperature, and north tropical Atlantic SST is under-predicted. WCI and SPInd regions have a moderate lead relationship with soil moisture and surface temperature along with ENSO teleconnection, resulting in improved skill compared to other subdivisions of India. The study shows that further improvement of the regional scale skill of CFSv2 for capturing SPI is possible mainly with the regional scale improvement of land surface processes and its lead teleconnection with SPI in the model.