Impact of model resolution in the Extended Range Prediction of Indian summer monsoon and 2013 North Indian Heavy Rainfall Event

A quantitative comparison of the extended range (~2–3 weeks) forecasts of monsoon intraseasonal oscillations (MISO) obtained from the Climate Forecast System (CFS) model version 2 developed at National Centre for Environmental Prediction USA at two different resolutions: T126 (~100 km) and T382 (~38 km) has been made in this study. It is observed that, higher model resolution (CFS382) has provided better basic state for MISO along with large reduction in climatological biases in June–September precipitation than the lower resolution forecast (CFS126). However, the reduction of climatological biases in the high resolution model are not necessarily translated into improvements in real-time forecast of rainfall over the homogeneous regions of India, or over the broader regions of Indian subcontinent. The lower resolution runs at T126 resolution has the similar potential to give operational ER forecasts as compared to CFS382. The area averaged forecast and the temporal evolution of MISO in the extended range show statistically similar skill in CFS382 as well as CFS126 run. Thus the forecast skill of large-scale MISO may not necessarily get improved with increased resolution. In this study, we assess whether the improved climatology in the high resolution version is useful in the prediction of extreme rainfall events. The Indian summer monsoon of 2013 covered the entire country by 16 June, one month earlier than its normal date. Around that period, heavy rainfall was experienced in the north Indian state of Uttarakhand, which is situated on the southern slope of Himalayan Ranges. The heavy rainfall and associated landslides caused serious damages and claimed many lives. In this study, we further investigate the scientific rationale behind the incidence of the extreme rainfall event as well as its extended range prediction in the backdrop of large scale monsoon environment. It is found that a monsoonal low pressure system that provided increased low level convergence and abundant moisture, and a midlatitude westerly trough that generated strong upper level divergence, interacted with each other and helped monsoon to cover the entire country and facilitated the occurrence of the heavy rainfall event in the orographic region. It is noticed that both CFS126 and CFS382 could predict the extreme event 10-12 days in advance, with CFS382 performing better in predicting the magnitude of rainfall. The formation and northwestward movement of the low pressure system over Bay of Bengal is also better predicted by CFS382. However, both the models failed to predict the extratropical influence on the event, which could be attributed to the inherent cold temperature bias in the CFS model. It is revealed that the anomalous moist flow of monsoonal surges (which are associated with the low pressure system) to the mountainous region generated the extreme event in the models, especially in CFS382 (evident from the stronger ascending motion over the region). If the models could have predicted the midlatitude influence reasonably, they could have predicted the amount of rainfall closer to the observations. The study advocates the usefulness of high resolution models in predicting extreme events.