Real Time Extended Range Prediction of 2013 Indian Summer Monsoon: An Assessment

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 8 January 2015
N. Borah, Indian Institute of Tropical Meteorology, Pune, Maharashtra, India; and A. Sahai, S. Abhilash, R. Chattopadhyay, S. Joseph, S. Sharmila, and A. Kumar

The Indian Institute of Tropical Meteorology (IITM), India has been disseminating the operational real-time extended range (3-4 pentad lead) prediction of Indian summer monsoon (ISM) since 2011, under the National Monsoon Mission Project, initiated by Govt. of India, in Indo-US collaboration. For this, an Ensemble Prediction System (EPS) has been indigenously developed at IITM based on the Climate Forecast System model version 2 (CFSv2) developed at National Centre for Environmental Prediction (NCEP), USA. In this study, we discuss several aspects of real time forecast of the ISM of 2013 to explore the operational capability of the EPS. 2013 summer monsoon was a near excess monsoon year in terms of seasonal mean and was unique due to the rapid progression of ISM over Indian region, incidence of extreme rainfall events, distinct and large amplitude of northward propagation of active and break phases, revival of monsoon at the time of withdrawal etc. For the 2013 monsoon season, IITM had generated and issued real-time ER forecast every 5 day, starting from 16 May initial condition and continued up to 28 September, using CFSv2 and the bias corrected CFS SST forced atmospheric component GFSv2 (GFSbc) at T126 resolution. The main objective of this paper is to document the salient forecast features during 2013 monsoon. A set of 11 (21) atmospheric initial conditions were created for CFSv2 (GFSbc) by perturbing the wind, temperature and moisture field at all vertical levels and more details on the perturbation method can be obtained from (Abhilash et al., 2014). An ensemble of 11 (21) members integration was performed for CFSv2 (GFSbc) for each start date and for the next 45 (25) days period. It is noticed that both, CFSv2 and GFSbc, models were able to predict the progression of ISM over the Indian region and the subsequent intraseasonal oscillations (active and break phases). The analysis for an extreme event (Uttarakhand flood) and monsoon revival towards the end of the season were also performed. Comparison between the two runs shows that active and break spells were predicted with good fidelity over the Indian region, though GFSbc outperforms CFSv2 on several occasions. Overall, the 3rd and 4th pentad lead time forecast shows a promise in capturing the intraseasonal variability of ISM. The probabilistic forecasts have the potential to give a reasonable outlook for the stakeholders. The forecast obtained from GFSv2 with bias corrected SST (i.e. GFSbc) shows a lot of promise and highlights the most important source of bias in ERP. This bias in CFSv2 SST is needed to be corrected for a better prediction of ERP of MISO. If SST bias is reduced, as shown using the GFSbc experiments, the forecast skill of CFSv2 can be improved. As study of Sharmila et al. (2013) suggests, the simulation of proper phase propagation of MISO requires a realistic air-sea interaction, this paper shows that for improvement of prediction using CFSv2 the phase of MISO may be better forecasted using a realistic mean state of SST. This realistic mean state of SST largely reduces the forecast bias and forecasts from GFSbc become comparable to that from CFSv2. Thus, this case-study for the year 2013 highlights the role of mean state bias that is crucial for realistic simulation as well as prediction of localized rainfall and MISO over the Indian region. This study may provide lead to an operational forecaster in analyzing and understanding the different events of monsoon season in NCEP-CFS/GFS framework. Finally, we wish to mention that this study corrects the mean bias on daily scale. With the aim to improve the dynamical models' forecast skill over the Indian region, effort will continue to correct the bias online as the ocean model runs and gets coupled to the atmospheric model. In future, a Multi-model-Ensemble system will be implemented for the real time prediction of ISM. This will be reported in a later study. References: Abhilash, S., Sahai, A.K., Pattnaik, S., Goswami, B.N., Kumar, A., 2014. Extended range prediction of active-break spells of Indian summer monsoon rainfall using an ensemble prediction system in NCEP Climate Forecast System. Int. J. Climatol. 34, 98–113. doi:10.1002/joc.3668 Sharmila, S., Pillai, P.A., Joseph, S., Roxy, M., Krishna, R.P.M., Chattopadhyay, R., Abhilash, S., Sahai, A.K., Goswami, B.N., 2013. Role of ocean–atmosphere interaction on northward propagation of Indian summer monsoon intra-seasonal oscillations (MISO). Clim. Dyn. 41, 1651–1669. doi:10.1007/s00382-013-1854-1