The Indian summer (southwest) monsoon is referred to as the lifeline of India, as variability in any of its aspects (onset, withdrawal and quantum of rainfall) can greatly influence agricultural yield, economy, water resources, power generation and the entire ecosystem. If variations in monsoon rainfall are known well in advance, it would be possible to reduce the diverse impacts related to excess or deficient rainfall. A reliable monsoon forecast with sufficient lead time is essential for policy makers and farmers for planning and sowing of crops, as also making long-lead plans for the future.
Yet, the accurate prediction of monsoon rainfall remained a challenge for decades. The conventional forecast in use so far is based on the statistical approach and is low skilled in forecasting rainfall anomalies. Keeping this in mind, the National Monsoon Mission (NMM) was envisaged in 2012 by the Ministry of Earth Sciences (MoES) to develop a dynamical coupled prediction system specially suited for the Indian region that would show high predictive skills with an anomaly correlation of 0.6 and above.
The future goals of the NMM would be not only to improve the weather and climate forecast during the four monsoon months of June, July, August, September (JJAS) but to provide skilful prediction throughout the year at different space and time scales right up to the regional levels.
The long range prediction of the seasonal and extended range mean monsoon rainfall depends on the dynamics of its year-to-year variations. Recent improvements in dynamical numerical models with ocean-atmosphere coupling have been found to be useful for improvement of the monsoon forecast skill. Changes in sea surface temperature provide the required memory for predicting the Indian monsoon rainfall at longer lead times and therefore acts as a source of predictability on seasonal time scales. In view of this, it is essential to represent the global ocean-atmosphere coupled processes in the model. Thus, the coupled, ‘ocean atmosphere general circulation model’ is the befitting option for attempting seasonal forecast. As the rise or fall of ocean temperature and its atmospheric response occurs at a slow frequency, the coupled dynamical model forecast provides sufficient lead time for precision planning.
Genesis of the National Monsoon Mission
Forecasts at leading global weather/climate forecast outfits—National Centre of Environmental Prediction (NCEP), USA; Meteorological Department, UK; European Centre for Medium Range Weather Forecast, UK; etc., are done using very high resolution ensemble prediction systems. Extreme events and probability statistics are better resolved with these systems. In India, until recently, scientists were using deterministic low resolution models to make forecasts.
Short term climate predictions (seasonal forecasts and extended range forecasts) using high resolution global dynamical coupled models can provide better representation of spatial distribution of climate parameters at longer leads (more than a season in advance). As against this, in India, operational seasonal forecasts are basically dependent on empirical models with limited operational prediction skill. The nation could not setup these state-of-the-art prediction systems mainly due to lack of computational infrastructure and trained manpower as also the lack of concerted efforts within the country.
Considering the above and to accelerate the forecast accuracies for weather and climate predictions over India, the NMM envisages to:
- build a working partnership between the academic research and development organisations and the operational agency to improve the monsoon forecast skill;
- setup a dynamical modelling framework for improving prediction skill of seasonal and extended range prediction system, and short and medium range prediction system; and,
- setup the infrastructure and train manpower required to improve the prediction skill in all time scales.
The NMM attempts to establish an improved operational prediction system for monsoon in short, medium, extended and long range time scales. The NCEP’s Climate Forecast System (CFS) has been identified as the basic modelling system for the mission, as it is one of the best among the currently available coupled models. However, this model is limited in its ability for retrospective forecast (hindcast) of seasonal monsoon rainfall. Hence, there is an urgent need to develop an Indian model based on the CFS coupled model with an improved hindcast skill for operational forecasting. IITM has currently invited proposals from national and international scientists/organisations towards fulfilling this need.
The IITM, IMD, and NCMRWF have considered using the following numerical models:
- The American CFS model developed by the NCEP, NOAA National Weather Service, USA. CFS is a coupled ocean-atmosphere modelling system that combines data from ocean, atmosphere and land for providing long range forecasting (seasonal and extended range prediction of Indian monsoon);
- Model developments on CFS implemented by IITM, with atmospheric initial conditions from National Centre for Medium Range Weather Forecasting (NCMRWF) and ocean initial conditions from Indian National Center for Ocean Information Services (INCOIS); and,
- The Unified Model (UM), developed by the United Kingdom Meteorological Office (UKMO), UK, utilised for short to medium range prediction.
Significant amount of research and development work has been carried out by scientists at IITM during the last two to three years in improving the CFS version 2.0. As a consequence, the present skill of CFSv2 with a horizontal resolution T382 (~38 km) stands at 0.55 which is considered to be fairly high as per the present day standards and as compared to other global models. There is also a significant improvement in the extended range forecast for a period of 15-20 days using the CFS system. This forecast has been specially useful for agricultural and town planning, and more. The extended range prediction system has also been found to be skilful in case of extreme rainfall events such as the Uttarakhand episode.
IITM has also started generating an ensemble based high resolution T574 (~25 km) probabilistic short range forecast system using Global Forecast System (GFS) model in collaboration with NCMRWF, Noida, from June 2016. This initiative will add further value to the existing daily forecasts of the IMD. Upon augmentation of the high performance computer (HPC) capacity to 10 petaflops by 2018 under priority activities of the present government’s development agenda, ensemble based prediction system forecast products at 12 km grid scale shall be rolled out on an experimental basis to upscale the monsoon rainfall prediction skills.
Research has also been carried out to improve the representation of physical processes such as the cloud convective, and radiation and land surface processes in the model. All these developments in the CFS/GFS system have significantly contributed to improving the efficiency of the model.
Also, several young scientists were trained on the dynamical modelling framework that required HPC infrastructure to carry out different experiments.
Though the Mission has been successful in achieving its goal of setting up a state-of-the-art dynamical modelling framework with moderate predictions skill, concerted efforts are required to further improve it to make it more useful. Predicting extreme events and obtaining reasonable prediction skills at regional levels at longer leads still remains a challenge.