Operational model for forecasting location specific quantitative precipitation and probability of precipitation over India

The National Centre for Medium Range Weather Forecasting was established in early 1988 with the major objective to develop operational medium range weather forecasting capability and agricultural meteorological advisory services (AAS) for each of the 127 agricultural climatic zones for the farming community in India. At present, medium range weather forecast of six surface weather parameters namely, average cloud cover, 24 h accumulated precipitation, average wind speed, predominant wind direction, maximum/minimum temperature trends (up to 4 days) is provided to 83 units in different agricultural climatic zones. In addition the forecast of weekly cumulative rainfall is also provided. An objective system for obtaining location specific forecast, in the medium range, of surface weather elements is evolved at NCMRWF. The basic information used for this is the output from the general circulation model (GCM). A T80L18 model operational at the centre since 1994 has been recently upgraded to a T170L28 model. However, it is well known that in spite of higher resolution, the global models are unable to account for the small-scale effects (e.g. of topography, local environmental features) important in predicting surface weather parameters like rainfall, temperature etc. This necessitates the development of statistical–dynamical models. Hence an operational system for forecasting rainfall (quantitative, probability of precipitation (PoP)) has been developed at the centre and implemented since 1994. A Perfect Prog Method (PPM) approach is followed for statistical interpretation (SI) of Numerical Weather Prediction (NWP) products. PPM model equations are developed by using analysis data obtained from European Centre for Medium Range Weather Forecasts (ECMWF) for a period of six years (1985–1990). Rainfall forecasts are subsequently obtained from these equations by using T80 model output. A comparative study of the skill of SI forecast and the direct model output (DMO) forecast has indicated that SI forecast improves over the DMO considerably and hence can be developed as a fully automatic operational weather forecasting system.