Status of monsoon forecasting in India – A scientific correspondence in Current Science

 The article dealt with the prediction of Indian summer monsoon rainfall by five atmospheric general circulation models (AGCMs) during 1985–2004.

Recognizing the need for mathematical modeling-enabled, forecast-centred research in dynamical long-range forecasting in India, the CSIR Centre for Mathematical Modeling and Computer Simulation (C-MMACS), Bangalore initiated an effort to bring long-range dynamical forecasting to India in early 2000. The present note by Goswami in the main deals with his objections to Gadgil and Srinivasan on the grounds that their reference to the C-MMACS forecasts and the methodology is erroneous and improper. The response by Gadgil and Srinivasan to this is also included at the end of the note.

The note by Goswami puts forward the following simple steps to ensure that the efforts put into SPIM project are not a complete waste -

• Define evaluation parameters relevant for seasonal forecasting even if they are challenging.
• Use the simulations by the models utilized in the respective platform as control experiments.
• The simulations on the CDAC platform then provide the (corresponding) test experiments.
• Use test–control to determine the degree of variability introduced by a change of platform; compare this variability to the one introduced by change of initial conditions (ensemble standard deviation) with respect to the selected evaluation parameters. Although the sample size is small, the results will provide some insight.

Goswami notes that in view of many omissions and technical errors in the article by Gadgil and Srinivasan, it is necessary to carry out an objective, scientific and in-depth study, by an agency like India Meteorology Department (IMD), of methodology and status on seasonal forecasting of monsoon in India.

The response by Gadgil and Srinivasan touches upon the following points -

• The SPIM project whose results were reported in the article, assessed the skill of prediction of atmospheric models used in the country for generating monthly/seasonal predictions. A major comment that has been made by Goswami is with regard to the omission of references to papers from C-MMACS. It is important to note that the article was not a review of the dynamical forecasting work done in India or elsewhere, and hence only references pertinent to the theme of the article were included.
• The methodology used is an internationally accepted technique of generating retrospective forecasts by running all the models with identical initial and boundary conditions for several years. Furthermore, the project envisaged that the models must be run on the same computational platform to avoid differences arising from different platforms.
• It has been shown that such an assessment of atmospheric models can lead to insight into the nature of the deficiencies and hence suggest directions for research for improvement of the models. However, it is clear that work has to be done towards improving the model to generate better predictions.

The original paper on Seasonal Seasonal prediction of the Indian monsoon – An assessment using atmospheric general circulation models (AGCMs), Current Science can be viewed here

Download the scientific correspondence here –