IMD Forecast For Monsoon rainfall (June-September 2000)

2.1 Operational Models

Since 1988, IMD has been issuing long range forecast for the monsoon rainfall over the country as a whole using the 16 parameter power regression model which was operationalised by the IMD in 1988. Recently this 16 parameter model was revised by replacing the 4 predictors (North India Temperature, 10 hPa zonal wind at Balboa, 500 hPa Ridge position and Darwin pressure (Spring) whose relationships with monsoon rainfall has declined in recent years) with 4 new predictors (Arabian Sea SST, South Indian Ocean SST, Europe Pressure Gradient and Darwin Pressure Tendency) which have shown significant relationship with the monsoon. It has been found that the performance of the new 16 Parameter Power Regression Model is better than the old model. Therefore, this year, official long range forecast for the country as a whole is issued based on this new 16 parameter (12 old + 4 new) Power regression and Parametric models.

2.1.1. Parametric Model

This model provides a qualitative forecast and utilizes the signals from 16 antecedent global and regional land-ocean-atmospheric parameters. Analysis of the signals from these parameters indicates that this year, 63% of them are favourable for a normal monsoon. The seasonal rainfall for the country as a whole is thus expected to be normal (defined as ± 10% of the long period average value)

2.1.2 Power Regression Model

This model provides quantitative forecast and is based on the physical relationship of monsoon rainfall with 16 different individual parameters. According to this model, the total amount of rainfall over the country as a whole during the monsoon season from June to September 2000 is likely to be 99% of its Long Period Average (LPA) with an estimated model error of ± 4%.

2.2 Experimental Models

Five different types of experimental models viz., Multiple Regression model with 6 parameters. Dynamic Stochastic Transfer Model with east coast India temperature and climatic forcings as inputs. Power Transfer model with 9 parameters, Principal Component Regression model with 10 parameters and Artificial Neural Network with 10 parameters also have been used. The forecasts obtained from different models are 93% of LPA from the multiple regression model, 103% of LPA from the DST model, 96% of LPA from the Power Transfer Model, 104% of LPA from the Principal Component Model and 103% of LPA from the Neural Network Model.

2.3 Summary of the Long range forecast for 2000 monsoon over the country as a Whole

The long range forecast for the 2000 Southwest Monsoon formulated on the basis of Parametric and Power Regression Model is given below :

a) In 2000, the Southwest Monsoon rainfall over the country is likely to be normal, (the normal being defined as rainfall within ± 10% of its long period average value) making 2000 the twelfth normal monsoon year in succession.

b) Quantitatively, the rainfall for the country as a whole for the entire monsoon season (June to September) is likely to be 99% of its long period average value with the model error limit of ± 4%.

3. FORECASTS FOR MONSOON RAINFALL OVER THE HOMOGENEOUS REGIONS OF INDIA.

Last year, IMD issued for the first time regionwise long range forecasts for three homogeneous regions of the country viz., Northwest India, Peninsula and Northwest India. This year also IMD issued long range forecasts for these three homogeneous regions of India.

3.1 Forecasts for Northwest India.

For the purpose of this forecast, Northwest India is taken to consist of Uttar Pradesh, Haryana, Chandigarh and Delhi, Punjab, Himachal Pradesh, Jammu & Kashmir and Rajasthan.

Four different types of models viz., Power Regression with 8 parameters, Multiple Regression with 6 parameters, Principal Component Regression with 10 parameters and Neural Network with 10 parameters have been used. The forecast obtained from different models are 102% of LPA from the Power Regression model, 109% of LPA from the Multiple Regression model. 107% of LPA from the Principal Component Regression Model and 107% of LPA from the Neural Network Model.

The performance of the Power Regression model is better than other models particularly in recent years. Thus based on the Power Regression Model, the monsoon rainfall over Northwest India for the entire monsoon season (June to September), 2000 is likely to be 102% of its long period average.

3.2 Forecasts for Peninsula.

For the purpose of this forecast, Peninsula is taken to consist of Gujarat, Madhya Pradesh, Maharashtra, Andhra Pradesh, Karnataka, Kerala, Tamil Nadu and Lakshadweep.

Five different types of models viz., Power Regression with 9 parameters, Multiple Regression with 7 parameters, Dynamic Stochastic Transfer with East Coast India temperature and climatic forcings as inputs, and Neural Network with 10 parametes have been used. The forecasts obtained from different models are 100% of LPA from the Power Regression model, 99% of LPA from the Multiple Regression model, 100% of LPA from the Dynamic Stochastic Transfer model, and 100% of LPA from the Neural Network Model.

The performance of the Power Regression model is better than other models particularly in recent years. Thus based on the Power Regression Model, the monsoon rainfall over Northeast India for the entire monsoon season (June to September), 2000 is likely to be 100%$ of its long period average.

4. SUMMARY OF THE FORECASTS

The long range forecasts for the 2000 Southwest Monsoon over the country as a whole and the three broad homogeneous regions of India, viz., NWS India, Peninsula and NE India are given below.

a) In 2000, the rainfall over the country as a whole for the entire southwest monsoon season (June to September) is likely to be normal thus making the year 2000 twelfth normal monsoon year in succession. The normal is defined as rainfall within ± 10% of its long period average.

Quantitatively, the rainfall over the country as a whole for the entire southwest monsoon season is likely to be 99% of its long period average with an estimated model error of ±4%.

b) Over the broad homogeneous regions of India, the rainfall for the entire southwest monsoon season of 2000 is likely to be 102% of its long period average over NW India, 98% of the LPA over Peninsula and 100% of the LPA over NE India with an estimated model error of 8%.