A case study on the prediction of Indian summer monsoon of 1987 with the help of a global spectral model and sensitivity of the model to vertical resolution

Abstract

A Global spectral model has been integrated for 240 hours for each day of an active epoch during the monsoon season of 1987. The mean error, root mean square error ,the correlation coefficient between analysis and forecast changes and root mean square vector wind error have been computed for each of the seven initial conditions for 24 hours to 240 hours forecast segments. The rainfall forecasts have also been evaluated subjectively against the realised rainfall values. An experiment on sensitivity has also been carried out with the same model for two vertical resolutions. It is seen that forecasts produced by the lower resolution limited physics model is closer to the analysis produced by using the 6-hr forecasts of a higher resolution, better physics model as first guess. This implies that higher vertical resolution produces changes which are in opposite direction.to that produced by adding more physics.