Prediction of winter minimum temperature of Kolkata using statistical model

Abstract

Five parameter multiple linear regression model for objective forecasting of minimum temperature of Kolkata (Alipore) with 12 hours lead period has been developed. The predictors are chosen from the available surface data of Alipore observatory and low level wind data of M. O. Kolkata. Model has been developed from data sample comprising of 237 days (in January and February, period: 1997 – 2000) after stationarity test of minimum temperature data of much longer period (1988–2004). The model is tested with independent sample of 178 days. Efficiencies of the model have been tested with statistical skill score and found to be positive. The model can be used by the forecaster for assessing prediction minimum temperature and modify if cloud cover and wind flow pattern are expected to change subsequently.