Forecasting seasonal mean temperature over Rangpur, Bangladesh

Abstract

The study was conducted by Climate Predictability Tools (CPT) to forecast (short-range forecast) the seasonal mean temperature over Rangpur for six Bengali seasons in Bangladesh. In this study, the sea surface temperature (SST) for the period of 1975 to the previous month of each season of 2008 was used as the predictor. This study also evaluated the difference between forecasted seasonal mean temperature and observed seasonal mean temperature for six seasons. To find the SST that is similar to the temperature in Rangpur, a correlation between the temperature of Rangpur and the sea surface temperature of various parts of the earth was performed through CPT using both data of 1975- 2008 years. The obtained SST through correlation that is more or less similar to the temperature in Rangpur was used as a predictor to forecast seasonal mean temperature of the year 2009. Statistical and mathematical methods were applied by CPT in this research which included canonical correlation analysis, covariance matrix, and eigenvalues equations.The study found that the forecasted seasonal mean temperature was higher in rainy and winter seasons than the temperature observed and was lower in summer, autumn, late autumn, and spring season than the observed temperature at Rangpur. The maximum overestimated temperature was found to be 0.52oC/day in winter and the maximum underestimated temperature was found to be 0.54oC/day in autumn. On the other hand, the minimum overestimated temperature was found during the rainy season having the value of 0.34oC/day and the minimum underestimated temperature was obtained during the summer season having the value of 0.25oC/day, which was the best-forecasted temperature. Therefore, the forecasted values of temperature in the summer and rainy seasons were found closer to the observed temperature during 2009. So, it can be said that it is possible to obtain good forecasting of temperature through CPT.