An analysis of statistical correlation for downscaled precipitation data

摘要

Among the few climate-related studies done in Brunei Darussalam, it is observed that previous research in related to the prediction of climate change scenarios using statistical downscaling method encounter difficulties in developing strong statistical correlations between precipitation and atmospheric parameters of Global Circulation Models (GCM). The statistical correlation between the observed precipitation (predictand) and the predictors were analysed that examined the correlation of 1, 2 and 3 days lagged precipitations and correlation with the data of surrounding GCM grids. Also, precipitation data were correlated with atmospheric data with and without rainfall (drydays). The results showed no strong statistical correlation exists between observed data with the reanalysed GCM data. The correlation coefficient between preditands with the predictors of National Centres for Environmental Prediction (NCEP) achieved a highest value of 0.18 and such data with high correlations was selected for rainfall prediction in Brunei Darussalam using multiple regressions.