Seasonality and spatial pattern of rainfall of Sri Lanka: Exploratory factor analysis

摘要

This analysis is based on monthly means of rainfall at a dense network of gauging stations in Sri Lanka. The mean monthly values of rainfall at 646 stations were used as variables to characterise the individual stations. These variables show a significant correlation among most of them. The highest correlations were found between months within the same meteorological season, with one exception. The exception is that of October which has a higher correlation with months of southwest monsoon (SWM) than with the inter-monsoon (IM) months. The IM months and November have moderate values of correlation with the months of SWM. All three months of northeast monsoon (NEM) are strongly correlated and form a clearly defined group. This pattern of correlation can be explained in terms of the spatial distribution of rainfall of the 12 months. The strongly correlated months have a similar spatial pattern. This indicates that the number of distinct spatial modes of rainfall is less than 12. To discover these modes, principal component analysis (PCA) and factor analysis (FA) were applied on the data set. Of the two ordination methods, FA produced more easily interpretable results than PCA. The factor solution identified four spatiotemporal rainfall modes - weak southwest (SW) mode (March-April), strong SW mode (May-October), strong NEM mode (December-February) and mixed mode (November). These modes have strong similarity to the monthly rainfall surfaces created using the original data of the same periods.