The high spatial resolution of precipitation distribution is a major concern for experts in environmental research and planning. This paper establishes a combination of multivariate regression algorithm and spatial analysis to predict distribution of precipitation, considering the four topographical factors of altitude, slope, aspect and location. Annual average and seasonal rainfall data were collected in nine rain gauges in Ulu Kinta Catchment in East Malaysia from 1974 to 2010. To examine records and fill gaps from long-term rain gauges, homogeneity analysis was performed using the double-mass curve method. Estimated missing rainfall data were also tested using index gauges from network rainfall stations. Multivariate regression analysis was conducted to propose an empirical equation for the study area. Topographical factors were considered from a 90 m resolution digital elevation model. The multivariate regression model was found to clarify 74% of spatial variability of precipitation on annual average and 78% during wet season. However, the correlation coefficient for the dry season decreased sharply to 63%. By using the kriging interpolation method, the estimated annual average improved to 78.4%; the average improved to 65.2 and 80.3% in the dry and wet seasons, respectively. This confirms the efficiency and significance of the model and its potential for use in other tropical catchments.

This content is only available as a PDF.