Soft Computing: Inferential Statistics of 3d Rainfall- Runoff Modelling in Peninsula Malaysia

Thorough understanding of the rainfall-runoff processes that influence watershed hydrological response is important and can be incorporated into the planning and management of watershed resources. Soft computing techniques and inferential statistics were used to assess 2 rainfall-runoff models and their runoff predictive accuracy in this article. The 1954 simplified SCS runoff model was found to be statistically in-significant under two Null hypotheses rejection and paved way for the model calibration study to produce regional specific runoff model through calibration according to regional hydrological conditions in Peninsula Malaysia. The new runoff model out-performed non-calibrated SCS runoff model and reduced its RSS by 27%. A 3D runoff difference model was created as a collective visual representation between the (SCS) non-calibrated and calibrated new model, it also showed that both under and over design risks were less significant at high CN (urban) area and more profound under higher rainfall depths. On average, rural and forest catchments of Peninsula Malaysia faced 7% (lower CN area as much as 22%) CN down scaling adjustment due to regional hydrological calibration in order to achieve better runoff predictions.