Future rainfall scenario over Orissa with GCM projections by statistical downscaling

Abstract

The article presents a methodology for examining future rainfall scenario using fuzzy clustering technique from the General Circulation Model (GCM) projections. GCMs might capture large-scale circulation patterns and correctly model smoothly varying fields such as surface pressure, but it is extremely unlikely that these models properly reproduce non-smooth fields such as precipitation. The model developed in the present study is a linear regression model for estimation of rainfall, using GCM outputs of mean sea-level pressure and geopotential height as explanatory variables/regressors. To reduce the dimensionality of the dataset, the Principal Component Analysis (PCA) is used. Fuzzy clustering technique is applied to classify the principal components identified by the PCA and the fuzzy membership values are used in the regression model, with an assumption that the effects of circulation patterns on precipitation in different clusters are different. The regression model is then modified with an appropriate seasonality term. A major advantage of the proposed methodology is that while being computationally simple, it can model rainfall with a high goodness-of-fit (R²) value. The methodology is applied to forecast monthly rainfall over Orissa.