Rainfall is one of the microclimatic variables that vary with space. The changes in vegetation characteristics may influence the microclimate elements. To demonstrate rainfall variation due to vegetation, the relationship between rainfall and vegetation should be spatially investigated over a local scale. This paper aims to explore the impact of vegetation on local variations of rainfall based on Geographically Weighted Regression (GWR) approach. The global and local relationship between rainfall and the extracted Normalized Difference Vegetation Index (NDVI) of Landsat 7 ETM+ are quantitatively estimated in 2000 and 2011 within the northern and east coast regions of the Peninsular Malaysia. Based on 277 rainfall stations, the Moran's Index (Moran's I) spatial autocorrelation and Ordinary Least Square (OLS) - GWR methods were applied to analyse the rainfall spatial patterns and to determine rainfall spatial variation, respectively. It was found that, the rainfall spatial patterns exhibit small clustering patterns which leads to non-stationarity. This indicator supports the use of local regression approach in exploring the variation of rainfall due to vegetation. The R-Squared (R2) from GWR (0.51 and 0.75) significantly improved the R2 from OLS (0.01 and 0.04) for both years. The approach of GWR in the relationship between rainfall and vegetation provides findings on rainfall spatial variation on a local scale.