Prakiraan Curah Hujan Stasiun Sukadana dengan Teknik Statistical Downscaling Berdasarkan Data Satelit TRMM

Abstract

Partial least squares regression (PLSR) is used in statistical downscaling based on TRMM data. This research uses monthly rainfall of Sukadana station and TRMM 3B43 data which are nearest from Sukadana using a grid (four TRMM data) and four grids (nine TRMM data) during period 1998-2008. Residual of PLSR shows the variance is not homogenous. Weighted Least Square Regression (WLSR) and Principal Component Regression (PCR) are used to solve the residual's. Weighting in WLSR is based on standard deviation and PCR adds dummy variables based on rainfall group. Residual's variance of WLSR is not homogenous but PCR. Forecasting of PCR with dummy variables is better than PLSR's and WLSR's.