Modified mean field bias and local bias for improvement bias corrected satellite rainfall estimates

Abstract

Corrected bias is often used to improve satellite rainfall estimates. The fastest corrected bias methods are mean field bias (MFB) and local bias (LB). Nevertheless, using the ratio between rainfalls observed and satellite rainfall estimates such as TRMM neglects no rain conditions. Whereas zero rainfall often happens in the tropical maritime region. The aim of this study focuses on improvement of correcting satellite rainfall estimates in using the ratio of MFB and LB. Modified MFB is done by classifying the ratio, then multiplied it to the pixel of TRMM rainfall estimates. While, classified the ratio of local bias is done before interpolated the ratios uses inverse distance methods. Implementation of this treatment uses rainfall data in surrounding of the Makassar Strait. For avoiding of failure of a ratio in zero rainfall observed, 1 mm is added to the rainfall data. Evaluation of this treatment is assessed by root mean square (RMSE), mean absolute error (MAE) and correlation. The result shows that performance modified local bias (LB) can improve RMSE and MAE. Based on value of correlation, modified LB with 20 classes can increase correlation than other methods except conditional merging (CM). Although LB is better methods than MFB in RMSE, but it is worse than CM. Moreover, modified LB can be considered as the best correction method for satellite rainfall estimates because of the stabilization of MAE. This modified, affirm assumed that the persistence of rainfall event or not, have an effect of satellite rainfall estimate performance.