Development of Hybrid Prediction System for Indonesia Precipitation

Abstract

Research on a method development of precipitation prediction system over Indonesia was conducted through a combination of three different methods. This hybrid type of method is applied to predict the land precipitation over Indonesia. It is predicted based on Step-wise Pattern Projection Method (SPPM) from the observation data, and the CFSv2 prediction is also used as the second prediction method. The third method which aims to correct the CFSv2 precipitation prediction is conducted by simply applying the CFSv2 prediction to the SPPM. Each method shows the different features of prediction skill in seasonal time-scales based on 1-month to 6-month time-lag prediction. The difference feature makes the prediction skill even better after combined as one. As the main objective of this research is to improve the precipitation prediction skill over Indonesia, it is improved for both wet and dry season compare to CFSv2 prediction. Although, the dry season improvement is considerably small but a notable improvement showed in the wet season. The wet season prediction skill improvement is related to the SST anomaly over Western North Pacific (WNP). Particularly, the Pacific Sea SST is the main contributor for February precipitation prediction due to a convergence zone in results of its warm SST. The dry season that seems to has no significant improvement brings a newly found precursor with a well explained dynamical process. A new reliable precursor is found over South Atlantic SST which related to the August prediction skill improvement of the Indonesia precipitation.