Estimasi curah hujan berdasarkan data CMORPH (CPC MORPHing Technique) wilayah Riau
Abstract
Rainfall is the most important climate element in Indonesia which has a very large variation compared with other climate elements. High variability will lead to limited rainfall data. The accurate data and information will support the development of rainfall prediction using remote sensing. CMORPH (CPC MORPHing Technique) is one of the new techniques developed by NOAA in predicting the amount of rainfall and global precipitation with high spatial and temporal resolution. Therefore, it is necessary to do a statistical downscaling technique approach based on partial least squares analysis (PLS). This research aims to analyze the potential for rainfall prediction data CMORPH surface.The model development is based on allegations precipitation method Partial Least Squares (PLS) in Pekanbaru, Japura Rengat, Tanjung Pinang, and Dabo Singkep. The research area is divided into five domains; domain 1x1, domain 3x3, domain 5x5, domain 7x7, and domain 9x9. The reliability model is shown by the allegations of the correlation coefficient value, Root Mean Square Error (RMSE), and Pearson test. Rainfall CMORPH is similar to the pattern of surface rainfall variation able to represent 10.7% to 47.5% variability of rainfall. Downscaling using Partial Least Ssquare technique (PLS) is better than downscaling techniques using simple regression, where using PLS can increase the determination coefficient value (R2) from 47.5% by using simple regression to 98.7% by using PLS. Based on model validation, precipitation allegations is similar to the pattern of rainfall area, except in the wet months. If observed from the region it can be seen that for Earth's surface a domain that provides the best result is the domain 5x5. Meanwhile, for the Islands, domain 3x3 offers the best performance. This is due to conditions in the region, where the Islands is influenced by the ocean. Therefore, rainfall CMORPH is better used to predict the precipitation on the continent surface.