digilib@itb.ac.id +62 812 2508 8800

Erna Sri Adiningsih
PUBLIC Alice Diniarti

Principle Component Analysis (PCA) application in modeling the soil moisture estimate using multispectral satellite data is meant to optimize spectral combination. PCA method has been applied to Landsat Thematic Mapper (TM) satellite data with good results. However, Landsat data have low temporal resolution (16 days) compared with daily NOAA-AVHRR (NOAA-Advanced Very High Resolution Radiometer) satellite data. So, NOAH-AVHRR data are able to provide better information on daily soil moisture. The objective of the study is to develop soil moisture estimation model based on daily 5-channel daily NOAH-AVHRR data using PCA method. The locations are West Java and Central Java as case study, while the period is August-September 1999. Some field soil samples were also taken from the two locations. The coefficient of variance shows that the three principle component (PC) can explain the variance of soil moisture of 0-20 cm depth better than of >20 cm depth. This is due to more dynamic surface soil moisture change rather than deeper soil layer. Among the three PCs, the first PC is the best parameter- to estimate soil moisture. The index resulted by the first PC can estimate soil moisture better than vegetation index.