To provide accurate land classification, we need to have a good and robust extimators. Minimum vector variance (MVV) is one of the robust method to produce them. In this paper, we implement the method to classify absorption land in Jakarta province. Our Experiment using 2002's satelite imahes of Jakarta area (band 1, 2, 3 and 4) has shown that MVV methods provides good classification of less tahn 5%. We also decreased the process time in order to occupy various types of land classification by enabling parallel technique on the solution. The experiment shows an average speedup almost two times of the sequential process of classyfing 2002's satelite images of Jakarta.