 This study proposes a new algorithm for producing land cover maps in Southern California based on remote-sensed images, which improves the accuracy of the baseline map by discarding potential bad pixels and dividing each land cover type into subclasses. The algorithm uses time series Landsat images to detect changed and unchanged areas between the baseline year and target year, and classifies each pixel accordingly. The results show that the land cover temporal pattern captures the observed successional stages of the ecosystem well, with overall classification accuracies ranging from 81% to 86% and overall kappa coefficients ranging from 0.79 to 0.83. This article was authored by Xingli Huang, Carlos Ramirez, Carmar Kennedy, and others.