Mangrove forests have important ecological functions in protecting the environment. However, the mangrove wetlands have been largely lost because of intensified human activities in the study area. Remote sensing can be conveniently used for the inventory of mangrove forests because field investigation is very difficult. In this study, a knowledge-based system is developed to retrieve spatio-temporal dynamics of mangrove wetlands using multi-temporal remote sensing data. Radar remote sensing data are also used to provide complementary information for the quantitative analysis of mangrove wetlands. Radar remote sensing is able to penetrate mangrove forests and obtain the trunk information about mangrove structures. The integration of radar remote sensing with optical remote sensing can significantly improve the accuracies of classifying mangrove wetland types and estimating wetland biomass.