Accurately obtaining information on seasonal wetlands not only helps improve the interdisciplinary wetland science system but also provides scientific basis for wetland restoration projects， which is of great significance for protecting wetlands， stabilizing the Earth's climate， and more. In this study， Landsat remote sensing images were used with the Google Earth Engine cloud platform to automatically extract detailed seasonal wetland information for the past 30 years in the East Dongting Lake area. Firstly， based on the typical characteristics of seasonal water-land alternation in East Dongting Lake wetlands， the random forest classification algorithm was used to obtain the range of seasonal wetlands. Then， a decision tree model was constructed to extract the subcategories of seasonal wetlands. The results show that：（1）the research can effectively extract wetland information， with an average overall classification accuracy and Kappa coefficient of 88.25% and 0.86 for seasonal wetland classification， and 93.28% and 0.92 for subcategories classification of seasonal wetlands. （2） From 1989 to 2020， the area of East Dongting Lake wetlands showed an increasing-decreasing-increasing trend， with a total increase of 154.02 km² in permanent water bodies and a decrease of 54.11 km² in seasonal wetland area. The seasonal wetland information extraction method developed in this study based on Landsat long time-series images can provide technical support for related research and decision-making on the dynamic changes of seasonal wetlands.