本研究以北部湾茅尾海滨海湿地为研究区采用随机森林算法对北部湾茅尾海滨海湿地信息进行提取旨在选择最佳方案改善湿地信息提取的效果。通过对比不同特征变量的影响和随机森林模型的性能选出最优的特征提取方案和分类模型提高湿地信息提取的准确度和可靠性为其他湿地信息提取和生态环境监测提供参考和借鉴。结果表明：1提高湿地信息提取的关键是有效地使用多种特征变量对于不同特征对湿地信息提取的贡献率而言红边指数植被指数和水

This study focuses on the Beibu Gulf Maoweidao coastal wetland as the research area, using the random forest algorithm to extract information from the wetland, aimed at selecting the optimal solution to improve the accuracy of wetland information extraction. By comparing the effects of different feature variables and the performance of the random forest model, the optimal feature extraction scheme and classification model were selected to improve the accuracy and reliability of wetland information extraction, providing reference and guidance for other wetland information extraction and ecological environment monitoring. The results showed that: (1) the key to improving wetland information extraction is to effectively use multiple feature variables, and in terms of the contribution rate of different features to wetland information extraction, the red edge index > vegetation index and water body index > spectral features > terrain features; (2) the feature variable extraction effect based on the optimized random forest algorithm was the best, with an overall accuracy of 84.53% and a kappa coefficient of 0.82, indicating that the random forest algorithm can effectively perform feature selection, while still ensuring the accuracy of wetland information extraction and improving operational efficiency. This study provides a new approach, method and technical means for wetland information extraction in terms of data source selection, feature selection and method selection