Monitoring of inter-annual variations in mangrove forests in the Bamen Bay area based on Google Earth Engine

Based on the Google Earth Engine (GEE) cloud platform and Landsat series data, this study classified the surface features of the Bamen Bay area using the support vector machine (SVM) classification method. Furthermore, the classification results were employed to monitor the inter-annual variations of mangrove forests in the area. The analysis reveals that mangrove forests and terrestrial trees exhibit extraordinarily similar reflectance spectral curves except for infrared bands. Hence, they were effectively distinguished using the infrared band feature index and topographic data, achieving an overall classification accuracy of 91%. The classification results show that mangrove forests in the study area manifested a trend of decrease followed by increase. Specifically, they decreased from 2009 to 2013, remained almost unchanged from 2014 to 2016, and increased slowly from 2017 to 2021. The increase in mangrove forests and the decrease in pits and ponds occurred following the wetland restoration policy that requires planting mangrove forests in South China and tamarix chinensis in North China, suggesting remarkable effects of the policy for returning ponds to forests. The transfer matrix analysis reveals a mutual transfer between mangrove forests and pits, ponds, suggesting that deforesting for ponds and returning ponds to forests constitute the primary factors influencing the variations in mangrove forests. The inter-annual variation monitoring results of mangrove forests enable detailed analysis of the evolutionary process of mangrove forests and accurate quantification of the transformation between mangrove forests and other land types. Therefore, the factors influencing mangrove forest evolution can be analyzed from the perspective of economy and policy for more effective preservation of mangrove forests.