The researchers used high-resolution imagery and various land factors to predict soil salinity in coastal wetlands. They focused on Yancheng National Nature Reserve, where Spartina alterniflora is invading and native vegetation is degrading due to high salinity. By analyzing factors like vegetation type and distance from the sea, they found that the random forest model was more accurate in predicting soil salinity. Over the years, soil salinity decreased, with the most important indicators being distance from the sea and vegetation type in predicting salinity levels in mudflats.