A Random Forest-Based Multi-Index Classification (RaFMIC) Approach to Mapping Three-Decadal Inundation Dynamics in Dryland Wetlands Using Google Earth Engine

Australian inland riparian wetlands located east of the Great Dividing Range exhibit unique, hydroecological characteristics. These flood-dependent aquatic systems in water-limited regions are declining rapidly due to competitive demand for water human activities, as well climate change and variability. However, there exist very few reliable data characterize inundation conditions quantify impacts loss deterioration wetlands. A long-term time record wetland maps can provide a crucial baseline monitor, assess, assist management conservation ecosystems. This study presents random forest-based multi-index classification algorithm (RaFMIC) on Google Earth Engine (GEE) platform efficiently construct temporally dense, three-decadal southeast The method was tested over Macquarie Marshes semiarid region NSW, Australia. results showed good accuracy when compared against high-spatial resolution imagery. total inundated area consistent with precipitation streamflow patterns, temporal dynamics vegetation agreement maps. analysed generate probability maps, which were frequently flooded areas simulated by hydrodynamic model distribution species. long-term, time-dense derived from RaFMIC key information assess condition health ecosystems have potential improve inventory spatially explicit regime information. be adapted other dryland wetlands, an effective semiautomated mapping dynamics.