The expansion of human activities and environmental changes in coastal desert and semi-desert ecosystems significantly impact habitat quality. This study used remote sensing data on the Google Earth Engine (GEE) platform to generate layers representing threats to mangrove ecosystems, including human construction, runoff, maximum daily temperature, wind speed, access roads, population density, drought severity, NO₂, SO₂, land surface temperature, soil moisture deficiency, actual evapotranspiration, and dust. These layers were analyzed for current and future conditions, and a target layer for mangrove change detection was created using a variable sampling method and the Random Forest algorithm for both time periods. The Mangrove Vegetation Index (MVI) and Enhanced Mangrove Vegetation Index (EMVI) were included as auxiliary data. The resulting data were analyzed using the InVEST model to generate maps of habitat quality and mangrove degradation, and habitat quality was further assessed using machine learning models. The results showed that Linear Trend, Neural Network and SVR models performed better than other machine learning models. Also, in the study area, habitat quality will decrease according to the Otsu threshold in the future years. The results reveal that, under current conditions, 35.43% of the mangrove forest area is classified as experiencing low-intensity degradation. Projections indicate that, under future conditions, 33.33% of the area will shift to high-intensity degradation, especially in southern regions near Qeshm Island. Habitat quality is also to decline, with the “good and suitable” habitat quality category expected to decrease by 3.52%. Additionally, human activity indices in medium and high categories are projected to increase by 5.38 km² and 4.48 km², respectively. These results can be used to guide future coastal management and conservation strategies by identifying areas at highest risk of mangrove degradation and declining habitat quality under increasing human pressure and environmental change.
Kazemi et al. (Tue,) studied this question.