Land degradation modeling of dust storm sources using MODIS and meteorological time series data

Land degradation affects environmental integrity and threatens sustainable development worldwide by contributing to a number of social, economic, and ecological problems. This study uses time series of MODIS products, meteorological data, and ground truth map from dust storm sources to assess the spatial-temporal behavior of land degradation through a Random Forest algorithm in the southwest of Iran. Spatial-temporal variations in land surface biophysical properties of dust sources and surrounding areas were modeled using different time-series data on a 16-days, monthly, seasonal, annual, and 20-year basis from 2000 to 2019. Land degradation was estimated using the Random Forest algorithm. Dust storm sources were used to evaluate land degradation estimates. Furthermore, sensitivity analysis was carried out to identify the most influential biophysical variables. The results revealed that (i) 16-day time series data performs best in assessing land degradation; (ii) Apparent Thermal Inertia (ATI) was identified as the most influential variable for land degradation assessment; and (iii) there is a marked tendency towards land degradation in areas with severe land cover changes, mainly change from wetland to other classes.