Stochastic modelling of dust emission sources probability based on satellite observations in a South American endorheic basin

Abstract

Northeastern Patagonia is a recognized vulnerable region for wind erosion and desertification. Identifying dust emission sources in the coastal regions of Argentine Patagonia is crucial to understanding micronutrient transport to the Southern Atlantic Ocean. This study proposes an innovative integrated approach combining species distribution modelling and optical satellite images of the land surface to identify potential dust emission hotspots related to wind erosion. To test the approach, we selected a study area in an endorheic basin in the southwestern Buenos Aires province, Argentina. We identified the source of particulate matter emission using MODIS Aerosol Optical Depth (AOD-MAIAC) values above a threshold of 0.06, which was validated by fieldwork and served as the dependent variable in the model. As explanatory variables, we derived several satellite-based products such as the Aridity Index, and Soil Organic Carbon, as well as topographic indices like Vertical Distance to the Drainage Network and land cover. The applied MaxEnt model shows an acceptable performance with Receiver Operating Characteristic curve integrals of 0.75. For further validation, the trained model was transferred to a neighboring region. The most significant variables were Soil Organic Carbon, Vertical Distance to the Drainage Network, and Landcover. The results indicate that the most susceptible areas generating dust/ salt plumes in the studied endorheic basin are the depressions and lowlands where fine sediments accumulate. The results yield valuable information for the proper management of Eastern Argentinian lowland areas and support general soil and water conservation practices and planning.