Dust is a type of natural disaster that in recent years has caused a lot of damage to nature in the first place and secondly to human life and various facilities. The first areas to be affected by dust are those that are closer to its source. Therefore, the first step to better manage and reduce its destructive effects is to identify these areas. Identifying active dust hotspots can be difficult for a variety of reasons, including accessibility, cost, time, and related factors. Therefore, knowing the remote sensing data and the capability of forecasting models such as Maxent, increases the speed of corrective operations of the relevant executive departments. Prediction of dust hotspots, like other research in Maxent modeling, is defined based on the relationship between presence data and environmental parameters. In order to eva‎luate the capability of Maxent model in determining the dust centers of Isfahan province from 890 presence data that were randomly extracted from the existing desertification map of the province in GIS environment, along with 14 effective environmental layers related to 2016 including altitude, precipitation, evaporation, net primary production, aspect, Rigi plain (desert pavement), land surface temperature, wind speed, lithology, normalized difference vegetation index, temperature and vegetation aridity index, slope, land use and soil type based on DEM map data of the province, PERSIANN satellite rainfall data, weather stations data, 1.250000 litology and 1.250000 soil texture maps, MODIS sensor data,1/100000 land use and land cover map of the province were used. The modeling results showed that the AUC value was equal to 0.88, which indicates the proper performance of this model in predicting dust centers. Also, the effect of all indicators on modeling was investigated using the Jacknife method, according to which the NPP factor was the most effective indicator. After that, TVDI, precipitation, evaporation, soil type and LST were the most important indicators, respectively, and the other indicators were less effective in identifying dust source regions. Maxent output prediction map in ArcMap was classified into 4 classes including very severe, severe, moderate and very low. 51% of the province was classified as very severe, 31% as severe, 16% as moderate and 0.8% as very low. Keywords: Jack Knife test, environmental parameters, remote sensing, dust source, Maxent model, ROC curve.

رضا جعفري

مصطفي تركش

حميدرضا كريم‌زاده، عليرضا سفيانيان