Investigation of the relationship between maximum dust and climatic variables using remote sensing data (Case study: Hormozgan province)

Dust phenomenon is one of the natural disasters that is considered as a serious environmental hazard, especially in arid and semi-arid regions due to the great damage it causes every year. The present study aimed to investigate the relationship between 14 climatic variables with the maximum monthly aerosol optical depth (AOD) due to the dust events in Hormozgan province. First, by coding in the Google Earth Engine (GEE) environment, a satellite image was retrieved from the MODIS aerosol products for each day, and while preparing the AOD time series, the average maximum monthly dust values for a 17-year period (2000-2017) was extracted. Also, monthly climate and water balance products of University of Idaho including actual and reference evapotranspiration, minimum and maximum temperature, precipitation accumulation, soil moisture, Palmer drought severity index, climate water deficit, downward surface shortwave radiation, vapor pressure, vapor pressure deficit, and wind speed, as well as land surface temperature (LST) and vegetation index (EVI) were extracted and, while sampling these images, the relationship between the average maximum monthly dust values with them were computed using the ordinary least squares (OLS) and geographic weighted regression (GWR) methods. Then, the global Moran's I statistics was employed to analyze the spatial autocorrelation and distribution of dust over the province. The results showed that the GWR model with the root mean square error of 0.14, the sum of residual squares of 11.3, the coefficient of determination of 0.82, and the corrected Akaike information criterion of -570.19, performed better than the OLS method. The evaluation of the coefficients of the GWR model showed that the variables of vegetation cover, soil moisture and precipitation had the greatest effect on the amount of dust, respectively. Also from the perspective of spatial autocorrelation, a cluster pattern was observed for dust distribution over the province.