Detection Of Dust Storms And Its Effect On Aerosol Retrieval Using NPP VIIRS Data

Aerosols are heterogeneous systems consisting of various types of gases and liquids and(or)solid particles suspended in the air.As aerosol particles can scatter and absorb electromagnetic waves,which can change the Earth's radiation budget,cause environmental pollution,and affect global climate change.Therefore,the quantitative description of aerosols is of great significance for studying global changes,pollution prevention,and restoration of the ecological environment.Dust storms,as an important source of aerosols,not only directly affect environmental pollution and climate change,but also the change of aerosol radiative characteristics,thus affecting the remote sensing retrieval of aerosol optical parameters.In this paper,the temporal and spatial effects of dust storm on aerosol retrieval were studied on the basis of dust storm detection and aerosol retrieval using NPP VIIRS data and combined with the aerosol optical properties data from AERONET(Aerosol Robotic Network)ground stations.The main contents include:(1)A new Simplified Dust Detection Algorithm(SDDA)was proposed,which can achieve automatic detection of dust storms.The accurate extraction of dust storm regions was the prerequisite of analyzing the temporal and spatial effects of dust storm on aerosol retrieval.Aim at detecting the temporal and spatial distribution of dust storms,a dust storm detected pixel database,consisting of cloud,vegetation,Gobi,ice/snow and dust,was constructed using the information on reflectance characteristics and emission characteristics of various types of ground objects.Based on the pixel database,the radiative characteristics of cloud,ice/snow,bright surface,dark surface,and dust were analyzed in single or multi-channel,and the extraction of dust storm information was realized.Visual interpretation technique,meteorological data and existing productions were used to verify the detection results.The results showed that the accuracy of the algorithm was up to 83.10%,and its robustness,and can achieve high precision dust storms detection.(2)Based on MODIS land surface reflectance products,the high precision retrieval of aerosol optical depth of VIIRS data were realized.The determination of aerosol type and surface reflectance is the key to the retrieval of aerosol over the land.In order to reduce the effect of aerosol types on the results,the aerosol optical parameters in the study area were analyzed using AERONET data from 2012 to 2017,and were used for the retrieval of aerosol.For improving the accuracy of surface reflectance determination,a database was constructed using MODIS surface reflectance products for the determination of land surface reflectance of VIIRS,and a spectral transformation model between different sensors was established to reduce the influence of spectral differences.The AOD(Aerosol Optical Depth)retrievals were validated using AERONET data,the results showed that the AOD retrieved using the proposed algorithm were in good agreement with the ground-based measurements.(3)In order to analyze the impact of dust storms on aerosol retrieval,multiple dust storm events from 2012 to 2017 were selected.By comparing the accuracy of retrieval results of two aerosol types within 5 days before and after the dust storm,the influence of dust storm on aerosol type was illustrated indirectly.In addition,statistics were also made on the retrieval results,respectively.Mean Absolute Error(MAE),Standard Deviation(STD),Relative Mean Bias(RMB)and Root Mean Square Error(RMSE)were analyzed,and the influence of dust storms on aerosol was described by polynomial fitting.The results showed that dust storms can cause changes in the optical properties of aerosols,which in turn lead to larger AOD retrieval errors.