Aerosol Optical Depth Automatically Retrieving Method And Validation

The air pollution problem seriously affects people’s living environment. More and more people have been concerned about it and aerosol pollution is an important aspect. In recent years, the satellite technology has a rapid development. According to its characteristics of fast imaging, low cost and small constraint by weather factors, many countries have begun to long-termly and real-timely monitor the wide range of aerosol pollution.According to the growth state of vegetation, growth periods of vegetation in the study area can be divided into the bare ground period, the dense vegetation period, and the sparse vegetation period. The former used improved time series algorithm for MODIS images of the bare ground period, Dense Dark Vegetation(DDV) algorithm for MODIS images of the dense vegetation period, surface spectrum model algorithm for MODIS images of the sparse vegetation period and finally realized the retrieval of aerosol optical depth in each period of the study area.However, the study above has the following two questions:1.There only exists this study area’s aerosol optical depth minimum that the bare ground period’s algorithm needs, we must artificially obtain when other study areas use this period’s algorithm.2.MODIS images of each vegetation growth period in the study area are filtered by hand.To solve these problems, research content of this paper is to make the manual part become automatic part to save time and manpower. Specifically:1.Statistic research to obtain global aerosol optical depth minimum.2.Method research to automatically filter vegetation growth periods’ MODIS images.We realize the retrieval of aerosol optical depth by automatically filtering vegetation growth periods’ MODIS images in the study area and conduct validation using the former’s ground observation site. For comparison, we add the validation result of the retrieval of aerosol optical depth by manually filtering vegetation growth periods’ MODIS images in the study area. It shows: the Root Mean Square Error(RMSE) between retrieval values and observation values in the former is 0.1779 and the fit factor(R2) is 0.9271; the Root Mean Square Error(RMSE) between retrieval values and observation values in the latter is 0.1500 and the fit factor(R2) is 0.9173. Thus, they have a good coherency.We complete the retrieval of aerosol optical depth and validation in tropical, arid and cold areas combining this method with global aerosol optical depth minimum(we use ppeno K(5)(5) ’s climate classification method here). The results show: the Root Mean Square Error(RMSE) between retrieval values and observation values in tropical area is 0.0278 and the fit factor(R2) reaches 0.9681; the Root Mean Square Error(RMSE) between retrieval values and observation values in arid area is 0.0943 and the fit factor(R2) is 0.8951; the Root Mean Square Error(RMSE) between retrieval values and observation values in cold area is 0.0345, and the fit factor(R2) is 0.7427.Thus, we can apply the method for the retrieval of global aerosol optical depth and will quickly monitor global atmospheric pollution.