Retrieval Of Aerosol Optical Depth Based On HJ-1 Satellite

In recent years, the climate effects of aerosols have caused worldwide concern. As responsible for the task of environmental monitoring and disaster mitigation, the application potential of Environment-1 (HJ-1) Satellite is yet to be discovered. In this paper, Aerosol Optical Depth (AOD) was retrieved using HJ-1-CCD data. MODIS aerosol products and ground-based CE318 sun photometer AOD measurements were used to test the accuracy of HJ-1 AOD inversion results. Finally, HJ-1 AOD and meteorological data were used to analyze a typical pollution process, and a multi-regression model for Air Pollution Index (API), AOD and meteorological factors was constructed. The following conclusions were obtained:1) Dark Density Vegetation (DDV) method was used to retrieve AOD from HJ-1 CCD sensor,6S radiative transfer model was used to construct the multi-wavelength aerosol lookup table (LUT), and NDVI was used to determine the dark pixels. The HJ-1-CCD AOD inversion result showed the spatial distribution characteristics of aerosols. However, in less vegetation season, or in cities with higher surface reflectance, the HJ-1 AOD result retrieved by DDV was not ideal.2) To test the accuracy of AOD retrieved results, MODIS aerosol products over Beijing area in Mar-Jun 2009 were used to make comparisons with HJ-1 AOD results,31 groups of comparative data showed a correlation coefficient of 0.791, the absolute error was about 0.06, while the relative error was about 13.6%. The standard deviation of HJ-1 AOD was slightly larger than MODIS AOD products. Spatial contrast between the re-sampled HJ-1 AOD and MODIS AOD products showed that the two AOD results reflect the consistent aerosol distribution characteristics, with nuances in the identification of few dark pixels. AOD scatter comparison showed that the correlation coefficient was 0.7265, the results were close. At the same time, scatter comparison of red band reflectance shows a correlation coefficient of 0.7275, the reflectivity of HJ-lwas slightly smaller than MODIS.3) Comparison between HJ-1 AOD and ground-based CE318 AOD observations in the same period showed a correlation coefficient of 0.8517. The average AOD observed by CE318 was 0.71, while the average AOD retrieved from HJ-1-CCD was 0.82, slightly higher than the former. Compared with CE318 observations, the absolute error of HJ-1 AOD was about 0.09, the relative error was about 12.5%.4) Based on the Air Pollution Index (API), a dust pollution happened in March 17-19,2009 was chosen, and HJ-1 AOD and ground-based observations were used comprehensively to analyze the typical pollution process. The changes of AOD from HJ-1 and CE318 observations corresponded well to meteorological factors. The continuity of ground-based observations and macroscopic of satellite observations was effectively complementary.5) Finally, the research attempted to establish a relational model among API, HJ-1 AOD and meteorological factors, and a validation was made. The result showed that the model could be a reference in calculating API using satellite data, but it was still far from practical applications.