Evaluation Of MODIS Aerosol Product Collection 006 And Its Application To Air Pollution Research In Zhejiang Province

At present, China’s urban air pollution manifests as compound pollution which is multi-pollutant coexistent and multi-pollution-source superimposed. Over the past decade, haze events which were caused by fine-mode aerosol particles have increased rapidly in eastern China. Thus, it’s necessary to strengthen the efforts to monitor aerosols especially fine-mode particles. Compared with ground-based observations, satellite-based remote-sensing has the advantage in large-scale simultaneously observation, can provide the spatial distribution and variation trends of air contaminants, and its cost is relatively low. MODIS （Moderate-resolution Imaging Spectroradiometer） aerosol products are currently the most widely-used satellite-derived aerosol product. In order to meet the requirements of local air quality applications, NASA released MODIS aerosol products Collection 006 （updated version over Collection 051） in early 2014, providing with improved 10km product and new 3km product which has a higher spatial resolution.This paper took Zhejiang Province as study area, verified the accuracy and applicability of MODIS aerosol products C006, carried out the research on the relationship between AOD and ground-based observations, analysed the long-term trend of AOD over the plain areas of northern Zhejiang province during year 2002-2013. The main conclusions of this paper are as followed:1. The accuracy of MODIS aerosol products C006 was validated, convicing that it can be applied in air quality research. MYD04_L2_C051, MYD04_L2_C006 and MYD04₃K_C006 showed consistency in AOD spatial distribution, and their AOD point values and regional average values correlated well. The monthly mean AOD of these three products correlated significantly, indicating the change of aerosol products had little effect on trend analysis the timeseries of AOD monthly mean. Algorithm updates had different impacts on different land types:the fitting coefficients between each product over urban areas are lower than those over vegetated areas; compared with those from MYD04_L2_C051 and MYD04₃K_C006, MYD04_L2_C006 demonstrated systematic higher AOD over vegetated areas. Validation with AERONET showed that MODIS aerosol products with certain accuracy were applicable over Hangzhou_City, Hangzhou_ZFU, Qiandaohu and LA-TM （R*=0.7110-0.8920, EE%=46.7%-84.1%）. Compared with MYD04_L2_C051, MYD04_L2_C006 and MYD04₃K_C006 improved in product precision over vegetated areas but deteriorated slightly over urban areas. Research on typical polluted days showed that, due to algorithm updates relating to cloud detection and land sea mask, the data coverage of MYD04_L2_C006 and MYD04₃K_C006 decreased significantly, with drops of 6.31%～10.96% and 32.37%～37.96% respectively when compared with MYD04 L2 C051.2. The relationship between MODIS AOD and ground-based observations suggested that, satellite remote sensing had the ability to monitor the near-surface aerosol particle concentration, and the correction of aerosol scale height （HA） and relative humidity （RH） improved the retrieval accuracy, however, the 3km product （relatively higher resolution） did not performed better than the 10km product （moderate resolution）. Analysis of the whole-year dataset showed that, with a smaller sampling box size the correlative coefficients between PM2.5 and AOD got higher; the HA and RH correction significantly improved the relevance between PM2.5 and AOD. The relationship between PM2.5 and AOD showed seasonal variations:the correlation was good during spring, summer and autumn （R=0.52～0.76,0.66～0.79 and 0.66～0.75）, but much lower during winter （R=0.51-0.53） when RH correction performed poor. Compared with 10km product,3km product showed a better correlation with PM2.5 during spring and summer, but during autumn and winter it was quite the contrary. Validation results from one autumn-time case displayed that, with Ha and RH correction the mean absolute errors between the retrieval values and observational values of PM2.5 decreased from 14μg/m3 and 16μg/m3 to 6μg/m3 and 9μg/m3 for 10km product and 3km product respectively, reflecting that aerosol scale height and relative humidity played a non-ignorable role on PM2.5-AOD relationship and 10km product performed slightly better than 3km product in PM2.5 retrieval.3. The long-term trend of AOD over the plain areas of northern Zhejiang province during year 2002-2013 was analysed using 10km product from C006. The timeseries of monthly mean AOD showed a periodic variation, but with no significant trend （P>0.05）. The regional AOD showed significant difference in seasons, for most years it was higher during spring and summer while lower during winter and autumn, except for year 2008 and 2013. Annual mean AOD climbed during year 2002-2008 and then declined. For the spatial distribution, aerosol pollution was heavier in the central part of the research area where AOD was higher than that of the surrounding area.