The Data Fusion Of Aerosol Optical Thickness Using Universal Kriging Method

As the problem of atmospheric pollution and air quality getting worse, the climate and environment effects of aerosols have attracted more and more attention. Aerosol Optical Depth (AOD) is one of the most important parameters which can indicate the atmospheric turbidity and aerosol load. High-quality AOD sources are significant for researches in the atmospheric environment, air quality and associated fields. Nowadays, satellite remote sensing products and ground-based products are widely used in the related studies. However, because of the limited spatial coverage and uncertain representativeness, they cannot meet the increasing demand in researches and applications. Using Universal Kriging (UK) method and regression analysis method, this paper fused different remote sensing AOD products with meteorological factors and Human Footprint (HFP) data. Besides, the fused AOD were validated by ground-based AOD retrieved from the AErosol RObotic NETwork (AERONET).This study indicates that we can acquired fused AOD products with a better accuracy from the MODerate-resolution Imaging Spectroradiometer (MODIS) AOD, Ozone Monitoring Instrument (OMI) AOD and AERONET AOD products by using UK data fusion method. We acquired ideal fused results by employing these three AOD products in May,2008. Furthermore, this paper confirmed that UK method can be applied to AOD fusing in East China although the source AOD products were retrieved from different algorithms and different meteorological conditions, etc. The root-mean-square error (RMSE) between AERONET AOD and daily source AOD products is 0.137 (month average in May,2008 is 0.812) while RMSE between AERONET and fused results is 0.101, decreased by 0.036,26.277 percent of 0.137.However, UK method cannot improve the data accuracy and coverage at the same time. The coverage of fused results retrieved from UK method is the same as the common coverage of all source AOD satellite products. Therefore, expand the common coverage of different source AOD products is of great benefits to improving the accuracy of AOD fused results. This paper had built the simple linear regression model by using the MODIS/Aqua AOD and MODIS/Terra AOD, and then used the MODIS/Terra AOD to predict the MODIS/Aqua AOD at the location where there is no MODIS/Aqua AOD value but has MODIS/Terra AOD value. Thus, daily MODIS AOD with wider coverage were acquired. Besides, the models which were used to predict daily MODIS AOD and OMI AOD also had been built, and we can acquired daily AOD with wider common coverage by applying these models to daily AOD and meterological factors, Human FootPrint. These are better fusion soure AOD data.AOD fused results with the widest coverage can be acquired by applying the UK data fusion method to daily MODIS AOD and OMI AOD which were predicted by corresponding prediction model stated avove. The accuracy of AOD fused results over different pixels can be divided into three situations because of the different AOD predict accuracy, that is, covered by MODIS AOD and OMI AOD, covered only by MODIS AOD, and covered only by OMI AOD, RMSE between UK fused results and AERONET AOD is 0.101,0.111, and 0.124, respectively. All less than that between AOD daily satellites products and AERONET AOD. This can decreased more if the HFP data was used to predict the daily AOD. Therefore, the human activities has impacts on AOD and human activities is one of the most important reasons why the AOD was increasing all the time in the past years.