The Validation Of MODIS Aerosol Products And The Improvement Of Merged Products Over China

The latest version of the Moderate Resolution Imaging Spectroradiometer（MODIS）Collection 061（C6.1）aerosol products was released in 2017.Due to the lack of sufficient ground-based measurements,few comprehensive validations of MODIS C6.1AOD products have been conducted over China.This paper firstly evaluated and compared the performance of the MODIS Aqua C6 and C6.1 Dark Target（DT）and Deep Blue（DB）aerosol optical depth（AOD）products at 10 km resolution against China Aerosol Remote Sensing Network（CARSNET）and Aerosol Robotic Network（AERONET）AOD data in China during the time period 2002–2016.In order to improve the retrieval accuracy of MODIS aerosol products,this manuscript developed a newly C6.1 merged AOD based on MODIS NDVI（MYD13A3）and land cover product（MCD12Q1）and the degree of improvement of the new merged product was evaluated.In recent years,Henan Province suffered frequent haze weather and serious air pollution.Therefore,the improved AOD products were used to analyze the spatio-temporal distribution of AODin Henan Province from 2003 to 2016.At the same time,the aerosol types and the source of pollutant based on the backward trajectory model in Jiaozuo were analyzed using the ground-based AOD from Jiaozuo-HPU and meteorological data,respectively,to study the source of pollutant and provide a scientific basis for air pollution control in Henan Province.The main conclusions are as follows:（1）For C6.1 DB,the significant improvements were observed in Middle East and Central Asia but not in most sites of Northwestern China.Aqua C6.1 DB slightly improved the retrieval accuracy with 2.3%more data falling within the Expected Error(EE:±(0.03+20%AOD_ground)),but only achieved 39.85%of retrieval accuracy.C6.1DB also shows considerable improvements over Gaolanshan and SACOL.The retrieval accuracy has increased from 23.55%and 35.99%to 48.98%and 75.04%.RMSE has decreased from 0.225 and 0.176 to 0.179 and 0.123.RMB has also increased from0.509 and 0.603 to 0.724 and 0.988.The retrieval accuracy of C6.1 DB over SACOL has exceeded 67%of the retrieval target.In addition,C6.1 DB also shows slightly improvements over Zhurihe,Ulate and Ejina.The retrieval accuracy of C6.1 DB at other sites has hardly improved or even decreased at some sites.For most sites,C6.1DB still has significant underestimation similar to C6,which indicates that the aerosol model selected by C6.1 DB for retrieving aerosol products in Northwest China needs to consider more absorption.（2）The results show that the C6.1 DT algorithm has a slightly better performance than the C6 algorithm,with a reduced RMSE of 0.171,a higher correlation coefficient（R:0.901）,and an increased retrieval accuracy over China from 54.94 to 59.03%.C6.1DT has adopted a new surface reflectance scheme in urban areas.The retrieval accuracy has been improved by 11.8%to 56.81%.Although the degree of overestimation has been reduced,the 11%of AOD overestimation shows that the hypothesis of the relationship between surface reflectance in visible and short-wave infrared bands in urban areas of China still needs to be improved.C6.1 DT has hardly improvement over forest,cropland and grassland sites.The improvement degree of C6.1 DT performance varies from cities.As far as the retrieval accuracy is concerned,the degree of improvement of C6.1 DT over Beijing and at other sites in North China is positively correlated with the proportion of urban pixels around the sites.（3）The number of collocations of the new merged AOD product(DTB_improved)is13.55%higher than that of DTB,which shows that the DTB_improved AOD significantly improves the spatial coverage of AOD in China.The proportion of DTB_improved AOD falling within the EE(±(0.05+15%AOD_ground))has increased from 64.86%to 67.95%and increased by 3.09%,exceeding the retrieval target of at least 67%of retrievals within the EE.The performance of DTB_improved is improved significantly at most sites,and the retrieval accuracy is increased,and the root mean square error is reduced.Especially at urban and cropland sites,DTB_improved almost achieved more than 67%of the retrieval target.It is more reliable to use DTB_improved AOD for investigating the distribution of aerosols because of improving spatial coverage and retrieval accuracy.（4）The long-term tendency of AOD over Henan Province from 2003 to 2016 was analyzed by C6.1 10 km DTB_improved AOD.In terms of time variation,the AOD inter-annual variation of Henan Province from 2003 to 2016 showed a tendency of rising first and then declining,and the maximum value was 0.78 in 2011.The seasonal average AOD changed obviously,the value in summer was the largest,but the overall trend is declining.On the contrary,the AOD in winter was relatively low,however,it presented an apparent upward tendency.Considering the spatial distribution characteristics,the AOD spatial distribution in Henan Province presents a layout of high in east and low in west separated by The Beijing-Guangzhou Railway.The high-value areas mainly are distributed in Central Henan,Southeastern Henan,Northeastern Henan and Southern Henan,which are hilly and plain areas,as well as southwest of Henan Province dominated by basin topography.The AOD value for the above areas is generally greater than 0.7.The aerosol types in Jiaozuo area are dominated by absorbent fine particles and absorbent mixed aerosol particles,mainly from carbon aerosols produced by industrial and agricultural activities.According to the analysis of potential sources of pollutants in Jiaozuo’s two pollution incidents,air pollution in Henan Province is not only related to local emissions,but also possibly affected by pollutant transport in surrounding areas and long-distance dust transport in Northwest China.