Improvement Of Aerosol Optical Depth Retrieval Method From Visibility Data

This paper introduced particle swarm optimization(PSO)into the M-Elterman model of AOD retrieval using visibility data and established the PSO-M-Elterman model.The PSO-M-Elterman model was able to successfully obtain the monthly mean AOD data series for 661 stations in China between 1960 and 2014 and for 2,435 stations globally between 1973 and 2019,through the use of visibility and meteorological data from the China Meteorological Administration(CMA)and global GSOD datasets combined with aerosol optical depth(AOD)data that was observed by satellite remote sensing(MODIS/MISR)and ground-based remote sensing(AERONET).The improved model enabled significant improvement to the accuracy and correlation between retrieval AOD and actual AOD that is observed by satellite remote sensing.The spatiotemporal characteristics of AOD in China and the world were analysed,and the reasons for the long-term changes of AOD in China were discussed.The main conclusions include the following:Prior to retrieval,the scheme of converting the visibility that was recorded by grade to horizontal distance before 1980 was successfully improved through the establishment of the correspondence between the visibility grade that was recorded by CMA in different months and the actual visibility distance.The relative error of visibility grade conversion to distance was decreased from over 10%to below 1%and the average absolute error was controlled at approximately 0.1 km.The maximum likelihood regression test and MannKendall test were both used for testing and revising the visibility data and a more homogeneous monthly visibility data series was obtained in China for the period 1960 to 2014.Compared to the nonlinear least square method,the PSO algorithm has the ability of obtaining the global optimal solution more quickly and with better stability.The data unavailable rate of retrieval AOD that was calculated from the PSO-M-Elterman model was reduced by over 10%,and the proportion of stations with smaller absolute error achieved 91.23%.In addition,the proportion of stations with a stronger correlation coefficient was greater than 54%,exhibiting an increase of approximately 20%compared to the M-Elterman model.The retrieval results show that AOD increased with a trend of 0.0011/yr between 1960 and 2014 in China,during which time rapid growth occurred prior to 2000.The value of AOD in large cities is greater than that in other cities,which is accompanied by a more rapid upward trend.The aforementioned facts confirm the increase of aerosol in China in the last 50 years is mainly a result of increased human activities.The fluctuation of AOD in the Taklimakan region is much more significant than in other regions,which is potentially related to intermittent sandstorms.Globally,high visibility values are mainly distributed in western parts of North America,north-east Asia,Australia,and southern Africa.Since the 1970s,visibility has generally decreased everywhere with the exception of Europe.The value of retrieval AOD in different countries and regions fluctuated slightly between 1973 and 2019 and AOD in northern and southern Africa and East Asia exhibited an upward trend,indicating that the aerosol content in these areas has increased in recent decades.