Influence Of Cloud And Aerosol In Atmospheric CO₂ Inversion And Its Correction Method

Human activities have led to an increase in atmospheric greenhouse gas content,which has caused global warming and frequent extreme weather problems.It is becom-ing a hot issue in today's society.CO2 as an important greenhouse gas,the basis of global climate change research is to detect its concentration distribution and changes.Satellite atmospheric CO2 remote sensing has the characteristics of long-term sequence and wide spatial range on a global scale,and cannot be replaced by other technologies.CO2 is a resident gas with low concentration and gradient in the atmosphere.Only high-precision observation can play a role in climate research.Therefore,the key to satellite atmospheric CO2 remote sensing needs to assure high inversion precision.So,the in-version method is one of the key technologies.As a technique for extracting CO2 con-centration information from remote sensing data,inversion is faced with many factors such as cloud pollution,atmospheric and surface parameter inaccuracy.Therefore,the current inversion results are difficult to improve the accuracy and the results are not stable enough.In this paper,the problems faced by atmospheric CO2 inversion are an-alyzed and proposed to overcome the effects of cloud,aerosol,and surface on the in-version.The process of CO2 inversion is completed on this basis.From the statistical results,most of the sky is covered by clouds.In atmospheric CO2 inversion,remote sensing data covered by clouds cannot be used for inversion.Therefore,cloud screening is a basic work to improve the efficiency of inversion and guarantee the quality of inversion.At present,there are many remote sensing data cloud detection methods,and all methods have certain advantages and disadvantages.Based on the analysis of the characteristics of existing methods,this paper proposes a multi-load collaborative screening method,which we call the CGDC algorithm(Combined GMI and DPC Cloud-screening).It combines the high homogeneity of oxygen in the atmosphere with the sensitivity of the polarization spectrum to cloud radiation to im-prove cloud screening efficiency in greenhouse gas inversion.The algorithm is applied to the global 16-day on-track measured data of the Greenhouse-gases Monitoring In-strument(GMI)in China,and 77,518 GMI observation points are detected,and 9,508 clear-air observation points are screened,accounting for 12.26%.Using the fused MODIS cloud mask and cirrus reflectance dataset,the correct rate of cloud detection can be 92.93%on land and 81.91%on the ocean.Aerosols and cirrus change the distribution and transmission path of atmospheric radiation processes.That is an important factor affecting the accuracy of inversion.Based on the influence mechanism,using MODIS global aerosol model and GHM cir-rus model to simulate the typical brightness surface of GMI observation data,we ana-lyze the CO2 inversion error distribution caused by the uncertainty of the main model aerosol and cirrus optical thickness from the light path changes.Furthermore,the effects of optical thickness and vertical distribution on the inversion results were further stud-ied,guiding the selection and processing of aerosol profiles during our inversion.In view of the important influence of aerosol model in the inversion process,this paper introduces cluster analysis technology to summarize and classify global aerosol char-acteristics and distribution,and builds an aerosol model database suitable for global inversion of GMI.And the temporal and spatial variation of aerosols in the four seasons,giving key parameters such as volume distribution of different aerosol modes,optical thickness conversion between bands,asymmetry factor,and single scattering albedo to reducing CO2 inversion errors and improved comparability of global inversion results.The complexity of surface reflection radiation has become an important reason for affecting the accuracy of atmospheric CO2 inversion.The complexity of surface reflec-tion radiation is manifested by the difference of spatial radiation distribution and the surface-atmosphere coupling effect,which changes the calculation of atmospheric ra-diation transmission of light,affects the accuracy of atmospheric CO2 inversion.There-fore,the improvement of inversion comes from accurate description of the reflection characteristics in different directions of surface.In view of the huge workload of global research,this paper takes the Beijing-Tianjin-Hebei region as an example to carry out related research.The Beijing-Tianjin-Hebei region is rich in surface types and has a complex spatial and temporal distribution of aerosols.It is a high-value aerosol region under the influence of strong human activities.Therefore,it is representative of the surface,atmosphere,and CO2 sources.In this paper,the research area is divided into urban and non-urban areas.Taking Beijing city as an example,using the MODIS bidi-rectional reflectance distribution function(BRDF)data from 2011 to 2016,a model suitable for inversion of single observation data is constructed,and a simultaneous in-version algorithm of BRDF parameters and CO2 content is proposed.The results show that when the aerosol optical thickness is less than 0.4,the inversion error of most GMI simulation data is controlled within 0.5%(?2ppm).The inversion results of GOSAT data using the GMI spectral bandwidth truncation were compared with the modified GOSAT NIES inversion results,with an average error of 1.25 ppm and a correlation of 0.85.For non-urban areas,the RPV BRDF model was used instead of the Lambertian model to retrieve the observations with aerosol optical thickness less than 0.4.The cor-relation was 0.788 and the average error was 1.50 ppm.The refinement of the regional inversion method satisfies the demand for high-precision CO2 inversion of GMI data in the Beijing-Tianj in-Hebei region and makes it possible to retrieve high-value AOD re-gion and improve the utilization of GMI observation data.Based on the equivalent theory,the transformation of the PPDF factor from O2-A to CO2-1 band and post-data screening method are proposed.The inversion method of three-layer PPDF model suitable for GMI inversion is constructed completely.The in-version results are compared with the international instrument's L2 products such as OCO-2 and GOSAT.The results show that the inversion value of the PPDF algorithm is slightly higher than GOSAT overall,and the average deviation from the GOSAT product is 1.72 ppm,the correlation is 0.926;the average deviation of PPDF algorithm inversion value and OCO-2 product is-2.25 ppm with an overall correlation of 0.911.The areas with large errors are concentrated in the high latitudes of the northern hemi-sphere.The comparison with the TCCON station of the high-precision ground-based observatory shows that the correlation between the GMI inversion results and six TCCON stations is 0.741,and the average deviation is-0.85 ppm.The inversion results are quite different from the data of the Sodankyla site near the Arctic Circle.Excluding the site from the comparison,the correlation between GMI and TCCON increased to 0.869,and the average deviation decreased to-0.3 ppm.Overall,the inversion results of the three-layer PPDF algorithm on GMI data are stable with a high correlation be-tween GMI and TCCON.The inversion result is slightly different from the TCCON measured results after the latitude is greater than 600 due to the decrease of the GMI's SNR,which indicates that the application ability of the inversion method is expected to be further improved with the improvement of the remote sensing detection technology.