Contrails Identification With Convolutional Neural Network And Simulation Study Of Radiative Forcing In Southeast Asia

In this paper,we use the artificial visual inspection,Convolutional Neural Network recognition algorithm and the traditional contrail recognition algorithm CDA method to calculate the contrail coverage in southern China,based on the Himawari-8 satellite data,and compare the results.Then select potential contrail parameterization scheme,verifying the scheme at the same time,combining reanalysis data,CCSM4 model data and QUANTIFY aviation data,simulating historical scenes in 2000 and 2050 and 2100 Southeast Asia under RCP4.5 and RCP8.5 future scenarios respectively The contrail coverage was finally simulated by the contrail radiative forcing scheme in 2000 and the net radiative forcing of contrails in Southeast Asia in 2050 and 2100 in the future scenarios of RCP4.5 and RCP8.5 respectively.Mainly got the following conclusions:(1)Based on the Himawari-8 satellite image statistics,the contrail coverage in southern China is based on the results of artificial visual inspection.The convolutional neural network recognition results are significantly better than the traditional contrail recognition algorithm CDA.(2)Comparing the observation results with the potential contrail parametric simulation,it shows that there is a high correlation between contrail occurrence and persistence COP and potential contrail coverage PCC,and the Convolutional Neural Network identification CNNI correlation coefficient is close to artificial visual inspection AVI.(3)The average contrail coverage and average net radiation of the contrail in2000 simulated in this paper are 0.11% and 5.54 respectively.The average contrail coverage and average net radiation of contrail in Southeast Asia are 0.22% and 10.41,respectively.(4)In Southeast Asia,in 2050 and 2100,the average coverage of contrail and the average net radiative forcing increased in different scenarios.In the RCP4.5scenario,the growth was more moderate,while in the RCP8.5 scenario,the growth was more moderate fast.(5)The average coverage and net radiative forcing of contrails in Southeast Asia are higher in summer than in winter.