作　　者： (李姗姗）; (邓小波）; (丁继烈）; (刘海磊）;

机构地区： 成都信息工程大学,四川成都610225

出　　处： 《成都信息工程大学学报》 2017年第3期276-281,共6页

摘　　要： 对于痕量气体二氧化碳浓度的精确测量问题,首先比较大气辐射传输的几种模型,然后选用SCIATRAN这种模型在二氧化碳的强吸收带进行晴空大气和卷云大气辐射的模拟计算,接着对计算结果进行敏感性测验,结果表明:晴空时的辐射强度比卷云时的辐射强度小;卷云条件下太阳天顶角和卫星观测角越小时,辐射强度越大;卷云条件下地表反照率越大时,辐射强度越大;卷云条件下光学厚度越小时,辐射强度越大。最后用神经网络的方法对SCIATRAN计算的结果进行训练和模拟并与实际结果进行对比分析,以期对于下一步二氧化碳浓度的反演模型和算法提供有力的理论依据和技术支撑。 For accurate measurement of trace gases CO2 concentrations,at first,this paper compares several of atmospheric radiation transmission models,then choose the SCIATRAN model to simulate the cloudless atmospheric radiation and cirrus atmospheric radiation in the strong absorption band of carbon dioxide,next do some sensitivity experiments with calculation data,the results show that： The intensity of clear sky is smaller than cirrus sky; Under the cirrus atmosphere,the smaller of sun zenith angles and satellite viewing angles,the stronger of the radiation intensity; Under the cirrus atmosphere,the bigger of ground albedo,the stronger of radiation intensity; Under the cirrus atmosphere,the smaller of Optical thickness,the stronger of radiation intensity.Finally,we use the method of neural network to train and the simulate the results of calculation,and compare with the real results of calculation,analyze the bias.In order for the carbon dioxide concentration inversion model and algorithm provide strong theoretical basis and technical support.

关 键 词： 大气辐射传输 模型 敏感性测验 神经网络