气溶胶对利用近红外波段反演CO2浓度精度的影响研究
The Influence of Aerosol Particles on Retrieval Accuracy of CO2 Concentration Using Near Infrared Bands
中文摘要：
近红外波段CO2浓度反演误差主要与气溶胶散射作用引起的吸收光程改变量难以精确估计有关。此外，不同的气溶胶模式及下垫面反射率条件下吸收光程改变量也不尽相同。针对不同气溶胶模式条件下，忽略气溶胶散射对CO2
浓度反演精度的影响进行分析。研究发现，对乡村型、对流层型和海洋型气溶胶来说，忽略气溶胶散射效应，在暗地表时（反射率<0.1）会导致CO2浓度反演结果低估；当地表反射率超过0.1后，CO2浓度反演结果均为高估，且随地表反射率的增加，反演误差增加。对城市型气溶胶模式来说，忽略气溶胶影响均会导致CO2浓度反演结果低估，随地表反射率增加，反演误差递减。在典型观测几何下，地表反射率（1.6μm）和气溶胶光学厚度（0.55μm）在0.1～0.3范围内时，忽略气溶胶影响，城市型、海洋型、乡村型以及对流层气溶胶分别会引入-0.1%～-0.5%、0.22%～1.92%、0.09%～1.46%及0.02%～0.45%的反演误差。在利用GOSAT观测光谱对XCO2（干燥空气下CO2平均混合比）进行反演时发现，相比GOSAT发布结果（389.814ppm，1ppm=10－6），基于地基实测气溶胶特性数据的XCO2反演结果（390.95ppm）与地基观测结果（390.737ppm）有更好的一致性。
浓度反演精度的影响进行分析。研究发现，对乡村型、对流层型和海洋型气溶胶来说，忽略气溶胶散射效应，在暗地表时（反射率<0.1）会导致CO2浓度反演结果低估；当地表反射率超过0.1后，CO2浓度反演结果均为高估，且随地表反射率的增加，反演误差增加。对城市型气溶胶模式来说，忽略气溶胶影响均会导致CO2浓度反演结果低估，随地表反射率增加，反演误差递减。在典型观测几何下，地表反射率（1.6μm）和气溶胶光学厚度（0.55μm）在0.1～0.3范围内时，忽略气溶胶影响，城市型、海洋型、乡村型以及对流层气溶胶分别会引入-0.1%～-0.5%、0.22%～1.92%、0.09%～1.46%及0.02%～0.45%的反演误差。在利用GOSAT观测光谱对XCO2（干燥空气下CO2平均混合比）进行反演时发现，相比GOSAT发布结果（389.814ppm，1ppm=10－6），基于地基实测气溶胶特性数据的XCO2反演结果（390.95ppm）与地基观测结果（390.737ppm）有更好的一致性。
中文关键词：
XCO2反演，近红外波段，GOSAT，气溶胶
KeyWords：
retrieval of XCO2, near infrared bands, GOSAT, aerosols
Abstract：
The main difficulty in retrieving CO2 concentration from near-infrared observations comes from the uncertainty of light path change brought about by scattering of the atmospheric aerosols. The scattering effect described by different aerosol models and surface reflectances may lead to different light path changes. This paper aims to study the influence of neglecting aerosol scattering on CO2 retrieval accuracy for different aerosol models. It is found that for rural, tropospheric and marine aerosol models, the neglecting of aerosol scattering results in underestimation of CO2 retrievals at low surface reflectances (<0.1), but
overestimation at surface reflectance over 0.1 and the increasing retrieval errors with increasing surface reflectances. For the urban aerosol model, the neglecting of aerosol scattering results in underestimation of CO2 retrievals at any surface reflection, and the retrieval errors decrease with increasing surface reflectances. For typical observation geometry, a surface reflectance of 0.1-0.3 at 1.6μm and aerosol optical depth (AOD) of 0.1-0.3 at 0.55μm, the neglecting of aerosol scattering leads to retrieval errors of (-0.1%) - (-0.5%), 0.22%-1.92%, 0.09%-1.46% and 0.02%-0.45% for urban, marine, rural and tropospheric aerosol models respectively. XCO2 (the dry air column averaged mixing ratio of CO2) was retrieved from GOSAT observations with in-situ measurements of aerosol properties as input parameters. Comparison with ground-based XCO2 measurements shows that the retrieved XCO2 (390.95ppm) is more consistent with the measured XCO2 (390.737ppm) as compared to that of GOSAT L2 product (389.814ppm).
overestimation at surface reflectance over 0.1 and the increasing retrieval errors with increasing surface reflectances. For the urban aerosol model, the neglecting of aerosol scattering results in underestimation of CO2 retrievals at any surface reflection, and the retrieval errors decrease with increasing surface reflectances. For typical observation geometry, a surface reflectance of 0.1-0.3 at 1.6μm and aerosol optical depth (AOD) of 0.1-0.3 at 0.55μm, the neglecting of aerosol scattering leads to retrieval errors of (-0.1%) - (-0.5%), 0.22%-1.92%, 0.09%-1.46% and 0.02%-0.45% for urban, marine, rural and tropospheric aerosol models respectively. XCO2 (the dry air column averaged mixing ratio of CO2) was retrieved from GOSAT observations with in-situ measurements of aerosol properties as input parameters. Comparison with ground-based XCO2 measurements shows that the retrieved XCO2 (390.95ppm) is more consistent with the measured XCO2 (390.737ppm) as compared to that of GOSAT L2 product (389.814ppm).