CO2作为影响气候变化最重要的温室气体，其反演精度达到1%是气候研究的基本要求。在反演中解决大气散射的影响，是提高CO2反演精度的关键问题之一。温室气体观测卫星为了实现高光谱分辨率，其光谱带宽通常较窄。高光谱分辨率对CO2浓度变化敏感，而窄带宽在采用传统差分吸收光谱(DOAS)法以快慢变分离的方式处理散射时难以保证反演精度。针对我国高光谱卫星CO2反演算法的开发需求，从光程的角度研究了散射对CO2反演的影响，并与传统DOAS方法在沙漠和草地两种区域进行了对比。结果显示相对于传统DOAS方法，该方法在沙漠和草地区域的应用均使CO2的反演精度得到提高，达到或接近1%的精度需求，反演结果的相关性和数据离散度也得到改善，这表明该方法能有效降低大气散射对CO2反演的影响。

CO2 as the most important greenhouse gas effecting climate change, inversion accuracy within 1% is essential requirement in climate research. One of key points to improve CO2 retrieval precision is account for atmospheric scattering in inversion processing. The spectral range of greenhouse gas monitoring satellite generally is narrow to achieve excellent spectral resolution. High spectral resolution has high sensitivity in CO2 concentration changing, however, traditional differential optical absorption spectroscopy (DOAS) methods are hard to assure accuracy in scattering atmosphere. Aim to meet inversion algorithm development, the influence of scattering in CO2 retrieval from path length is studied, compared with traditional DOAS method in desert and meadow. Result indicates that path length method can improve retrieval accuracy and obtain precision close to 1%, and correlation and dispersion of inversion results also can achieve improvement. It is implied that path length method is capable of reducing the impact due to atmospheric scattering in CO2 retrieval.