中文摘要：

为了克服利用太阳散射光作光源的被动差分吸收光谱（DOAS）系统中异方差对反演结果的影响,对DOAS反演算法进行了研究。鉴于局部多项式回归良好的统计性质,且当采用非参回归时不必人为设定异方差的结构,能适应几乎所有回归估计,提出利用非参稳健局部多项式回归方法估计DOAS系统测量谱中的慢变吸收部分,并据此反演痕量气体斜柱浓度。把该方法应用于外场实验获得光谱信息处理中,结果表明,本文方法比最小二乘多项式回归估计反演效果更好,降低了被动DOAS系统测量误差。

英文摘要：

In order to avoid the influence of heteroscedasticity on retrieval results in passive differential optical absorption spectroscopy（DOAS） using scattered sunlight,the retrieval method of passive DOAS is discussed in this paper.There is better statistical characteristic in local polynomial regression method,and the non-parameter regression does not need to give heteroscedastic structure and can be applied in almost all evaluations.Therefore the non-parameter robust local polynomial regression is applied to evaluate the slowly varying absorbed component and the slit column density of trace gases is retrieved.This method is used to process the measured spectra in field campaign.Results show that the new method obtains better retrieval effects than the least square polynomial regression evaluation and the measurement error of passive DOAS is reduced.