基于线性调频Z变换的差分吸收光谱数据处理方法研究

差分吸收光谱法(differential optical absorption spectroscopy，DOAS)是一种常用的污染气体监测方法，对所监测的光谱数据去噪可以提高反演精度。可采用傅里叶变换(fast Fourier transform, FFT)滤波法滤除光谱数据中的噪声，但该算法本身会引入误差。提出一种线性调频Z变换法(chirp Z transform，CZT)，通过对傅里叶变换之后的频谱进行局部细化，能够在保留傅里叶变换滤波法去噪效果的基础上，对算法的误差进行补偿，从而进一步提高反演精度。实验配置了SO₂及NO₂进行浓度反演，结果表明，直接采用相除法反演浓度时误差较大且很不稳定，线性调频Z变换法能够获得比傅里叶变换滤波法更高的反演精度。模拟了SO₂和NO₂混合气体实验，频谱分析结果表明FFT算法无法解决特征吸收结构被扭曲、削弱等问题，CZT算法能完成特定频段频谱的精细化重构。

Study on Differential Optical Absorption Spectroscopy Data Processing Based on Chirp-Z Transformation

Differential optical absorption spectroscopy (DOAS) is a commonly used atmospheric pollution monitoring method. Denoising of monitoring spectral data will improve the inversion accuracy. Fourier transform filtering method is effectively capable of filtering out the noise in the spectral data. But the algorithm itself can introduce errors. In this paper, a chirp-z transform method is put forward. By means of the local thinning of Fourier transform spectrum, it can retain the denoising effect of Fourier transform and compensate the error of the algorithm, which will further improve the inversion accuracy. The paper study on the concentration retrieving of SO₂ and NO₂. The results show that simple division causes bigger error and is not very stable. Chirp-z transform is proved to be more accurate than Fourier transform. Results of the frequency spectrum analysis show that Fourier transform cannot solve the distortion and weakening problems of characteristic absorption spectrum. Chirp-z transform shows ability in fine refactoring of specific frequency spectrum.