The combination of Variational mode decomposition （VMD） and Savitzky-Golay filtering （SG filtering）， abbreviated as VMD-SG filtering， was suggested for the reduction of noise in signals containing noise， with a particular emphasis on the absorption spectroscopy of tunable semiconductor lasers and the issue of noise in the second harmonic during the measurement of low concentration gases. The optimal selection of VMD balance parameters was analyzed， and the set frame length adjustment factor P was utilized to select the optimal frame length for SG filtering. When compared with other denoising algorithms， it was found that the VMD-SG filtering denoising algorithm performed better in terms of signal-to-noise ratio， root mean square error， waveform recovery， and the error of the second harmonic feature point. At signal-to-noise ratio （SNR） of -7.838 9 dB， an improvement of 22.047 dB in SNR was achieved by the algorithm， a root mean square error of 0.015 4 was recorded， and a correlation coefficient of 0.985 with the standard second harmonic was determined. The results were indicative of the fact that the tunable diode absorption spectrum composite denoising technology， which was based on the VMD-S-G filtering denoising algorithm， could more effectively enhance the quality of signals in weak signal conditions. It was also found to be beneficial for enhancing the accuracy of gas inversion and the sensitivity of system detection.