Abstract <p>This paper presents a method that integrates transmitted laser intensity and absorption pulse signal models to achieve a balance between the complexity, robustness and accuracy of gas concentration estimation using tunable diode laser absorption spectroscopy (TDLAS). Mathematical models take into account absorption, scattering, and reflectance effects and the relationship between the phase shift and the time delay and the absorption pulse signal quality. The Levenberg–Marquardt algorithm is applied to define optimal model parameters, enabling an efficient and effective of gas concentration estimation process. Experimental results demonstrate a strong correlation between the harmonic ratio feature and methane concentration under varying laser power conditions. Furthermore, the phase shift effect on the absorption signal is investigated, revealing its impact on waveform symmetry and peak distribution. The proposed method has been proven to require only a few signal cycles for gas concentration estimation.</p>