Improved Data Processing and a Prior Profile Generation Method for Precise Retrieval of Atmospheric CO2 Based on a Laser Heterodyne Radiometer

The laser heterodyne radiometer (LHR) is a promising technique for atmospheric remote sensing due to its exceptionally high spectral resolution and sensitivity. A model based on a random forest algorithm is proposed to generate highly accurate prior atmospheric profiles using real-time meteorological parameters. In addition, a locally weighted scatter plot smoothing (LOWESS) method is applied for baseline correction during data preprocessing. An inversion algorithm is implemented using the Py4CAtS radiative transfer model, in which quadratic baseline parameters are included in the iterative process. Continuous measurements of the atmospheric CO2 absorption spectrum were made in our laboratory (Hefei, China, 31.9°N, 117.16°E), and the dry mixing ratio (XCO2) was obtained after data processing and inversion. The results demonstrate that this research improves the accuracy of LHR signal inversion. The implemented Python-based framework shows potential for real-time atmospheric CO2 monitoring.