Improved DQ-1 spaceborne IPDA lidar XCO2 retrieval accuracy by an enhanced inversion and post-processing framework

Abstract

Global climate change and the “Dual Carbon Strategy” have created an urgent demand for high- precision atmospheric carbon dioxide (CO₂) monitoring. In response, Atmospheric Carbon Dioxide Lidar (ACDL) onboard China’s Atmospheric Environment Monitoring Satellite (DQ-1) employs Integrated Path Differential Absorption (IPDA) lidar technology, enabling continuous day- and- night global observations and significantly improving the capability to detect global CO₂ concentrations. However, the retrieval accuracy of ACDL is constrained by factors such as atmospheric state uncertainty and random measurement noise.To address these challenges, this paper proposes a CO2 column-weighted dry-air mixing ratio (XCO₂) retrieval algorithm tailored for the DQ- 1. The algorithm integrates ERA5 reanalysis data for atmospheric profile reconstruction with an efficient particle filter-based inversion of CO2 for single observation (EPICSO) post- processing noise suppression method, while simultaneously correcting the Differential Absorption Optical Depth (DAOD). Using this algorithm, we processed global observation data from the DQ- 1 satellite from May to August 2023 and validated the results against Total Carbon Column Observing Network(TCCON). The results show that the root mean square error (RMSE) between the retrieved XCO₂ and TCCON data is approximately 0.86 ppm. Ablation experiments indicate that using ERA5 profile reconstruction alone primarily reduces systematic bias, while using EPICSO post- processing alone primarily suppresses random noise; combining the two achieves optimal accuracy and stability. The retrieved XCO₂ data effectively capture seasonal variations, land- ocean contrasts, and day- night differences. Specifically, XCO₂ in spring is about 3.13 ppm higher than summer, XCO₂ of land is about 4.21 ppm higher than that of the ocean, and the monthly mean nighttime XCO₂ is about 0.65 ppm higher than its daytime value. The fusion algorithm proposed in this paper demonstrates strong generalizability and practicality, providing an important reference for the global active CO2 remote sensing and the application of DQ- 1 data.