Robust constrained retrieval for 1550 nm SNSPD-based single-channel elastic lidar via maximum correntropy unscented Kalman filtering

Eye-safe 1550 nm single-channel elastic lidar is attractive for routine aerosol monitoring because of its practical deployment advantages, yet its inversion remains fundamentally underconstrained and can be easily destabilized by large effective residuals in the preprocessed photon-counting observation space. To address this challenge, we develop a robust constrained retrieval framework for SNSPD-based 1550 nm elastic lidar that formulates the problem as sequential state estimation, jointly retrieves the aerosol backscatter profile and a time-varying effective column lidar ratio, and combines maximum correntropy criterion (MCC)-based unscented Kalman filtering with weak aerosol optical depth (AOD) closure. The framework is evaluated using mechanism-oriented observing system simulation experiments (OSSEs), an MPLNET-based product-level consistency test, and real daytime case studies from a 1550 nm SNSPD lidar system. In the OSSE ablation, the standard UKF is found to be highly vulnerable to the imposed contamination stress test, whereas MCC-based robust updating prevents contamination-driven divergence and yields stable retrievals across all four stages. Once robust updating is in place, weak AOD closure provides a further 11.8%–28.6% reduction in median aerosol backscatter RMSE relative to MCC-UKF alone. Cross-framework and real-case results further show temporally coherent retrievals and product-level vertical-structure consistency, with median profile-wise correlations exceeding 0.98 across four representative case-study days. These results show that the proposed framework improves the robustness, temporal continuity, and operational usability of single-channel elastic-lidar retrieval, while the real SNSPD cases should be interpreted as operational consistency demonstrations rather than independent truth-level validation of extinction accuracy.