CSDWC: A Cross-Source Waveform Cooperation Method for Estimating Forest Canopy Height Change From GEDI and Airborne LiDAR-Simulated Waveforms

Abstract

Integrating airborne light detection and ranging (LiDAR) with the global ecosystem dynamics investigation (GEDI) offers an effective way to characterize forest canopy height changes. However, the accuracy of conventional change estimation methods is often limited by systematic geolocation errors, intersensor inconsistencies, and signal coupling among canopy, ground, and noise components. To address these challenges, we develop a footprint-level canopy height change extraction framework by integrating GEDI observations and airborne LiDAR-simulated waveforms. First, a two-stage optimization algorithm that combines digital terrain model (DTM) matching and ground-return waveform matching is introduced to correct systematic geolocation errors in GEDI footprints. Second, a novel cross-source differential waveform cooperation (CSDWC) method is proposed to retrieve footprint-level canopy height changes by jointly analyzing GEDI waveforms and airborne LiDAR-simulated waveforms. This method reduces intersensor bias while improving the detection of weak canopy signals. Results show that the proposed two-stage optimization algorithm effectively corrects systematic geolocation errors, compensating for a mean absolute spatial offset of 10.07 m. The CSDWC method achieves reliable footprint-level canopy height change estimation over the five-year study period, with an overall root mean square error (RMSE) of 3.72 m and mean absolute error (MAE) of 2.07 m, representing reductions of 1.92 m (34.04%) in RMSE and 1.09 m (34.49%) in MAE compared with the conventional relative height (RH) metrics-based differencing approach. Overall, the proposed framework substantially improves the accuracy of forest canopy height change monitoring and provides a robust methodological basis for forest carbon stock and dynamics research.