Improving the application of UAV-LiDAR for erosion monitoring through accounting for uncertainty in DEM of difference

Successful control of erosion requires effective monitoring and an in-depth understanding of erosion processes over large areas. Data from repeated unmanned aerial vehicles (UAV) light detection and ranging (LiDAR) acquisitions have been one important option to derive erosion-related information across large areas for example by calculating a digital elevation model of difference (DoD). However, the lack of an efficient and practical DoD uncertainty derivation method constrains the use of repeated UAV- LiDAR acquisitions for erosion monitoring, particularly for erosion that induces relatively small changes (i.e. field scale erosion). This study employed repeated UAV-LiDAR acquisitions to promote the understanding of DoD uncertainties and integrate the gained understanding in an erosion monitoring approach. The uncertainty of DoD was derived based on the impacting factors (point density and topography) using the fuzzy inference system (FIS), with the rules established according to the empirical relationships between DoD uncertainty and impacting factors for one sloping field. The proposed method was then validated on another sloping field by using terrestrial laser scanning (TLS) data. Results showed that the absolute value of DoD uncertainty (DoD_ua) linearly increased with local roughness and decreased exponentially with increasing point density. The established DoD uncertainty derivation method was found to be able to account for over 96% of raw DoD uncertainty. A validation with TLS results showed that our method slightly underestimated erosion and deposition volume, respectively, (-0.29 m³ and 0.74 m³ compared to −0.94 m³ and 1.34 m³ measured by TLS) but the trends were comparable. The distribution of deposition (volume and area) in different parts of the validation slope was well captured by our method, while the distribution of detected erosion was less satisfactory. This may be explained by the fact that our method did not well capture subtle erosion, suggesting a direction for future improvement.