Monitoring soil erosion on field slopes by terrestrial laser scanning and structure-from-motion

Terrestrial laser scanning (TLS) and structure-from-motion (SfM) photogrammetry are prevalent erosion monitoring techniques. However, their relative performance for field erosion monitoring remained unclear. In the study, five souring experiments were implemented on each of the three plots, consisting of hillslope and gully area, established on field slopes. The input flow for three hillslopes was 25, 55, and 85 L min(-1) respectively while that for the gully area was 10 L min(-1). TLS and SfM were used to monitor erosion processes, while their performance was assessed using manual measurements. Results showed that TLS and SfM produced generally sound and comparable soil loss estimates, with relative errors (REm) for the cumulative and consecutive sediment yield derived by the two techniques being -24.10% to +84.50% and -29.60% to +73.10% respectively. TLS estimated sediment yield more accurately than SfM for more intensively eroded areas, while the estimation accuracy of TLS was lower than SfM for less intensively eroded areas. On gully slopes, the two techniques broadly captured the spatial pattern of erosion and deposition, with inconsistency/errors emerging on areas with small topographic changes or caves; TLS generally produced more accurate rill length than SfM, with the REm for TLS being -2.18% to -0.16% and the REm for SfM being -7.09% to -1.49% respectively. TLS- and SfM-derived rill networks were broadly consistent (Kappa coefficient was 0.42-0.79) and different in details. The SfM- and TLS-derived level of detection was generally at the same order of magnitude. However, their spatial pattern on hillslopes was rather different, which informed a different erosion monitoring capacity of TLS and SfM on field hillslopes and needs further study. Overall, SfM provides a cost-effective and powerful alternative to TLS for field erosion monitoring and a combined use of them may produce more satisfactory results.