Vegetation-sediment interactions and bank stability in Atlantic rivers: Linking multispectral approaches and LiDAR data

Riparian and channel vegetation play a crucial role in fluvial sediment dynamics and riverbank stability. We investigated vegetation-sediment interactions along the Saja and Besaya rivers (northern Spain) through two complementary analyses combining LiDAR and multispectral imagery.

First, we assessed the influence of riparian vegetation on bank stability by delineating 100-m longitudinal reaches classified as erosive or stable. LiDAR data were used to quantify total woody vegetation cover, tree cover, and the mean vegetation height (as a proxy for maturity) across riparian strips of varying widths (10, 20, 30, and 50 m). Logistic regression models showed that riparian vegetation significantly reduced the probability of bank erosion, with minimal variability among predictors. The strength of these vegetation metrics was highest when calculated at a 10-m stripe, and it broadly decreased at 50 m, which suggests that vegetation adjacent to the bank was sufficient to mitigate erosion risk. These findings support the conservation and restoration of riparian woody vegetation as a potential nature-based solution for bank erosion prevention in sensitive areas.

Second, we examined in-channel sediment balances at 10x10 m pixels using multitemporal LiDAR (2018-2023) and Sentinel-2 imagery. Sediment balance was derived from LiDAR elevation changes and expressed as continuous (m³) and binary (erosion vs. non-erosion). We analyzed Sentinel-2 series to derive pixel history (yearly frequency and persistence of in-channel vegetation) and harmonic metrics (decomposing the NDVI series). Random Forest models indicated that pixels with decreasing vegetation persistence and negative NDVI trends were more likely to experience erosion. While predicting the magnitude of the balance was challenging, classification into erosion vs. stability achieved better performance. These results highlight the capabilities of multispectral image series to assess erosion and sediment processes in fluvial ecosystems and complement LiDAR data in river monitoring.

Overall, our study shows that riparian vegetation strongly influences bank stability and that vegetation dynamics within the channel are linked to sediment deposition and erosion. By integrating high-resolution LiDAR and multispectral imagery, we provide evidence that maintaining riparian vegetation can serve as an effective nature-based solution for reducing bank erosion processes. Furthermore, our approach underscores the value of combining remote sensing techniques to improve the understanding and management of sediment processes in highly dynamic river systems.