LIDAR DERIVED SALT MARSH TOPOGRAPHY AND BIOMASS: DEFINING ACCURACY AND SPATIAL PATTERNS OF UNCERTAINTY

Abstract. As valuable and vulnerable blue carbon ecosystems, salt marshes require adaptable and robust monitoring methods that span a range of spatiotemporal scales. The application of unmanned aerial vehicle (UAV) based remote sensing is a key tool in achieving this goal. Due to the particular characteristics of tidal wetlands, however, there are challenges in obtaining research and management relevant data with the requisite level of accuracy. In this study, the spatial patterns in uncertainty stemming from scan angle, binning method, vegetation structure and platform surface morphology are examined in the context of UAV light detection and ranging (LiDAR) derived digital elevation models (DEM). The results demonstrate that overlapping the UAV flight paths sufficiently to avoid sole reliance on LIDAR data with scan angles exceeding 15 degrees is advisable. Furthermore, the spatial arrangement of halophyte species and marsh morphology has a clear influence on DEM accuracy. The largest errors were associated with sudden structural transitions at the marsh channel boundaries. The DEM_mean was found to be the most accurate for bare ground, while the DEM_min was the most accurate for channels and the middle to high marsh vegetation (MAEs = −0.01m). For the low to middle vegetation, all the trialled DEMs returned a similar magnitude of mean error (MAE = ± 0.03m). The accuracy difference between the two vegetation associations examined appears to be connected to variations in coverage, height and biomass. Overall, these findings reinforce the link between salt marsh biogeomorphic complexity and the spatial distribution and magnitude of LiDAR DEM error.