A comparison of hillslope drainage area estimation methods using high‐resolution DEMs with implications for topographic studies of gullies

Abstract

Topographic models provide a useful tool for understanding gully occurrence in the landscape but require reliable estimates of gully head drainage areas. Modern highresolution topography data (collected using structure from motion photogrammetry or light detection and ranging) is increasingly used for topographic studies of gullies, but little work has been done to assess the variability of gully head drainage area estimates using different methods. This study evaluated alternative approaches to using high-resolution digital elevation models (DEMs) so that gully topographic models can be more readily applied to any area with suitably high-resolution data. Specifically, we investigated the impact of single-or multiple-direction flow routing algorithms, DEM hydrologic-enforcement procedures and spatial resolution on gully head drainage area estimation. We tested these methods on a 40 km 2 site centred on Weany Creek, a low-relief semi-arid landscape draining towards the Great Barrier Reef, Australia. Using a subroutine to separate gully heads into those with divergent or convergent flow patterns upslope, we found that divergent flow conditions occurred at half of 484 studied gullies. Drainage areas estimated by different flow routing algorithms were more variable in these divergent cases than for convergent cases. This variation caused a significant difference between topographic threshold parameters (slope b and intercept k) derived from single-or multiple-direction flow routing algorithms, respectively. Different methods of hydrologic enforcement (filling or breaching) also affected threshold analysis, resulting in estimates of the exponent b being $188% higher if the DEM was filled than if breached. The testing of the methods to date indicates that a finer resolution (≤2 m) DEM and a multiple-direction flow routing algorithm achieve the most realistic drainage area estimates in low-relief landscapes. For Weany Creek we estimated threshold parameters k = 0.033 and b = 0.189, indicating that it is highly susceptible to gully erosion.

Ephemeral Stream Network Extraction from Lidar‐Derived Elevation and Topographic Attributes in Urban and Forested Landscapes

“…For example, Walker et al. (2021) found various flow direction algorithms and hydrologic correction methods had an influence on estimating drainage area upstream from gullies. Thomas et al.…”

“…less than 5 m resolution). For example, Walker et al (2021) found various flow direction algorithms and hydrologic correction methods had an influence on estimating drainage area upstream from gullies. Thomas et al (2017) found 0.25-m horizontal resolution DEMs tended to over-predict runoff risk, and a horizontal resolution of 1-2 m is more optimal for modeling hydrologically sensitive areas in agricultural watersheds.…”

Delineation of the Luvuvhu Headwater Catchment using high-resolution Digital Elevation Models (DEMs)

“…To delineate the headwater catchment, the mosaicked dataset was resampled to 10m x 10m resolution. Higher resolution grids visualise better the headwater catchment delineated (Walker et al, 2021). The DEM processing was done using ArcGIS (ESRI ® ArcGIS Desktop: Release 10.8, Redlands, California, USA).…”

Achieving change through gully erosion research

“…However, further research is required to refine computational methods (e.g., Walker et al, 2021) and to develop monitoring routines. Image analysis based on machine learning holds promise for directly mapping gully occurrence across large areas from datasets that are more broadly available such as satellite imagery (Gafurov & Yermolayev, 2020;Wang et al, 2020), and current and historic air photographs (Ding et al, 2020;Phinzi et al, 2020).…”