A GIS focal approach for characterizing gully geometry

Abstract

AbstractIn this work we examined the explanatory factors of gully geometry in the Campiña landscape in southern Spain using a geographic information system (GIS) focal approach with regional topographic products (1.5 m LiDAR [light detecton and ranging] digital elevation model [DEM] for elevation data and 0.5 m orthophotography for gully limit delineation). We evaluated the performance of a focal approach (based on neighbourhood calculations) using topographic factors for characterizing the gully linear extent and cross‐section (XS) morphology (i.e., depth, D, width, W, mean slope, S and width‐to‐depth ratio, WDR). As for the longitudinal analysis, focal variables (e.g., using the 11 × 11 mean thalweg slope and 75‐m downstream mean slope) outperformed the pixel‐based approach, with accurate spatial predictions for gully headcuts and mouth locations. The threshold between gullied and non‐gullied areas in a slope‐area diagram took the form of a break‐line rather than a straight limit. For estimating the main XS dimensions (D and W), the Half‐width method was applied by selecting the closest point to the thalweg at the gully limits. The tendency to underestimate XS depths was corrected by applying the equations derived from an assessment of the sampling errors in DEMs datasets. The 50 m adjacent slope to the gully was the best explanatory descriptor of both XS dimensions (due to energetic and geometric reasons) along with the channel density, drainage area and total gully length. The XS mean slope and WDR were controlled by the resistance of the materials (investigated through geotechnical and vane‐test surveys) where incision processes were dominant (V‐shaped XSs in upstream and middle reaches). As sedimentation started to become significant, both morphological parameters changed abruptly and XSs tended to show a trapezoid shape. An adaptative focal strategy with varying window sizes produced better results than the pixel‐based approach, requiring larger calculation scopes in areas of greater DEM uncertainty.