Geospatial scaling of runoff and erosion modeling in the Chihuahuan Desert

Authors: BALL, GRADY - Us Geological Survey (USGS); Mankin, Kyle

Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/17/2019
Publication Date: 11/15/2019
Citation: Ball, G.P., Douglas-Mankin, K.R. 2019. Geospatial scaling of runoff and erosion modeling in the Chihuahuan Desert. Applied Engineering in Agriculture. 35(5):733-743. https://doi.org/10.13031/aea.13275.
DOI: https://doi.org/10.13031/aea.13275

Interpretive Summary: Natural resource managers often assess the potential effects of planned land-use and management changes using models, such as the Rangeland Hydrology Erosion Model (RHEM). Topographic, plant and ground-cover data collected from small plots are used to define the specific model parameters used in RHEM to describe a small-scale area. We developed and tested a method to extend these plot-scale model parameters to a large, arid, grass-shrub rangeland landscape in southern New Mexico using publicly available geospatial data. We compared plot data to geospatial data for the same "cell" area. Plot and geospatial cell values generally agreed for elevation, surface slope, and percent total foliar cover but agreed less well for percent litter cover and not well for percent rock cover. These percent-cover values influence model parameters, which influences simulated runoff and soil erosion. As a result, RHEM-simulated runoff was similar between the plot-data model and the cell-data model, but simulated soil loss was different between the two models. These results demonstrate that we can use common geospatial data to extrapolate plot data to large landscape areas and provide reasonable simulation of runoff using RHEM. However, these same methods do not work for simulating soil loss, mainly due to the poor representation of litter and rock cover. Future research should assess whether fine-scale data collected using unmanned aerial systems (drones) could be a practical method to improve model parameters to simulate rangeland soil erosion for larger landscape scales.

Technical Abstract: Large-scale assessments of rangeland runoff and erosion require methods to extend plot-scale parameterizations to large areas. In this study, Rangeland Hydrology Erosion Model (RHEM) parameters were developed from plot-scale foliar and ground cover transect data for an arid, grass-shrub rangeland in southern New Mexico, and a method was assessed to upscale transect-plot parameters to a large landscape. The transect-plot data compared favorably to corresponding cell data generated from publicly available geospatial data for total foliar cover but less favorably for litter cover and poorly for rock cover. The RHEM effective hydraulic conductivity (Ke) parameter was comparable between transect-plot and geospatial-cell methods but the splash and sheet erosion factor (Kss) had poor agreement between the two methods. Simulated runoff and erosion reflected differences in transect-plot and geospatial-cell-based RHEM parameterizations, with low bias and very good agreement for runoff but high bias and poor agreement for soil loss. These results demonstrate that Ke parameters developed using geospatial data calibrated to plot data can be extrapolated to large spatial areas and provide reasonable simulation of runoff using RHEM. However, these same geospatial methods do not provide reasonable estimation of Kss or simulation of soil loss. Poor representation of litter and rock cover variables, which are highly spatially heterogeneous at the plot scale, were inadequate to accurately represent Kss or soil loss using RHEM. Future research should assess the potential to use higher-resolution litter and rock cover data, such as from unmanned aerial systems, to improve parameterization of Kss and, ultimately, arid rangeland soil erosion simulation.