Erosion modeling
The Water Erosion Prediction Project (WEPP) Model is a physically based
soil erosion model (Laflen Elliot, Flanagan, Meyer & Nearing, 1997),
and is particularly suited to modeling the conditions common in forests
(Elliot, 2013, Elliot & Robichuad, 2001;). The accuracy of WEPP
predictions, or predictions of any other erosion model, can be no
greater than the accuracy of the input data. A study of 33 agricultural
soils found the coefficient of variation of soil erodibility values, the
mean divided by the standard deviation, was around 30 percent (Elliot,
Liebenow, Laflen & Kohl, 1989). Because of the natural variability of
soils, Elliot, Hall and Scheele (2000) state that “The accuracy of a
predicted runoff or erosion rate is, at best, plus or minus 50 percent”
for a WEPP application for forest and rangeland erosion prediction. For
predictions of erosion on burned forest hillslopes, Robichaud et al.
(2007) have incorporated this variability into post fire erosion
predictions in the Erosion Risk Management Tool (ERMiT) application of
the WEPP model.
A geospatial application for the WEPP Model, the GeoWEPP ArcMap Wizard,
allows users to import and utilize their own detailed terrain, soil and
landuse information or to access publicly available spatial datasets
(Flanagan, Frankenberger, Cochrane, Renschler, & Elliot, 2013). The
GeoWEPP model version 10.3 was adopted in the ArcGIS 10.3.1 environment
for soil erosion modelling in the Emerald fire area. Topography, soil
and landuse data were preprocessed before running the model to ensure
the GIS projections were compatible. According to our results based on
local terrain, the 0.5-m LiDAR DEM would likely generate too many flow
paths, exceeding the maximum memory limitation in GeoWEPP running on a
PC. Therefore the 0.5-m DEM was resampled as a 2-m DEM with the ArcMap
Resample tool using the “Bilinear” method for hydrological process
simulation and soil loss prediction. The 30-m resolution soil and
landuse data were also resampled as 2-m raster files to be compatible
with the 2-m DEM using the “Nearest” method to maintain consistent
pixel values. The original downloaded soil data did not describe the
road surface erodibility conditions, so the soil properties for the road
surface pixels were revised to describe a paved road for those cells.
The GeoWEPP GIS wizard can be run in two modes, “watershed” and
“flowpath.” (Flanagan et al., 2013) When running in watershed mode,
GeoWEPP lumps the shallow lateral flow, surface runoff and erosion
response for each 2-10 ha hillslope polygon, and then routes the runoff
and sediment through the stream network for each storm. The output from
a watershed run is sediment loss and runoff for each hillslope and
channel segment. In the flowpath mode, GeoWEPP estimates the erosion for
each pixel in the DEM abstracting the erosion for that point from the
distributed erosion predicted by the WEPP model. Each individual
flowpath starts with a pixel adjacent to the channel and follows the
flowpath up to the ridge from that pixel, draining an area of 0.2 to 0.5
ha. The flowpath method is not intended to be used to predict sediment
delivery, but the output from a GeoWEPP flowpath run will include an
average erosion rate for each hillslope polygon. The “flowpaths” as
defined by GeoWEPP are not the same as the flow path lengths as defined
by the ArcMap Spatial Analyst for routing runoff as described later, and
although the GIS steps in their derivation are similar, they are not
identical (Flanagan et al., 2013;
http://desktop.arcgis.com/en/arcmap/) .
GeoWEPP first delineates channel networks using the TOPAZ model
(Flanagan et al., 2013). The TOPAZ model requires the user to specify a
critical source area necessary for a channel to form. A critical source
area (CSA) of 0.5 ha was necessary in order to generate a channel
network to link most of the hillslopes within the Emerald fire area. For
some hillslopes that were hydrologically independent from the adjacent
areas, a CSA of 0.3 ha was used to generate more detailed channels to
include these slopes. GeoWEPP then builds a watershed of hillslope
sub-catchments and individual flow paths that, along with channel
segment topography, serve as inputs to the WEPP Watershed version.
Offsite sediment yields for both pre-fire and post-fire conditions in
the study area were estimated based on sub-catchments (select
“Watershed” as the GeoWEPP simulation method). For the post-fire run,
a climate file was built for the year 2016 using the observed daily
temperature and precipitation data from the Echo Peak SnoTel station.
The pre-fire erosion modelling was carried out using long-term observed
data from the same station and run for 20 years to simulate the normal
forest conditions. For the purpose of evaluating the effect of road
segments on post-fire surface flow accumulation and sediment transport,
a flow path simulation was carried out and the onsite soil loss was
estimated for each 2-m pixel.