2. Background
WEPP hillslope validation
WEPP and RUSLE2 study
• Part 1 – Climate comparisons
• Part 2 – WEPP & RUSLE2
simulations
• Part 3 – Results
Summary
3. Erosion prediction technologies are often used
to assess soil loss rates under current land
management practices, and effects of changes
to that management.
RUSLE2 (Revised Universal Soil Loss Equation
version 2) is the current technology being used
by USDA-NRCS for erosion prediction and soil
conservation planning.
WEPP (Water Erosion Prediction Project) model
is being implemented by USDA-NRCS to replace
RUSLE2.
NRCS and others need to understand potential
differences when using these 2 different models.
4. Zhang et al., 1996. Evaluation of WEPP runoff and soil
loss predictions using natural runoff plot data.
WEPP tended to over-predict soil loss for small
events with low erosion rates and under-predict soil
loss for large events with higher erosion rates.
Means of event and annual soil loss were well
predicted.
Tiwari et al., 2000. Evaluation of WEPP and its
comparison with USLE and RUSLE.
Compared WEPP average annual soil loss without
calibration with USLE and RUSLE.
WEPP performed quite acceptably, at similar levels
to both USLE and RUSLE.
5. WEPP continuous simulations were conducted
using NRCS WEPP for 11 USLE validation sites
(1930–1970s).
Bethany, MO; Castana, IA; Geneva, NY; Guthrie, OK; Holly
Springs, MS; Madison, SD; Morris, MN; Pendleton, OR; Presque
Isle, ME; Tifton, GA; Watkinsville, GA
Climate files were in breakpoint format.
Managements: Tilled-fallow, single crop or crop
rotations.
Erodibility, critical shear, and effective hydraulic
conductivity values were based on WEPP
parameterization equations (NO CALIBRATION).
6.
7.
8.
9. 1. Srivastava et al., 2017. Comparison of soil loss predictions from RUSLE2 and
WEPP in the U.S. under different cropping systems. (21 locations x 3 soil
types x 2 managements = 126 runs)
10. Cooperative effort by the USDA-ARS National Soil
Erosion Research Laboratory (NSERL) and the
National Sedimentation Laboratory (NSL).
The study is composed of 3 parts:
1. Climate evaluations for the two models using inputs for each
derived from the same observed weather station data.
2. Detailed WEPP and RUSLE2 model simulations at the same
locations using the same slope length, slope gradient, soil,
and cropping/management inputs.
3. Evaluation of results, including comparisons of long-term
average annual soil loss.
11. Obtain climate data from weather stations in Iowa with
available 15-min or finer resolution precipitation information
to create breakpoint precipitation inputs for WEPP.
Monthly EI30 values, EI distribution and average annual R
values will be determined, using RUSLE2 rules.
WEPP will be run using base CLIGEN input files and
breakpoint input files using various precipitation
resolutions.
RUSLE2 will be run for the same locations using the base
RUSLE2 climate inputs as well as newly derived values
from the 15-min precipitation data.
Unit plot conditions will be used, with a silt loam soil under
tilled fallow. Comparisons of each model’s results for
different climate inputs, as well as between models will be
made.
12. Average annual soil loss and runoff from WEPP forced
by CLIGEN (simulated) and NCDC (observed) data.
AREA SOIL LOSS (T/ac-yr) RUNOFF (in/yr)
(County) (CLIGEN) (NCDC) (CLIGEN) (NCDC)
All Stations 45.0 31.4 9.8 7.4
Adair 48.4 37.0 9.8 8.1
Des Moines 50.0 36.1 11.0 8.7
Hardin 53.0 31.3 11.0 8.3
Jackson 38.9 29.3 10.1 8.2
Plymouth 34.7 32.6 7.3 7.7
WEPP soil loss predictions were reduced by 6% – 69%
when using observed breakpoint climate inputs (15-
min data)
13. Average annual EI for Iowa from minimally screened
NCDC stations (all storms included) for 1970-2013.
270
260
250
240
230
220200
150
150
RUSLE2
database
EI = 150 RUSLE2 EI values derived
from the 15-min data were
44% - 60% greater than
those in the existing RUSLE2
database
25. WEPP validation
WEPP validation was performed using NRCS WEPP interface for 11
USLE plots consisting of different landuses.
On an average annual basis:
WEPP predicted runoff and soil loss ~ measured data.
WEPP and RUSLE2 comparisons
A study to compare soil erosion predictions by 2 different USDA
technologies in Iowa was developed.
The first part of the study on climate inputs to WEPP and RUSLE2 is
incomplete.
Preliminary results show that soil loss from:
CLIGEN-generated data > observed 15-min precipitation data
Newly derived RUSLE2 EI > existing RUSLE2 EI
Relative differences in soil loss predictions between WEPP and
RUSLE2 increase with increasing model complexity
fallow-tilled < terrain and management < cropping systems
26. Using existing climate inputs:
WEPP predictions > RUSLE2 predictions, except for
no-till soybean management systems.
For tilled-fallow conditions,
WEPP predicted soil loss values were 24% greater
than RUSLE2 predicted soil loss across all climates
and soils.
Differences in mean soil loss between WEPP and
RUSLE2 increased as slope length and slope gradient.
More work is needed, especially on climate input
evaluations and comparisons, and slope effects.
27.
28.
29. Part 2 – WEPP & RUSLE2 simulations (Group 2)
WEPP soil loss was 78% higher than RUSLE2 soil loss
across all climates, soil textures, and managements.
Both WEPP and RUSLE2 showed trends of decreasing
soil loss with increasing crop yields for each soil.
WEPP showed more variability in soil loss with climate
for different soil textures compared to RUSLE2.
WEPP soil loss for corn and soybeans with fall plow,
fall chisel, spring plow, and spring chisel tillage
systems were higher compared to RUSLE2 soil loss.
Under no-till soybean cropping systems, RUSLE2
showed higher soil loss, whereas under no-till corn
cropping systems, WEPP and RUSLE2 showed similar
ranges of soil loss.
30.
31. Part 1 – Climate comparisons
Only part of this work has been completed. We are also
still processing finer resolution (1-min) precipitation data,
to use in more comparisons.
Generally, results indicate that WEPP model simulations
using the breakpoint precipitation inputs (observed 15-min
precipitation data from 1970-2013) are less vigorous than
those predicted using CLIGEN-generated inputs to WEPP.
WEPP soil loss predictions were reduced by 6% - 69% when
using observed breakpoint climate inputs (15-min data).
In terms of computed RUSLE2 EI factors using observed
15-min precipitation data from 1970-2013), values are
substantially more vigorous than those in the existing
RUSLE2 database.
RUSLE2 EI values derived from the 15-min data were 44% - 60%
greater than those in the existing RUSLE2 database.
WEPP soil loss predictions are generally greater than those of RUSLE2 for both Benchmark and Alternative management. This may partially be due to the influence of the updated climate data used.
Compared to RUSLE2 results, WEPP tends to predict greater effectiveness from Alternative managements relative to Benchmarks in reducing soil loss.
43% higher soil loss from CLIGEN
32% higher runoff from CLIGEN
WEPP and RUSLE2 soil loss were similar for the Silt Loam (SiL), Loam (L), and Clay Loam (CL) textures, whereas WEPP predicted soil loss for Sand (S), Sandy Loam (SL), Silty Clay (SiC), and Clay (C) textures were higher compared to RUSLE2 soil loss.
Differences in WEPP and RUSLE2 predicted soil loss for the tilled-fallow conditions might be arising from the combined effects of differences in precipitation characteristics and soil erodibility, and approaches used in both models.
52% of simulation runs were between -2 and +2 T/ac/yr differences in soil loss. 80% were within -10 to +10.
Distribution of differences was not normal. Some few (<20% of simulations) where WEPP predictions were much greater than RUSLE2 predictions affected numerical statistical results.
Overall, WEPP predicted soil loss values were 43% greater than RUSLE2 predicted soil loss across all climates, soil textures, and managements.
Both models showed trends of increasing soil loss with increasing slope lengths and slope gradients.
WEPP predicted soil loss was higher for corn and soybeans under all tillage systems except no-till, where RUSLE2 predicted soil loss was higher.
WEPP and RUSLE2 soil loss at lower slope lengths (≤72.6 ft) were similar. WEPP predicted soil loss was higher than RUSLE2 for slope lengths >72.6 ft.
Differences in mean soil loss between WEPP and RUSLE2 increased as slope length and slope gradient increased possibly due to increases in rill detachment contributions (as opposed to interrill dominant conditions on gentler slopes) in WEPP soil loss predictions.
WEPP predicted soil loss was higher for corn and soybeans under all tillage systems except no-till, where RUSLE2 predicted soil loss was higher.
WEPP and RUSLE2 soil loss at lower slope lengths (≤72.6 ft) were similar. WEPP predicted soil loss was higher than RUSLE2 for slope lengths >72.6 ft.
Differences in mean soil loss between WEPP and RUSLE2 increased as slope length and slope gradient increased possibly due to increases in rill detachment contributions (as opposed to interrill dominant conditions on gentler slopes) in WEPP soil loss predictions.
Both WEPP and RUSLE2 showed trends of decreasing soil loss with increasing crop yields for each soil.
WEPP showed more variability in soil loss with climate for different soil textures compared to RUSLE2.
WEPP soil loss for corn and soybeans with fall plow, fall chisel, spring plow, and spring chisel tillage systems were higher compared to RUSLE2 soil loss.
Under no-till soybean cropping systems, RUSLE2 showed higher soil loss, whereas under no-till corn cropping systems, WEPP and RUSLE2 showed similar ranges of soil loss.
increased possibly due to increases in rill detachment contributions (as opposed to interrill dominant conditions on gentler slopes) in WEPP soil loss predictions.
Management #1 involved chisel disk tillage ahead of each crop.
Management #2 involved chisel disk tillage after corn harvest, but no-till corn planting following soybean harvest.
Management #3 involved no-till management for both crops.