The document summarizes Iowa Daily Erosion Project 2, which aims to improve upon IDEP 1 by using remote sensing data to better estimate soil erosion at a higher resolution watershed scale. IDEP 2 will use LiDAR data to define hillslope profiles, NEXRAD radar to provide hourly precipitation data at a 1km resolution, and Landsat imagery to identify crop rotations and residue levels on agricultural fields across Iowa. The new model will run at the HUC12 watershed level to provide more localized estimates of sheet and rill erosion, runoff, and soil moisture on a daily basis.

1. Iowa Daily Erosion Project 2: Real time
soil and water resource inventory
Brian K. Gelder, PhD
Associate Scientist
Iowa State University
D. James, D. Herzmann, R. Cruse, J. Laflen, W.
Kraszewski, J. Opsomer, D. Flanagan, J. Frankenberger

2. Iowa Daily Erosion Project
• Our mission:
To help farmers, land managers, and the public
better understand the dynamics and magnitude of
runoff and soil erosion through daily estimation of
these processes on agricultural areas and
dissemination of the estimates via the web
These estimates are made using WEPP, the Water
Erosion Prediction Project model, a mechanistic
model designed for agricultural and forest plots.

3. • Township
level data
– Soils
– Slopes
– Manage
ment
NEXRAD
Precip
data

4. Iowa Daily Erosion Project 1
• Estimating
– Sheet and rill erosion
– Soil moisture
– Runoff
– Rainfall
• On 18,000 hillslopes
– 6‐20 per township
• Weather from NEXRAD/
Iowa Mesonet
• Slope, soils, and
management
information from 1997
USDA Natural Resource
Inventory (NRI)
– Extrapolated to present
• Estimates aggregated to
township level
• Debuted in 2003

6. Iowa Daily Erosion Project 2
• IDEP continues to function, but
– Rotations are out of date (1994‐97)
– Hillslopes are simple, uniform slope
– Actual locations not known
• Impossible to assess current accuracy
• Township structure not ideal for water issues
• Research in remote sensing makes it possible to
generate all model inputs
– Run at HUC12 watershed scale on approximately 250
flowpaths per watershed
• 1700 HUC12 watersheds, each approximately 100 km2

7. NEXRAD
Precip
LiDAR
Slopes
SSURGO
Soils
Remote
Management

8. Iowa Daily Erosion Project 2
• New Interface
– Similar to Google Earth, Bing Maps
• Selectable basemaps, overlays, data layers
• Transition to remotely sensed inputs
– Rainfall – 1x1 km 5 minute NEXRAD
• Upgraded from 4x4 km 15 minute NEXRAD
– Management – Remote sensing of crops/residue cover
– Hillslope profile – LiDAR derived
– Soils – SSURGO derived
• Enables field and watershed monitoring comparisons
for estimating model accuracy and precision
– Eventually stream delivery and channel erosion

9. IDEP 2 Rainfall
• Upgrade to Level II
NEXRAD radar data
from the NWS and UI
– 1 x 1 km rainfall
– Every 5 minutes
• Previously
– 4 x 4 km rainfall
– Every 15 minutes
• Improved wind, temp,
and solar spatial
resolution

10. IDEP 2 Management
• Estimates are currently made on
agricultural land parcels greater
than 10 acres
• To accurately delineate
agricultural land we began with
pre‐2008 publically available
USDA FSA Common Land Units
• A CLU has
– Permanent, contiguous boundary
– Common land cover &
management
– Common owner
– Common producer
• Does not represent crop
boundaries
• Aerially truthed to 2009 field
boundaries

11. IDEP 2
Management
• To obtain crop
rotations we use
USDA NASS
Cropland Data
Layer
– Annual map of
crop cover
– Based on Landsat
& other images
– Does not delineate
common
management
practices
2011
20092008
2010
20132012

12. IDEP 2
Management
2011
20092008
2010
20132012
For each 2009 field boundary the
majority land cover and fraction is
calculated for each year from
2008‐2013 and the pattern is
assigned to one of nine major
rotation types and extrapolated to
the current year.
+

13. IDEP 2 Management
2008‐2013 Crop Rotations

14. IDEP 2 Management
• Remotely sensed residue
cover estimates will come
from Landsat TM and
ETM+ sensor
• Soil and residue reflect
differently in middle
infrared
• Index relationship can
define tillage intensity
• Index uses crop cover
from previous step to
improve estimates

15. IDEP 2 Management
Post Fall Residue Polygons NDTI = Band 5 – Band 7
Band 5 + Band 7

16. IDEP 2 Management
• Determining residue cover (RC)
– Identify and download all imagery
from the Landsat archive over Iowa
from October 15 to June 15 from 2000‐
01 to 2011‐12
– Use Fmask, an automated process for
classifying imagery into snow, cloud,
cloud shadow, water, and clear sky
pixels
– Visually inspect Fmask results to find
any missed clouds and create
additional cloud/shadow masks if
needed
– Calculate Normalized Difference Tillage
Index (NDTI) value to estimate RC

17. IDEP 2 Management
• Steps to determine residue cover (RC)
– Utilize date, crop, growing crop (and
potentially soil) adjusted NDTI values to
calculate residue cover
• 100% NDTI value decreases with time
• 100% decreases from corn ‐> beans
• 0% ‐ consistent across date & crop
• Growing crop increases values
• No till class may be needed
– Utilizing all available imagery minimum
residue cover for each field is estimated in
two time periods
• After fall tillage (before March 15)
• After planting (before June 15)
– Based on residue cover estimates, corn
and soybean tillage practices are assigned
to 1 of 4 possible regimes
• No‐till
• Mulch‐till (2 levels)
• Conventional tillage
y = 507.95x - 23.333
R² = 0.8063
0
10
20
30
40
50
60
70
0.000 0.050 0.100 0.150 0.200
AVG_RCPLAN
NDTI_Corn
y = 165.95x + 0.0596
R² = 0.1142
0
10
20
30
40
50
60
70
0.000 0.050 0.100 0.150 0.200
AVG_RCPLAN
NDTI_Soybean

18. IDEP 2 Hillslope Profiles
• LiDAR topographic data
can better predict water
flow across the landscape
but LiDAR is not perfect –
– Doesn’t always flow
• Roads/railroads create digital
dams
• Areas of no returns (water,
heavy trees) create errors
– Enforcement is needed to
get water to the channels
or depressions and not
back up in fields

19. Fill Depth at 3 m Resolution

20. Best Connection(s)

21. Remaining Fill Depth

22. IDEP 2 Hillslopes (Cont.)
• Once surface is enforced
further processing begins
– Too many flowpaths to run
them all
• Random sampling
– Stratify HUC12s into
subcatchments
• Using Douglas‐Peuker constant
drop stream analysis
• 100‐250 subcatchments per
HUC12
– Select 1 ag flowpath per
subcatchment

23. Catchments and Channels

24. Field Level Flowpaths

25. Field Level Flowpaths and Soils

26. HUC12 Framework

27. IDEP2 Rainfall

28. IDEP2 Erosion

29. IDEP2 Runoff

30. Questions?
• Acknowledgements:
– Iowa State University Agronomy Department
Endowment
– USDA Agricultural Research Service
– Environmental Defense Fund

31. IDEP 2 Management
June 1 2011 Residue Cover Index