1. J A M E S A . K L A N G , P E
A N D R E W FA N G , P E
K I E S E R & A S S O C I A T E S , L L C
5 3 6 E . M I C H I G A N A V E . , S T E . 3 0 0
K A L A M A Z O O , M I 4 9 0 0 7
J K L A N G @ K I E S E R - A S S O C I A T E S . C O M
Does Precision Agriculture
Result in Consistent and Predictable
Nutrient
Loading Reductions?
2. Natural Resources Conservation Service
Conservation Innovation Grant Project
Project Lead:
American Farmland Trust
Project Title:
Coupling Precision Agriculture with Water Quality Credit Trading
Project Objectives:
Create, test, and define a Water Quality Credit Trading credit estimator to
incorporate Variable Rate Technology-based nutrient management crediting
into wastewater treatment plant trading programs.
Project Area:
Within Ohio, Kentucky, Indiana, and/or Illinois
This material is based upon work supported by the Natural Resources Conservation
Service, U.S. Department of Agriculture, under number 69-3A75-12-177. Any opinions,
findings, conclusions, or recommendations expressed in the this presentation are those of
the authors and do not necessarily reflect the views of the U.S. Department of Agriculture.
3. Project Team Collaborators
American Farmland Trust
Indiana State Department of Agriculture
John Deere
Kentucky Division of Conservation
Kieser & Associates, LLC
Ohio Department of Natural Resources
Ohio Farm Bureau
Ohio State University
Purdue University
Trimble
USDA Natural Resource Conservation Service
University of Kentucky
4. Overview
What is Water Quality Credit Trading?
Trading credit characteristics and requirements
Project approach
Precision Ag technologies
Limiting factors for nutrient management credit
generation
Assessment methodology
Project status
Next steps
5. What is Water Quality Credit Trading?
Water Quality Credit Trading (WQCT) is a flexible
U.S. EPA National Pollutant Discharge Elimination
System permit compliance option
Allows a new effluent limit to be met by purchasing
credits from other locations with equal or greater
reductions
Trading options:
Point source to point source trading
Point source to nonpoint source trading
Nonpoint source to nonpoint source trading
6. Water Quality Credit TradingWater Quality Credit Trading
Trading uses a Watershed Approach
Treatment plants treat to a baseline level before
being allowed to trade
Trade for a specific parameter, plant treats all others
WQCT allows flexibility and cost savings
WQCT provides greater protection of the ecosystem
than conventional treatment
7. Past Uncertainties for Trading Credits
Generated by Nutrient Management
Establishing a baseline (e.g., field history, county averages, or
comprehensive nutrient management plans?)
Weather variability introduced uncertainty
Differences in crop uptake over time
Yearly yield increases
Changes in crop rotation
Different application rates and timing from
Equipment upgrades
Fertilizer purchases
Manure management systems
Adequate data and record storage
Leakage (e.g., manure management must be for whole farm)
Drainage (numerous complications when present)
8. Midwest Nutrient Estimation Method;
Fields Experiencing Sheet & Rill Erosion
Region V model (a.k.a. STEPL)
Explanation in Michigan DEQ “Pollutants Controlled
Calculation and Documentation for Section 319
Watersheds Training Manual”
Revised Universal Soil Loss Equation (RUSLE)
Chemicals, Runoff and Erosion from Agricultural Management
Systems (CREAMS) nutrient enrichment algorithm
Default nutrient concentration values applied
9. Nutrient Enrichment
Particle size distribution changes during transport
Upland particles Edge-of-field particles
Sand, silt and clay bind phosphorus at different rates
10. Midwest Sheet and Rill Erosion Method
CREAMS enrichment algorithm:
For sediment-attached nutrients (includes
organically bound nutrients)
Estimates increase in soil nutrient concentrations
due to redeposition of coarse materials
Inputs: erosion rate, delivery ratio, and upland
nutrient concentration
Soil nutrient concentration default values:
Sand: 0.85 pounds of P per ton of sediment
Silt: 1.0 pound of P per ton of sediment
Clay: 1.15 pounds of P per ton of sediment
11. Precision Ag Technologies
Many different forms of Precision Ag exist:
● VRT nutrient applications
● On-the-go
● Zone mapping
● GPS tractor guidance systems
● VRT pesticide applications
● VRT seeding
● VRT irrigation controls
Project Focus is nutrient controls; credit estimator development:
● VRT nutrient applications
● Zone mapping
● GPS guidance systems
12. Project Approach
Collect data from operators that have a long-term VRT
history with records
Preference for sites with edge-of-field water quality
monitoring (difficulty finding VRT fields with monitoring)
Select a field-scale watershed model:
Considers: Provides field or edge-of-field:
● Agricultural inputs ● Yield response
● Soil characteristics ● NPS volume of runoff
● Crop dynamics ● NPS sediment loading
● Climate variability ● NPS nutrient loading
13. Modeling Approach
Use model to create multiple scenarios:
Vary weather patterns
Simulate different VRT and uniform application
nutrient rates
Perform a sensitivity analysis to identify which
inputs the model is most responsive to
Create a multiple linear regression equation based
on field-modeled estimates of NPS loading to create
an edge-of-field phosphorus and nitrogen credit
estimator
14. Model Selection Criteria
Primary project needs:
Appropriate for the Ag setting (e.g., considers
timing of equipment passes, application rates, crop
rotations,…)
Edge-of-field nonpoint source loadings for
sediment, nitrogen, and phosphorus
Additional desired attributes:
Robust crop yield estimates
Ability to model under extended weather datasets
15. The “4 R’s” of Nutrient Management
4 R’s for Nutrient Management
Right Source (balanced nutrients in management plan)
Right Rate (for N & P applied, based on crop needs)
Right Time (placed when the crop needs it)
Right Place (applied where the plant uptake occurs)
VRT crediting focus on changing the rate:
Assumed producer uses the right balance of all nutrients
Illustrated load reductions from timing and placement
16. The KY Farm Site (124 acres)
No-till over a decade;
VRT phosphorus
application in 2010.
Hydrologic Response
Unit #851 is a Lowel Silt
Loam with 5 to 10 %
slopes; low STP with
higher application rates
Hydrologic Response
Unit #1933 is a
Nicholson silt loam with
2 to 5 % slopes, high STP
with low application
rates
17. 4 R’s as Seen Through SWAT
Testing of 4 R’s
Develop equation using multiple linear regression
Sensitivity of nonpoint source edge-of-field loading
to changes 4 R input scenarios
Check for input statistical significance
Check multicollinearity of inputs
Estimate equation’s ability to
explain edge-of-field loading
18. SWAT Estimated Reduction for Right Rate
(Averaged Across Entire Field)
SWAT Scenarios
2010 & 2011 Cropping Years
Right Rate
P2O5 (lbs/ac)
Based on farm records includes VRT rates 59
Increased VRT rate to county average 99
2010 loading difference (Corn) 0.7
(+12.1%)
2011 loading difference (Beans) 1.1
(+8.9%)
19. SWAT Scenarios
2010 & 2011 Cropping Years Right Time
Based on farm records, applications
in the spring
Spring
2010
Switched nutrient application to fall
of prior year
Fall
2009
2009 fall application;
Increase in 2009
0.036 lb P/ac
+0.7%
2010 loading difference (Corn) 0.0084 lb P/ac
(+0.2%)
Increase 2010
2011 loading difference (Beans) -0.2488 lb P/ac
(-2.0%)
SWAT Estimated Reduction for Right Time
(Averaged Across Entire Field)
20. SWAT Scenarios
2010 & 2011 Cropping Years
Right Place
(Magic -
Incorporation into
no-till!)
Based on farm records includes VRT rates Broadcast
Increased VRT rate to county average Incorporation
2010 loading difference (Corn) -0.8 lbs P/ac
(-13.7%)
2011 loading difference (Beans) -1.4 lbs P/ac
(-10.9%)
SWAT Estimated Reduction for Right Place
(Averaged Across Entire Field)
21. Modeled Field Characteristics
Calibrated on yield
Sediment roughly calibrated to 2.6 tons/acre/yr
SWAT algorithms used to estimate water quality
results at edge-of-field
Highly SEDP and ORGP dominated NPS loading
(e.g., average for all corn years: 38% SEDP, 53 %
ORGP, 9% SOLP)
Silt loams modeled with 2 to 10 percent slopes
Phosphorus depletion driven by both erosion and
crop uptake
22. Expanded List of Scenarios;
Focus on Two Different Zones
Varied all hydrologic resource units to experience:
1. An initial available soil P at the 2007 soil test
value
2. An initial available soil P at the highest 2007
soil test result
3. An initial soil soluble P level at the lowest 2007
soil test result
4. A one-year precipitation and temperature shift
5. A VRT rate reduced by 5%
6. A two-week shift of precipitation and
temperature plus a VRT rate increase by 10%
23. Fluctuations Observed During a 40-Year
Weather Simulation
Years
Erosion
Rates
(tons/acre)
Mehlich 3
STP Test
Estimation
Results
Application
Rates
(lbs P2O5)
NPS TP
Edge-of-field
Loading
(lbs. TP/ acre)
1995 versus 1998 1.2 & 1.8 Low & Low 52 & 31 5.5 & 4.1
1973 +1975 + 1978
versus same
9 & 9
105 & 71
(Averages)
Same across
all six years
26.1 & 22.5
Supports long-term NPS loading reductions occur when practicing 4 R’s.
Therefore, any confusion occurs within the crediting constraints.
1975 versus 1975
(Two Scenarios)
2.7 & 2.7 105 & 72 31 & 31 10.4 & 9.1
1975 versus 1975
(Two Scenarios)
3.2 & 1.7
(Est. by 1-yr
weather shift)
68 & 64 52 & 52 8.9 & 2.1
The edge-of-field loading is dominated by SEDP and ORGP. Therefore, variability in
erosion create larger variability in loading compared to variability of STP.
24. SWAT OutputsSWAT Outputs Available Field EstimatorsAvailable Field Estimators
NPS Edge-of-field
Sediment Yield
Sediment Phosphorus
Organic Phosphorus
Soluble Phosphorus
USLE, RUSLE,
RUSLE2
STEPL, Region V
(CREAMS model
nutrient enrichment
estimate added to
USLE family estimates)
SWAT-Based Multiple Linear Regression
25. SWAT OutputsSWAT Outputs Available Field EstimatesAvailable Field Estimates
Cropping
Crop yield
Plant uptake
Fresh organic to mineral
P
Organic P to labile P
Labile to active P
Active to stable P
Application rate
Average yield
Soil test phosphorus
Estimates of average P
uptake per bushel
SWAT-Based Multiple Linear Regression
26. Validation of Selected Equation
Multiple linear regression equation developed on one HRU
and tested on a second
Setup on HRU #851, has higher slopes and lower STP initial
values
Validation on HRU #1933 with lower slopes and higher initial
STP values
Both Loam soils
Equation developed on High Mehlich 3 STP results
TPeof = 1.608 – 0.03 (STP) + 2.81 (SED)
Regression Statistics: R2 = 0.84, F = 77.4, Significance F = 2.35 E-12
Independent Variable Statistics: STP P-value = 0.0008
SED P-value = 1.35 E-12
27. Validation of High STP Based Equation
Validation
Site
Erosion
Rate Range
(tons/acre)
Mehlic3
STP test
Estimate
Results
Multiple Linear
Regression
Equation Result
Ranges
(lbs TP/ acre)
SWAT
Result
Ranges for
Same Years
Average Error
Across Ten
Corn Years (%)
0.6 to 3.3
(Average 2.3)
Very High
1.3 to 9.9
(Average 6)
1.6 to 11
(Average 7.1)
17% Under
Estimated
1.5 to 4.1
(Average 2.5)
Very Low
5.6 to 13
(Average 8.8)
2 to 6
(Average 3.8)
131% Over
Estimated
TPeof = 1.608 – 0.03 (STP) + 2.81 (SED)
Compared to SWAT model HRU #1933 results
28. Comparison Findings for
VRT Based STP Management
Long-term VRT applications reduce long-term
nutrient loading but not always yearly loading
Lowering STP results takes time (Randle, 1997),
(Mallarino, et al., 2011), (Hanson et al., 2002)
Field erosion rate has the greatest influence on NPS
nutrient edge-of-field loading
Application rate increases show up for two years
Prediction equations like Region V model need
calibration
29. Implications for Trading
Current use of long-term erosion averages is
appropriate
Verification of credits can be done by STP
measurements
Use of default inputs introduce higher uncertainty
Nutrient management practices and VRT based on
zone mapping can be credited (STP part of the mgt)
On-the-go VRT application rates need to be
associated with erosion predictions if trading
30. GPS Applicator Guidance Controls for
Section Boom; Courtesy of John Deere
NH3 Swath Control Pro
Purpose
Develop multi-section on/off system to allow for greater
control of NH3 and High Speed Low Draft machine
Goals
Increased application precision
Product savings
Reduced operator strain
Retain Original distribution accuracy
Instant on/Instant off (eliminate gassing on ends)
Develop monitoring system
31. Swath Control System Overview
Each opener has an additional on / off valve
9 section system on 15 opener bar
Six two-opener sections on outsides
Three single opener sections in middle
32. Swath Control System Overview
• On/off valve every opener
• Pressurized NH3 hose from
manifold to valve
35. $325 Product Savings or $3.25 / Acre
(5 acres of overlap removed;
1000 lbs less NH3 applied)
Swath Control System Results
What does this mean?
100 acre actual field size
201 lbs NH3 (165 units)
$650 / ton NH3
Without Swath Pro
105.5 applied acres = 10.60 ton
10.60 x $650/ton = $6890.00
With Swath Pro
100.5 applied acres = 10.10 ton
10.10 x $650/ton = $6565.00
36. Next Steps
Add a scenario that balances STP results across the 40-
year weather simulation
Complete data gathering from a field in Illinois
VRT practiced on N and P
Long-term record with tile water monitoring
Setup and calibrate SWAT
Test Kentucky equations on the Illinois field
Develop crediting protocol for GPS guidance systems
Make recommendations for calibration of multiple linear
regression equation and Region V model
Develop crediting protocol for zone mapping VRT