Yil Me Hu Spring 2024 - Nisqually Salmon Recovery Newsletter
August 31 - 0153 - Zhiming Qi
1. Modeling Phosphorus Losses from Tile-Drained
Cropland using RZWQM2-Phosphorus Model
Peng Pan1
Zhiming Qi1
Tiequan Zhang2
Liwang Ma3
1. Department of Bioresource Engineering, McGill University, Sainte-Anne-de-Bellevue, Quebec, Canada.
2. Harrow Research and Development Center, Agriculture and Agri-Food Canada, Harrow, Ontario, Canada.
3. USDA-ARS Rangeland Resources and Systems Research Unit, Fort Collins, Colorado, USA
Presented by
Zhiming Qi
Associate Professor
McGill University
Canada
2. OUTLINE
• Introduction
• RZWQM-P Development
• P Pools
• P model Inputs
• P model parameters
• RZWQM-P Evaluation
DRP & PP in tile drainage and surface runoff in a
corn-soybean field treated with:
• Inorganic Fertilizer
• Liquid Cattle Manure
• Conclusions and Future work
2
3. INTRODUCTION
• Phosphorus pollution
• Nitrate, ammonia and phosphorus (P) are the top 3 substances
released to Canadian water bodies with 55,723.6 tonnes,
51,209.7 tonnes, and 6,053.4 tonnes per year (Environment
Canada, 2010).
• P is the limiting factor of algal bloom in surface water bodies
• Canadian farmers used 1.07 million tonnes phosphate mineral
fertilizers per year, usually with additional 180.96 million tonnes
of livestock manure in which 0.30 million tons of P according to
the 2006 census (Statistics Canada, 2011; 2018).
3
4. INTRODUCTION
• Phosphorus pollution
• Subsurface drainage is a major conduit
for P transported from agricultural soils
(International Joint Commission, 2018).
• In Lake Erie basin, 49% of soluble P and
48% of total P lost via tile drainage
(Smith et al., 2015b). In an Ontario study
this values were 95 to 97% (Tan and
Zhang, 2011)
From Dr. Gary Sands
4
5. INTRODUCTION
• Review of Existing Models
For simulating P loss through tile drainage (adapted from Radcliffe et al., 2015, JEQ)
Models
Drainage Macro-
pore
Phosphorus in tile Maure P
Pools
Manage-
ment
Free Control DP PP
ADAPT Y -- Y Y -- -- Limited
APEX Y Y Y Y -- -- extensive
DRAINMOD Y Y -- -- -- -- extensive
HSPF -- -- -- -- -- -- limited
HYDRUS Y -- Y -- -- -- limited
ICECREAMDB* Y -- Y Y Y Y extensive
P Indexes Y -- Y -- -- -- extensive
PLEASE Y -- -- Y Y -- Limited
SWAT Y -- Simple -- -- -- extensive
RZWQM2 Y Y Y -- -- -- extensive
* Not available yet. As of 2015, DRAINMOD-P was not available either.
5
6. INTRODUCTION
• Using RZWQM2 to develop a new P model
• Excellent management practices: water table management with
free/control tile drainage (by date), macropore, irrigation, manure (19
kinds), tillage (33 kinds), crop residue, N, DSSAT crop models
• Excellent hydrologic subroutine: Green-Ampt for infiltration, Richards
for soil water redistribution, Hooghoudt’s equation for drainage,
Shuttleworth-Wallace ET, freeze-thaw SHAW model, etc.
6
• Justification
• No model available (as of 2015) to simulate P losses via tile drains;
• Existing P models are based Jones et al. (1984) which is obsolete.
7. RZWQM-P Development
• P Pools
Plant
Uptake
Fresh Org P
Active Inorg P
Labile P
Stable Inorg P
Stable Org P
PP
DRP ManwiP
ManwoP
MansiP
MansoP
Manure P
Applied
Fertilizer P
Applied
AvFert P ResFert P
7
8. • Inputs
Precipitation
Temperature
Radiation
Wind
INPUTS
Weather
pH
Soil texture
Field Capacity
Saturated Conductivity
Organic Matter Content
Soil Porosity
Wilting Point
N and P Content
Soil
Fertilization
Harvest
Tillage
Crops
Agri.
Management
Location
Elevation
Field width
Field Length
Field Slope
Drain Depth
others
RZWQM-P Development
8
9. • Parameters
• Dissolved Reactive P (DRP)
• P Extraction Coefficient.
• Bubbling Pressure.
• Pore size distribution index.
• Particulate P (PP)
• USLE Parameters.
• Soil Replenishment Rate Coefficient
• Soil Detachability Coefficient.
• Soil Filtration Coefficient.
• Macroporosity, Fraction of Dead end pores, Average radius of the
macropore.
• Other Parameters
• Plant P uptake distribution parameter.
• Soil Root growth factor.
RZWQM-P Development
9
10. 10
• Phosphorus mitigation practices
RZWQM-P Development
•Water table control (controlled vs. free)
•Tillage: reduced tillage, no-till etc.
•P fertilizer application rate
•Cropping system
11. 11
Objectives
• To evaluate the performance of RZWQM2-
P in simulating P losses through tile
drainage water
• To assess long-term impacts of tillage on P
losses through tile drainage flow
12. • Field Experiment (compost & tillage)
• AAFC experimental site, Ontario (42° 12′ 15″ N, 82° 44′ 50″W)
• Drainage design: spacing 8.7 m; depth 0.6 m; 5 pipes per plot
• Factorial experimental design (compost x tillage)
Compost rate (CMP0 and CMP75):
0 and 70 Mg dry weight ha-1
Tillage (NT and CT) :
no till and
conventional tillage
RZWQM-P Evaluation Methods
12
Treatments:
NT-CMP0, NT-CMP75
CT-CMP0, CT-CMP75
15. • Field Experiment: Field Management
15
Year Crop
Tillage date Compost (leaf)
disk
Mold-
board
Date Rate
Organic
matter
N (g kg-1) P (g kg-1) C: N
(Mg ha-1) (g kg-1) Total NH4-N Total
Water
extracta
ble
1998 soybean 10-May 5-Nov 10-Dec 75 196 17.4 0.471 2.96 0.067 6.53
1999 soybean 2-May 15-Oct 21-Oct 75 480 16.0 0.033 2.08 0.082
17.4
0
2000* maize 10-May 15-Nov 8-Dec 75 338 16.7 0.25 2.51 0.075
11.9
7
2001 soybean 2-May 20-Oct No compost application
Table 1. Details on tillage and compost application at both experiment farms.
* Additional commercial fertilizers at the
rates recommended locally (200 kg N ha-1
and around 17 kg P ha-1) were surface-
applied around 30 April each year.
16. • Field Experiment: Drainage Water Sampling
16
• ISCO model 2900 (Lincoln, Nebraska, USA)
• One sample every 10,000 L in Conventional Tillage
one sample every 25,000 L in the No Till plots
• Water sample composition
24 samples (max.) per period
• P analysis: Dissolved and total P
Year Collection Period Year Collection Period Year Collection Period
Start Date End Date Start Date End Date Start Date End Date
1998 9/15/1998 2/3/1999 2000 4/25/2000 5/23/2000 2001 2/1/2001 2/14/2001
1999 2/3/1999 3/8/1999 5/23/2000 6/26/2000 2/14/2001 3/19/2001
3/8/1999 4/1/1999 6/26/2000 7/31/2000 3/19/2001 4/4/2001
4/1/1999 4/14/1999 7/31/2000 8/9/2000 4/4/2001 4/18/2001
4/14/1999 4/20/1999 8/9/2000 9/25/2000 4/18/2001 5/15/2001
4/20/1999 4/27/1999 9/25/2000 10/12/2000 5/15/2001 5/30/2001
4/27/1999 8/6/1999 10/12/2000 11/14/2000 5/30/2001 8/21/2001
8/6/1999 4/25/2000 11/14/2000 12/20/2000 8/21/2001 10/16/2001
12/20/2000 2/1/2001 10/16/2001 11/14/2001
Table 2. Aggregated periods for water samples
From Dr. Gary Sands
17. • Model Initialization
17
Soil layer Initial Soil P Calibrated soil hydraulic parameters
Labile Total Pb λ ksat klat
(m) g kg-1 g kg-1 (cm) (mm h-1) (mm h-1)
0-0.01 0.023 0.90 -20.00 0.22 4.5 2.5
0.01-0.20 0.021 0.90 -21.00 0.20 5.0 5.0
0.20-0.40 0.011 0.65 -21.50 0.20 5.0 5.0
0.40-0.60 0.005 0.50 -21.50 0.20 5.0 5.0
0.60-1.10 0.005 0.40 -16.64 0.20 1.9 1.9
1.10-3.00 0.001 0.10 -16.64 0.19 1.9 1.9
3.00-3.09 0.001 0.10 -16.16 0.19 0.1 0.1
Table 3. Initial soil P concentration and Calibrated soil hydraulic
parameters used in model
18. • Model Calibration & Validation
18
Table 4. Calibrated parameters for soil, tillage, and phosphorus cycle
• Parameters were calibrated manually against tile flow and P losses data
• Calibration treatment (CT-CMP75); validation (NT-CMP75 CT-CMP0 NT-CMP0)
Parameters
Calibrated
values
Parameters
Calibrated
values
Albedo Soil replenishment coefficient 1
Dry soil 0.5 Initial DRP in ground water reservoir (kg ha-1
) 14
Wet soil 0.7 Initial PP in ground water reservoir (kg ha-1
) 13
Crop at maturity 0.8 Plant P parameters
Fresh residue 0.22 Maize
Tillage Biomass P Fraction at Emergence 0.002
Moldboard -intensity 1 Biomass P Fraction at 50% Maturity 0.001
Moldboard -mix efficiency 0.25 Biomass P Fraction at Maturity 0.0008
Disk-intensity 0.4 P uptake distribution parameter 5
Disk-mix efficiency 0.5 Soybean
Macroporosity (m3
m-3
) 0.009 Biomass P Fraction at Emergence 0.004
P extraction coefficient 1 Biomass P Fraction at 50% Maturity 0.002
Soil filtration coefficient 0.1 Biomass P Fraction at Maturity 0.001
Soil detachability coefficient 0.4 P uptake distribution parameter 5
19. • Model Evaluation Criteria
19
Table 5. Statistical model performance evaluation criteria
Rating
Model accuracy evaluation statistics
|PBIAS| R2 IoA
Drainage Water flow
Satisfactory 10 - 15% 0.6 - 0.7 0.75 - 0.85
Good 3 - 10% 0.7 - 0.75 0.85 - 0.9
Very Good < 3% > 0.75 > 0.9
Phosphorus Loss
Satisfactory 15 - 30% 0.4 - 0.65 0.75 - 0.85
Good 10 - 15% 0.65 - 0.80 0.85 - 0.9
Very Good < 10% > 0.80 > 0.9
PBIAS: percent bias of the mean
R2: coefficient of determination
IoA: index of agreement
20. 20
RZWQM-P Evaluation Results
• Model Performance
Statistics
Calibration Validation
CT-CMP75 NT-CMP75 CT-CMP0 NT-CMP0
Drainage (mm)
Obs. mean 112.37 99.02 75.84 104.51
Sim. mean 102.71 107.07 86.21 96.49
Rating good good satisfactory good
DRP (g ha-1)
Obs. mean 181.62 361.83 45.89 57.38
Sim. mean 219.95 262.42 164.57 195.59
Rating satisfactory satisfactory unsatisfactory unsatisfactory
PP (g ha-1)
Obs. mean 347.11 323.32 274.1 361.73
Sim. mean 277.43 338.79 209.29 253.45
Rating satisfactory good satisfactory unsatisfactory
TP (g ha-1)
Obs. mean 555.32 760.92 331.38 433.60
Sim. mean 570.70 688.68 428.72 514.24
Rating very good good satisfactory unsatisfactory
Table 6. Model performance on simulating annual drainage
flow and P losses
21. 21
Statistics
Calibration Validation
CT-CMP75 NT-CMP75 CT-CMP0 NT-CMP0
Manure P 567 567 0 0
Fertilizer P 54 54 54 54
Residue P 23.11 22.30 23.09 22.38
Plant uptake P 51.30 47.41 51.30 47.36
DRP loss
Runoff 21.60 43.76 1.95 3.95
Drainage 0.84 1.02 0.62 0.78
PP loss
Runoff 3.17 5.82 0.97 1.93
Drainage 1.00 1.24 0.75 0.98
• Simulated P Balance
Table 6. P input and simulated output (P uptake and losses through tile flow and
surface runoff). Unit kg P/ha
22. 22
• Long-term impacts of tillage on P loss
Table 7. Tillage practices applied in RZWQM-P long-term simulation
Implement name Tillage intensity Mix efficiency
No till 0.00 0.00
Paraplow 0.20 0.05
Row cultivator
(Standard treatment)
0.25 0.10
Moldboard 0.93 0.30
One-way disk 0.40 0.40
Tandem disk 0.50 0.50
23. 23
a)
Tillage Intensity
0.0 0.2 0.4 0.6 0.8 1.0
Minimum
Bulk
Density
(g
cm
-3
)
1.00
1.05
1.10
1.15
1.20
1.25
1.30
1.35
c)
Tillage Intensity
0.0 0.2 0.4 0.6 0.8 1.0
Average
Actual
ET
(mm
day
-1
)
1.06
1.08
1.10
1.12
1.14
1.16
1.18
1.20
1.22
d)
Tillage Intensity
0.0 0.2 0.4 0.6 0.8 1.0
Average
Surface
Residue
Mass
(kg
ha
-1
)
3000
4000
5000
6000
7000
8000
b)
Tillage Intensity
0.0 0.2 0.4 0.6 0.8 1.0
Average
Infiltration
(mm
day
-1
)
1.64
1.66
1.68
1.70
1.72
1.74
1.76
y=0.3995x2
-0.6644x+1.337
R2
=0.9912
y=-0.0585x2
+0.1450x+1.6601
R2
=0.9972
y=--0.0943x2
+0.2230x+1.0759
R2
=0.9900
y=1804x2
-5734x+7075
R2
=0.9994
• Long-term impacts of tillage on P loss
24. 24
Conclusion
• RZWQM2-P model performed well in simulating annual
PP and TP loss through tile drainage compared with
observed data
• Simulation on DRP loss through tile drainage was
unsatisfactory for the no compost plots as it may
overestimated DRP loss from soil matrix
• model application showed that tillage could reduce tile
drainage and P loss in tile drainage compared with no-
till management.
• DRP loss simulation should be improved by adjusting
the manure P mineralization parameters and winter
drainage.
• Effects of tillage or other management practices on P
losses in RZWQM2-P model still need to be further
tested using more data.
25. 25
Presented by
Zhiming Qi
Brace Associate Professor
Department of Bioresource Engineering
McGill University
Montreal Area, Canada
zhiming.qi@mcgill.ca
Editor's Notes
Main connectivity factors: (a) surface drainage density and (b) tile drainage density between agricultural land and water bodies in the Great Lakes region of Ontario.
Temporal trends of risk of water contamination by phosphorus from agricultural land in the Great Lakes Watersheds of Canada. Canadian J. Soil Science 2011, 91(3): 443-453
Harmful Algae Blooms Plague Lake Erie Again
https://blog.nationalgeographic.org/2013/04/24/harmful-algae-blooms-plague-lake-erie-again/
In ICECREAM or any other models (except Surphos) the manure/fertilizer can be simulated just like adding P to the system. They don't have dedicated manure/Fertilizer P pools (Like we have four manure P pools, two fertilizer P pools) to simulated manure/fertilizer decomposition. They assume that when manure/fertilizer is added to the system they are instantaneously added to the soil P pools. In our case, it is at first added to manure/fertilizer P pools, then slowly with the decomposition, it is added to the soil P pools.
In ICECREAM or any other models (except Surphos) the manure/fertilizer can be simulated just like adding P to the system. They don't have dedicated manure/Fertilizer P pools (Like we have four manure P pools, two fertilizer P pools) to simulated manure/fertilizer decomposition. They assume that when manure/fertilizer is added to the system they are instantaneously added to the soil P pools. In our case, it is at first added to manure/fertilizer P pools, then slowly with the decomposition, it is added to the soil P pools.