# Guidance for Tier I Estimation of Aqueous Pesticide Concentrations in Rice Paddies

May 8, 2007

## Tier I Rice Model v1.0 for Estimating Pesticide Concentrations in Rice Paddies

May 8, 2007

1. ### Introduction

This document describes a Tier I Rice Model (Version 1.0) for estimating surface water exposure from the use of pesticides in rice paddies. This screening-level model is based on the Interim Rice Model, which has been used in EFED for four years to estimate pesticide concentrations in rice paddies. The single, screening-level concentration calculated with this model represents both short and long term surface water exposure and can be used for both aquatic ecological risk assessments and drinking water exposure assessments for human health risk assessment. The formula of the Tier I Rice Model v1.0 is as follows:

Cw = (mai) / (0.00105 + 0.00013Kd)

and, if appropriate:

Kd = 0.01Koc

where:

Cw = water concentration [µg/L]
mai = mass applied per unit area [kg/ha]
Kd = water-sediment partitioning coefficient [L/kg]
Koc = organic carbon partitioning coefficient [L/kg]

Top of page

2. ### The Conceptual Model

The Tier I Rice Model v1.0 relies on an equilibrium partitioning concept to provide conservative estimates of environmental concentrations resulting from application of pesticides to rice paddies. When a pesticide is applied to a rice paddy, the model assumes that it will instantaneously partition between a water phase and a sediment phase. The aqueous concentration that results from such partitioning is described as follows:

(1)

Cw = (mai) / (Vw + msedKd)

where:

Cw = water concentration [mass/volume]
mai = mass of active ingredient applied to paddy [mass]
Vw = volume of water column plus pore water [volume]
msed = mass of sediment at equilibrium with water column [mass]
Kd = water-sediment partitioning coefficient [volume/mass]

Note that neither the degradation of the pesticide nor the mass transfer from the aqueous phase to the sediment is considered in this conceptualization, which greatly simplifies the model. The absence of degradation adds conservatism to the model (i.e., estimated concentrations should be higher than those usually found in rice paddies). The absence of mass transfer processes can either add or reduce conservatism depending on numerous conditions such as whether the pesticide is soil or water applied and whether actual degradation occurs preferentially in the sediment or the water compartment; however, consideration of this parameter is a refinement beyond a Tier I conceptualization.

Because it is more customary to describe a rice paddy in terms of depth rather than in terms of volume and mass, the following equations are defined:

(2)

msed = dsedb

(3)

Vw = dwA + dsedθsedA

where:

dsed = sediment depth [length]
dw = water column depth [length]
A = area of the rice paddy [area]
θsed = porosity of sediment [-]
ρb = bulk density of sediment [mass/volume]

Further, the input mass per unit area is defined as:

(4)

mai = mai / A

where:

mai = mass applied per unit area [mass/area]

Substitution of equations (2), (3), and (4) into (1) produces the conceptual model with commonly understood parameters:

(5)

Cw = (mai) / (dw + dsedsed + ρbKd) )

Top of page

3. ### Parameterization

The physical parameter values that describe the rice paddy for the Tier I Rice Model v1.0 represent a reasonably conservative system and are summarized in Table 1. The selection of these values was based to a large extent on typical properties of rice paddies. In some cases where such information was not available (e.g., sediment depth), the parameter values were calibrated from field data. The rationale for choosing these values is described in Sections III A, B, and C of this document.

Table 1. Physical Values of the Tier I Rice Model v1.0
ParameterSymbol ValueReference
Water column depth dw0.10 mUSDA (1998,2002)
Sediment depth dsed0.01 mCalibration described in Section III C of this document.
Organic content of sediment foc0.01Justification provided in Section III B of this document.
Sediment bulk density ρb1300 kg/m3Brady (1984), Hillel (1982)
Grain density ρg2650 kg/m3Density of quartz, typical assumption; Brady(1984), Hillel (1982)
Sediment porosity θsed0.509θsed = 1- (ρb / ρg)
1. #### Water Depth, Bulk Density, Porosity

The water column depth (dw) of 10 cm is a typical average rice paddy water depth at the time of flood release (USDA, 1998; 2002). Bulk density is assumed to be roughly the same as agricultural soils (Brady, 1984; Hillel, 1982). Grain density is assumed to be that of quartz (2.65 g/ml), and thus the porosity of the sediment is 0.509 (see relationship in Table 1).

2. #### Organic Carbon Content

The organic carbon content (foc) represents the mean of the organic carbon content in sediments taken from 15 rice fields (see Table 2). These sediments are described in 17 studies submitted in support of the registration of rice pesticides, and are assumed to be typical for rice paddies. Thirteen of the rice fields are located in the southeastern U.S. and two are located in California. The mean organic carbon content of the sediment in these 15 rice fields is 0.01.

The Interim Rice Model assumes an organic carbon content of 0.02. Interim Rice Model exposure estimates are compared with those of the Tier I Rice Model v1.0 for applications of 1 kg/ha of compounds of differing Koc to show how the exposure estimates are affected by the different modeled organic carbon contents (Appendix II). The Tier I Rice Model v1.0 exposure estimates are approximately twice those of the Interim Rice Model for compounds that have a Koc > 100,000. As compounds of decreasing Koc are modeled, partitioning to sediment becomes a less important dissipation process and Tier I Rice Model v1.0 exposure estimates approach those of the Interim Rice Model.

3. #### Water Calibration of Sediment Depth

The sediment interaction depth (dsed) was estimated by calibration using direct paddy measurements from 15 sites in 17 studies. The results of these studies are summarized in Table 2. The maximum observed concentration was that of the overlying water at any time after application. The application rate was the total amount of pesticide applied during the study. Multiple pesticide applications were typically made within one season in these studies. During a multiple application study, no attempt was made to associate concentration measurements with any particular application. Thus, there are cases where the highest concentration was observed prior to the final application. This situation is consistent with the Tier 1 model concept in which the timing of a measurement is not a consideration. Also consistent with the Tier 1 concept is the lack of differentiation among various application methods such as applications prior to flooding.

Table 2. Summary of Studies Used for Calibration of Sediment Interaction Depth.
SiteChemical Koc (L/kg)OC % ∼Depth (cm)Application (kg/ha) Observed Maximum (ppb)Reference / MRID
TXIprodione 4260.521.1248943718301
MSIprodione 4260.521.1255043718301
MSQuinclorac 361.26unknown0.841341063565
ARPropaconazole 6481.8100.3786142560501
ARPropaconazole 6480.63101.51221442560501
TXPropaconazole 6481.157.60.3784842560502
TXPropaconazole 6481.157.61.51213842560502
MSBentazon 350.7572.24207343431901
LABentazon 350.5372.24154543431901
MSLambda cyhalothrin 3000000.4790.18344367403
ARTrifloxystrobin 35001.1130.386945080803
ARMolinate 1860.64unknown10.1420040391706
TXMolinate 1860.29unknown10.127040391707
CAMolinate 1861.571011.258141421803
TXMolinate 1860.29unknown3.3624044970003
MSMolinate 1860.35103.36344970002
CAMolinate 1865.8153.361800Soderquist et al., 1977

The model sediment depth was calibrated by comparing maximum observed concentrations to model outputs based on a range of depths (0.5 cm to 3 cm). These results are shown in Table 3. The values at the bottom of the table summarize the number of simulated concentrations that are less than the observed concentration. A depth of 1 cm appears appropriate as the model simulates only three concentrations that are below the observed value, and two of these values underpredict the observed concentration by only 1 %.

Table 3. Calibration of the Sediment Depth (Concentrations Expressed as ppb).
Study ChemicalObserved Concentration Model Concentration
dsed = 0.5 cm
Model Concentration
dsed = 1.0 cm
Model Concentration
dsed = 2.0 cm
Model Concentration
dsed = 3.0 cm
Iprodione 489860698507398
Iprodione 550860698507398
Quinclorac 13801765703650
Propaconazole 61261200136103
Propaconazole 2141045799543411
Propaconazole 48261200136103
Propaconazole 1381045799543441
Bentazon 20732137204318781738
Bentazon 15452137204318781738
Lambda cyhalothrin 30.920.460.230.15
Trifloxystrobin 69115683726
Molinate 42008801780563645372
Molinate 2708801780563645372
Molinate 5819779867270715969
Molinate 2402934260221211791
Molinate 31293426022121 1791
Molinate 18002934260221211791
Number of modeled concentrations less than the observed 1346

1This concentration was observed 14 days after application.

Top of page

4. ### Formula of the Tier I Rice Model v1.0

Substituting the values in Table 1 into equation (5) and using common units for Kd (L/kg), mai (kg/ha), and Cw (µg/L), the Tier I rice conceptual model is simplified to the Tier I Rice Model v1.0 (see Appendix I for derivation):

(6)

Cw = mai / (0.00105 + 0.00013Kd)

where, in this case:

Cw = water concentration [µg/L]
mai = mass applied per unit area [kg/ha]
Kd = water-sediment partitioning coefficient [L/kg]

Model input values for the water-sediment partitioning coefficient (Kd) should represent a mean Kd of relevant soil (or sediment). A mean Koc value should be used to generate model input values for Kd in cases where sorption Kd values correlate with soil organic matter content. In these cases, Kd model input values should be calculated from the mean Koc using a fraction of organic carbon (foc) of 0.01 (Table 2). Kd can thus be estimated by the following equations:

(7)

Kd = focKoc

(8)

Kd = 0.01Koc

Alternatively, mean Koc values can be directly entered into equation (9):

(9)

Cw = mai / (0.00105 + 0.0000013Koc)

Top of page

5. ### Assumptions

Most of the assumptions used in this model help ensure that the outputs are protective of most environments associated with rice agriculture. The model assumptions include the following:

1. Movement of pesticide on suspended sediment is not considered.

3. Volatilization and other dissipation processes are not considered.

4. Partitioning to sediment is instantaneous.

5. Water is available for human or wildlife exposure instantaneously.

6. Water column depth is 10 cm.

7. Sediment depth is 1 cm.

8. All pore space is saturated with water.

9. Organic carbon fraction is 0.01.

10. Bulk density is 1300 kg/m3.

11. Grain density is 2650 kg/m3.

Top of page

6. ### Model Evaluation for Drinking Water Exposure

Tier I model estimates are screening estimates and, as such, are expected to exceed peak values found in the environment in most cases. This expectation is due to the protective assumptions of the model listed in Section V of this document, such as exclusion of degradation as well as dilution with uncontaminated water outside the paddy. In order to evaluate the Tier I Rice Model v1.0 for drinking water exposure estimation, modeled estimates, which are calibrated to indicate exposure within rice paddies, were compared to measured concentrations of pesticides in rice-growing environments well downstream of the actual rice paddies. The results of these comparisons are summarized in Table 4 and discussed in Sections VI A, B, C, and D of this document.

Table 4. Summary of Comparisons of Within-Paddy Simulations with Downstream Observations.
SiteChemical Koc (L/kg) Application Rate (kg/ha)Observed Maximum (ppb) Model Estimate (ppb)MRID or Reference
LA 1Fipronil 7270.0568.4128.1LDAF, 2000
LA 2Fipronil 7270.0562.1228.145349901
CAMethyl parathion 5230.890.1151244632501
CAThiobencarb 9094.512.3201044632501
CAMolinate 1861025.7781044632501
1. #### Louisiana Downstream Measurements 1 (LA 1)

The Louisiana Department of Agriculture and Forestry collected data from 23 monitoring sites where surface water bodies received rice paddy discharge water in Acadia, Calcasieu, Cameron, Jefferson-Davis, and Vermilion Parishes (LDAF, 2000). The maximum detected fipronil concentration from March 6, 2000 to May 15, 2000 was 8.41 µg/L. Although the actual use history of fipronil is unknown in this area, the labeled maximum application rate of 0.05 lbs a.i./acre was assumed in order to compare rice model results. The rice model gives an estimate that is 3.5 times greater than the maximum detected value.

2. #### Louisiana Downstream Measurements 2 (LA 2)

Fipronil surface water monitoring data were submitted for the Mermentau River and Lake Arthur (MRID 45349901). The Mermentau River drains a large portion of the rice acreage in southern Louisiana from the mouths of Bayou Plaquemine and Bayou Nezpique. There are 250,000 acres of rice in this area, 70% of which was treated with fipronil in 1999. The monitoring program was designed to provide a snapshot of concentrations on May 11, 1999, just after a 0.45-inch rainfall event, at 0 to 1 feet and 4 to 6 feet depth. The maximum detected concentrations were sampled from the mouth of the Bayou Plaquemine, where fipronil was detected at a maximum concentration of 2.118 µg/L and total residues of fipronil and three of its degradates were detected at a maximum concentration of 3.509 µg/L. The rice model gives an estimate that is approximately one order of magnitude greater than the maximum detected value for fipronil.

3. #### California Indirect Measurements (CA)

The California Department of Pesticide Regulation collected data from six monitoring sites in the Sacramento Valley that were sampled from March 31, 1997 to May 29, 1997 as part of a cooperative water quality monitoring program (MRID 44632501). Residues of molinate, thiobencarb, and methyl parathion were monitored in water bodies near rice agriculture that may not have necessarily contained concentrations representative of those found in rice paddy water or released paddy discharge water. Maximum concentrations of 0.11 µg/L of methyl parathion, 12.3 µg/L of thiobencarb, and 25.7 µg/L of molinate were detected in the Colusa Basin Drain monitoring site near State Highway 20 on the 13th , 20th, and 22nd of May 1997, respectively. The rice model gives estimates that are two to three orders of magnitude larger than the respective maximum detected values of these pesticides.

4. #### Evaluation Summary

Evaluation of the Tier I Rice Model v1.0 indicates that modeled within-paddy estimates are conservative, exceeding peak measured concentrations of pesticides in water bodies well downstream of rice paddies by less than one order of magnitude to multiple orders of magnitude. The range of modeled estimate exceedances over measured concentrations is most likely in response to chemical, environmental, and study-specific factors.

Top of page

7. ### Model Characterization

Estimates from the Tier I model generally do not represent typical concentrations found in human drinking water, as they represent paddy discharge water. However, these concentrations may be a reasonable estimate of acute concentrations for use in ecological assessment where exposure occurs at or near the rice paddy. In both cases, human drinking water and ecological exposure, the chronic concentrations as well as offsite concentrations are expected to be conservative. A higher tier rice model should be used to estimate chronic exposure to compounds that degrade rapidly into degradates that are not of risk concern.

If Tier I estimates calculated by this screening method do not exceed the level of concern in a risk assessment, there is high confidence that there will be little or no risk above the level of concern from exposure through water resources. However, because of the uncertainties associated with a screening method, when a level of concern is exceeded it cannot be determined whether the exceedance will in fact occur or whether this method has overestimated the exposure.

Sample input and output tables for use in ecological and/or drinking water exposure assessments are provided in Appendix III.

Top of page

8. ### References

Brady, N. C. 1984. The Nature and Properties of Soils, Ninth Edition. Macmillan Publishing Company, New York.

Hillel, D. 1982. Introduction to Soil Physics. Academic Press, Orlando, Florida.

LDAF. 2000. Summary of Actions Taken in the Monitoring of ICON® (Fipronil) Use in Rice Culture in South Louisiana. Louisiana Department of Agriculture and Forestry, AES Pesticide and Environmental Programs. June 15, 2000.

USDA. 1998. Crop Profile for Rice in California. United States Department of Agriculture, Pest Management Centers. Oct., 1998. Online at: http://www.ipmcenters.org/CropProfiles

USDA. 2002. Crop Profile for Rice in Louisiana. United States Department of Agriculture, Pest Management Centers. Sep., 2002. Online at: http://www.ipmcenters.org/CropProfiles

USDA. 2006. Physical Soil Properties: Sharkey Clay. United States Department of Agriculture, Natural Resources Conservation Service, Soil Data Mart. Jun. 19, 2006. Online at: http://soildatamart.nrcs.usda.gov

1. #### Submitted Studies

MRID 41063565. Winkler, V. 1987. Confined Field Aquatic Dissipation Study of 14C - BAS 514 H in Rice Paddy Water. Unpublished study performed and submitted by BASF Corporation Chemicals Division, RTP, NC. Study Number: 87/5064. May 1987. 20 pp.

MRID 42560501. Krueger, H., A. Hosmer, and S. McIninch. 1992. Dissipation of Tilt® in Two Arkansas Rice Fields. Unpublished study performed by Wildlife International Ltd., Easton, MD and EN-CAS Analytical Laboratories, Winston-Salem, NC; sponsored and submitted by CIBA-GEIGY Corporation, Greensboro, NC. Study Number: 108-261. November 9, 1992. 251 pp.

MRID 42560502. Krueger, H., A. Hosmer, and S. McIninch. 1992. Dissipation of Tilt® in Two Texas Rice Fields. Unpublished study performed by Wildlife International Ltd., Easton, MD and EN-CAS Analytical Laboratories, Winston-Salem, NC; sponsored and submitted by CIBA-GEIGY Corporation, Greensboro, NC. Study Number: 108-262. November 6, 1992. 248 pp.

MRID 43431901. Evans, J. 1994. Basagran® Herbicide Dissipation in a Rice Paddy: 1992 Study: Final Report. Unpublished study performed by Jensen Ag. Consultants, Inc., Washington, LA, Stoneville Associates, Inc., Greenville, MS, BASF Corporation, RTP, NC, and ALTA Laboratories, El Dorado Hills, CA; sponsored and submitted by BASF Corportation, RTP, NC. Study Number: 92111. October 19, 1994. 578 pp.

MRID 43718301. Chancey, E. 1995. An Aquatic Field Dissipation Study with Iprodione. Unpublished study performed by Rhone-Poulenc Ag Company, Leland, MS; Agvise, Inc., Northwood, ND; and South Texas Ag Research, Inc., Pattison, TX. Study Number: EC-92-187. June 20, 1995. 374 pp.

MRID 44367403. Patterson, S., P. Francis, J. Robbins, H. Storoni, and C. Spillner. 1997. Lambda-cyhalothrin: Aquatic Field Dissipation after Application to Rice in Mississippi during 1996. Unpublished study performed and submitted by Zeneca Inc., Richmond, CA. Study Number: LCYH-96-SD-01. August 19, 1997. 143 pp.

MRID 44632501. Department of Pesticide Regulation. 1997. Information on Rice Pesticides Submitted to the Central Valley Regional Water Quality Control Board. Cooperative Water Quality Monitoring Program review submitted by California Department of Pesticide Regulation. December 23, 1997. 94 pp.

MRID 45080803. Manuli, P. and B. Jacobson. 2000. Aquatic Field Dissipation of CGA-279202 under Field Conditions with Rice in Arkansas. (Volume 1). Unpublished study performed by Waterborne Environmental, Inc., Leesburg, VA, Mid-South Ag Research, Inc., Proctor AR, and Centre Analytical Laboratory, Inc., State College, PA; sponsored and submitted by Novartis Crop Protection, Inc., Greensboro, NC. Study Number: 283-98. January 5, 2000. 533 pp.

MRID 45349901. Theissen, R. 2001. ICON (Fipronil) Rice - Louisiana Surface Water Monitoring. Unpublished study performed, sponsored, and submitted by Aventis Cropscience, RTP, NC. Study Number: ICON-SWM-5/11/99. February 7, 2001. 36 pp.

Top of page

• ### Appendix I: Units Analysis

The Tier I Rice Model with commonly understood parameters (equation (5)) is:

Cw = mai / (dw + dsedsed + ρbKd) )

Substituting the parameter values for sediment depth (dsed=0.01 m), water depth (dw = 0.10 m), sediment bulk density (ρb=1300 kg/m3), and porosity (θsed = 0.509), we can develop the following expression with unit conversions in place:

(10)

Cw[µg/L] = { (mai[kg/ha]10-4[ha/m2]109[µg/kg]) } { (10-3[m3/L]) } / {0.10m + 0.01m(0.509 + 1300[kg/m3]Kd[L/kg]10-3[m3/L])}

Simplifying equation (10) forms the Tier I Rice Model expression (with units shown):

(11)

Cw[µg/L] = {mai[kg/ha]} / {1.05[10-3] [(L)(kg) / (ha)(µg)] + 1.3[10-4][kg2 / (ha)(µg) ]Kd[L/kg]}

The Tier I Rice Model v1.0 (equation (6)) is equation (11) written with implied units:

(6)

Cw = mai / (0.00105 + 0.00013Kd)

where:

Cw = water concentration [µg/L]
mai = mass applied per unit area [kg/ha]
Kd = water-sediment partitioning coefficient [L/kg]

Top of page

• ### Appendix II: Model Comparison

Table II. Comparison of Interim Rice Model Exposure Estimates with Those of the Tier I Rice Model v1.0.
CompoundKoc Interim Rice Model Output (ppb) Tier I Rice Model v1.0 Output (ppb) Ratio of Model Outputs
(-)
Bentazon 358769131.04
Molinate 1866527741.19
Propaconazole 6483665281.44
Trifloxystrobin 350098.51791.81
Lambda cyhalothrin 3000001.282.562.00

Top of page

• ### Appendix III: Sample Input and Output Tables

#### Sample Input Table:

Table III.1. Tier I Rice Model v1.0 input parameters for [chemical name]. Source data are in Table [X].
Input ParameterValue JustificationSource
Application Rate (kg a.i./ha) 0.06220.0555 lbs a.i./A x 1.121 (kg)(A) / (ha)(lbs) = 0.0622 kg a.i./ha Proposed label
Soil-to-Water Partition Coefficient (Kd) (L/kg) 4.00Mean Koc (400 L/kgoc) x 1 % organic carbon content = 4.00 L/kg[ MRID #(s)]

#### Sample Ecological Exposure Assessment Output Table:

Table III.2. Tier I surface water estimated exposure concentrations (EEC) of [chemical name] from use on rice.
SourceApp. Rate (lbs a.i./A)Peak & Chronic EEC (µg/L)

#### Sample Drinking Water Exposure Assessment Output Table:

Table III.3. Tier I drinking water exposure estimates for [chemical name] use on rice.1
SourcePeak Exposure (µg/L) Annual Mean Exposure (µg/L)
Surface Water39.6< 39.6
Ground Water[SCI-GROW value]< [SCI-GROW value]

1Surface water concentrations calculated by the Tier I Rice Model v1.0 and ground water concentrations calculated by SCI-GROW do not distinguish between peak and chronic concentrations.

Top of page