JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Haith, D. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Haith, D. A.
Agricola
Right arrow Articles by Haith, D. A.
Related Collections
Right arrow Water Quality
Right arrow Turfgrass Pesticides
Right arrow Other Models
Right arrow Water Pollution
Journal of Environmental Quality 30:1033-1039 (2001)
© 2001 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORT
Surface Water Quality

TurfPQ, A Pesticide Runoff Model for Turf

Douglas A. Haith

Agricultural and Biological Engineering, Riley–Robb Hall, Cornell Univ., Ithaca, NY 14853

Corresponding author (dah13{at}cornell.edu)

Received for publication June 26, 2000.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Environmental assessments of golf courses and other turf systems must often rely on mathematical modeling. However, in the case of pesticide runoff, successful modeling applications are rare. Available models were developed for agricultural applications and have seen very limited testing for turf. TurfPQ is a pesticide runoff model developed exclusively for turf. The model is based on a curve number calculation for runoff volume and linear partitioning of pesticide into adsorbed and dissolved components during a precipitation or irrigation event. Calibration is optional, so the model can be applied, using default parameter values, to situations where runoff and chemical loss data are unavailable. TurfPQ was tested with default parameter values for 52 pesticide runoff events involving six pesticides measured in plot studies in four states. The model typically produced conservative overpredictions of pesticide runoff, particularly with strongly adsorbed pesticides. Mean predicted pesticide runoff was 3.2% of application, compared with an observed mean of 2.1%. TurfPQ captured the dynamics of the pesticide runoff events well with R2 = 0.76. Sensitivity analyses indicated that prediction errors could be reduced by better estimates of adsorption parameters and runoff curve numbers. However, even with default parameters, TurfPQ predictions are at least as accurate as those produced by more complex models.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
PESTICIDE runoff from turf has uncertain effects on the quality of surface waters such as streams, ponds, and lakes. Given the high evapotranspiration and infiltration rates of turfgrass systems, runoff water may seldom be sufficient to wash significant amounts of chemicals off-site. Nevertheless, a recent review of water quality data indicated that 31 pesticides have been detected in surface waters on or near golf courses (Cohen et al., 1999). Pesticide levels in runoff events measured in field experiments have ranged from negligible to 25% of applications (Cole et al., 1997; Evans et al., 1998; Hong and Smith, 1997; Ma et al., 1999; Smith and Bridges, 1996; Watschke et al., 2000).

Pesticide runoff from turf depends on chemical properties, soil and cover conditions, timing of applications, and weather patterns (Balogh and Anderson, 1992). Field evaluations of more than a few combinations of these factors are unlikely, and environmental assessments of golf courses and other turf systems must often rely on mathematical modeling. However, in the case of pesticide runoff, successful modeling applications to turf have been very limited. Rosenthal and Hipp (1993) used the EPIC model to simulate pesticide runoff from four different turf management systems. Model results were not validated by comparisons with field data. Wauchope et al. (1990) applied the GLEAMS model to estimate pesticide runoff from Texas turf plots. The model was calibrated by adjusting runoff curve numbers to produce the measured runoff volumes. Predicted total pesticide runoff was 56% of the measured value for cyanazine (2-[[4-chloro-6-(ethylamino)-1,3,5-triazin-2-yl]amino]-2-methylpropionitrile), and 64 and 109% of measurements for two different formulations of sulfometuron-methyl (methyl 2[[[[(4,6-dimethyl-2-pyrimidinyl)amino]carbonyl]amino]sulfonyl]benzoate). Ma et al. (1999) simulated 3 yr of 2,4-D (2,4-dichlorophenoxyacetic acid) runoff from Georgia fairway plots using the OPUS model. The runoff curve number was calibrated using the first year of data, and pesticide runoff predictions were compared with measurements for the entire 3-yr period. Depending on the form of the sorption model used in OPUS, 85 to 95% of the predicted 2,4-D concentrations in runoff events were within a factor or two of measured values. Durborow et al. (2000) calibrated GLEAMS for the same sites used in Ma et al. (1999). Although the authors indicated that "pesticide predictions also compared well with observed concentrations," results were only presented for two of seven pesticides, and the predicted and observed concentrations often differed by more than a factor of two and, in several cases, by an order of magnitude.

Based on these studies, it appears that neither EPIC nor GLEAMS has been tested for turf conditions with independent observations (i.e., data not used in calibration). The OPUS model has seen such testing, but for only one pesticide and one site, with associated errors of roughly 100%.

Even if these models could be shown to be reasonable predictors of pesticide runoff from turf, their extensive data and calibration requirements would limit usefulness in water quality studies and environmental assessments where runoff and chemical loss data are lacking. Moreover, the models were originally developed for agricultural applications, and selection of parameter values for turf systems is problematic. Given these difficulties, it is reasonable to investigate whether simpler modeling approaches based on the physical characteristics of turf might provide a more accurate and useful model.

The model described in this paper, TurfPQ, estimates pesticide in runoff events from turf. The model is based on a curve number calculation for runoff volume and linear partitioning of pesticide into adsorbed and dissolved components during an event. Calibration is optional, so the model can be applied, using default parameter values, to situations where runoff and chemical loss data are unavailable. Succeeding portions of this paper describe the model, the five plot studies in four states used for model testing, and the comparisons of model estimates with field observations.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Runoff Model
Runoff is determined by a straightforward application of the U.S. Soil Conservation Service curve number equation (Ralston, 1985; Soil Conservation Service, 1986) as described by Haith and Andre (2000):

[1]
where Qt, Rt, and St = runoff, rain or irrigation, and water detention, respectively, on day t (mm), and St is related to a curve number, CNt, by:

[2]

Curve numbers are selected as linear functions of 5-d accumulated antecedent rainfall, At (cm):

[3]


[4]


[5]
where CN1, CN2, and CN3 are curve numbers for dry, average, and wet antecedent moisture, and AMC1 and AMC2 are breakpoints or limits that determine which of the three equations ([3], [4], or [5]) are used. Suggested moisture limits are AMC1 = 13 and 36 mm for dormant and growing seasons, respectively, and AMC2 = 28 and 53 mm for dormant and growing seasons, respectively (Ralston, 1985). However, Haith and Andre (2000) showed that growing season limits of AMC1 = 22 cm and AMC2 = 53 cm provide better estimates of turf runoff.

The single required model parameter is CN2, the runoff curve number for average moisture conditions. The other two curve numbers are given in tables in the associated manuals, or from the fitted equations given in Hawkins (1978). Haith and Andre (2000) proposed the set of turfgrass curve numbers given in Table 1. These curve numbers are based on values given for pasture, range, and meadow in Ralston (1985), and assume that thatch has an effect on runoff comparable with the effects of plant residues with conservation tillage. The soil hydrologic groups in Table 1 are the standard Soil Conservation Service categories ranging from very high infiltration rates (A) to very low rates (D).


View this table:
[in this window]
[in a new window]
  		Table 1. Curve numbers for turfgrass at average antecedent moisture (CN2).

 
The runoff model was tested, as described in Haith and Andre (2000), by comparing model estimates with measured runoff data from turf field plots in six different states, including three different soil hydrologic groups, four turfgrasses, and a range of antecedent moisture and turf conditions. Mean modeled runoff from the 69 events was within 3% of the observed mean, and the predicted runoff event values explained 77% of the variation in measured runoff. Also, the model mean for the largest observed events was within 6% of the observed mean.

Chemical Model
The chemical model is based on a mass balance of the pesticide in the turf foliage and thatch. Once the pesticide is leached into the soil it is assumed to be unavailable for runoff. Sediment losses from turf are relatively small, and it is thus assumed that all pesticide runoff is in the dissolved form. The general mass balance is:

[6]
in which Pt = turf pesticide at the beginning of day t, {Delta}Pt = pesticide application on day t, PQt = pesticide in runoff on day t, and PLt = pesticide leached into the soil by wash off and infiltration on day t (all in g/ha). The model presumes first order, or exponential decay of the pesticide with rate {alpha} (d-1), given by {alpha} = -0.693/ln({tau}1/2), where {tau}1/2 is the decay half-life (d).

During a precipitation (rain or irrigation) event, the available pesticide Pt* = Pt + {Delta}Pt is partitioned between dissolved and adsorbed forms. This partitioning is conceptualized as a two-stage process, corresponding with the division of precipitation first into infiltration (Rt - Qt) and subsequently into runoff (Qt). During the first, or infiltration stage of the event, we have:

[7]
where A1t = pesticide adsorbed to vegetation (foliage and thatch) and D1t = pesticide dissolved in infiltration during Stage 1 (g/ha).

Assuming a linear equilibrium partitioning,

[8]
with a = adsorbed concentration (µg/g), d = dissolved concentration (µg/cm3), and K = partition coefficient (cm3/g), Eq. [7] and [8] can be combined as:

[9]
where M = vegetation dry mass (kg/ha). Solving Eq. [9] for d, we can obtain the pesticide leached from the turf by infiltration:

[10]

Partition coefficients are generally determined from Koc, the organic carbon partition coefficient, and Eq. [10] is more conveniently written as:

[11]
in which OC = organic carbon in vegetation dry matter (kg/ha).

The remaining pesticide Pt*- D1t is assumed to reach an equilibrium with runoff water,

[12]
where A2t = pesticide adsorbed to turf vegetation and D2t = pesticide dissolved in runoff (g/ha). Through a process similar to Eq. [8] to [11], we obtain the pesticide in runoff,

[13]

The complete TurfPQ model consists of Eq. [1] to [6], Eq. [11], and Eq. [13]. It is a very simplified description of pesticide behavior in turf. The division of event water first into infiltration (Rt - Qt) and subsequently runoff (Qt) approximates a more complex process in which runoff is minimal early in the event and gradually becomes more significant in the later stages. The model neglects adsorption kinetics, volatilization, pesticide incorporated into plant tissue, and dissolved pesticide in water remaining on vegetation following the precipitation event. As a result, TurfPQ should generally overestimate pesticide runoff losses.

Default Parameter Values
Model input parameters are runoff curve number, pesticide decay half-life ({tau}1/2), organic carbon partition coefficient (Koc), and the organic carbon in turf vegetation (OC). Curve numbers are given in Table 1, but the remaining parameters would ideally be measured at the site of interest. More realistically, default values based on secondary sources will be necessary in most applications.

Default values of {tau}1/2 and Koc for many pesticides are available from the USDA Pesticide Properties Database (Agricultural Research Service, 2000) or compendiums such as Tomlin (1994) and Ahrens (1994). These parameter values are typically for soils, and {tau}1/2 is generally the aerobic biodegradation half-life. Similar comprehensive data sources do not exist for turf vegetation. Although there is some evidence that soil and turf values are comparable (Dell et al., 1994; Cisar and Snyder, 1996) other studies have found substantial differences (Carroll et al., 2000; Horst et al., 1996; Lickfeldt and Branham, 1995). Nevertheless, soil values will generally be the only practical alternative. Soil Koc values and biodegradation half-lives for the six pesticides included in this study are listed in Table 2.


View this table:
[in this window]
[in a new window]
  		Table 2. Adsorption and decay parameters for pesticides.

 
Lickfeldt and Branham (1995) measured organic carbon contents of 38% in dry matter of both thatch and grass foliage from an 8-yr-old Kentucky bluegrass (Poa pratensis L.) turf. A bulk density of 0.38 g/cm3 was measured for the 40-mm depth of thatch, and this results in an organic carbon content of 0.144 g/cm3. Comparable values can be inferred from data collected by Horst et al. (1996) for two Kentucky bluegrass turfs. The first turf, which was 6 to 12 mm thick, had a bulk density of 0.75 g/cm3 and 18% organic matter. Using the ratio of 0.46 organic carbon to organic matter measured by Lickfeldt and Branham (1995), the thatch organic carbon = 0.062 g/cm3. For the second turf, which was 12 to 24 mm thick, with bulk density = 0.73 g/cm3 and 39% organic matter, we obtain 0.131 g/cm3 of organic carbon. Although these values seem to reflect a depth effect, three samples are insufficient to measure it, and an average value of 0.112 g/cm3 is a plausible default value. This is equivalent to 1120 kg/ha per mm of thatch depth.

Organic carbon in turf foliage can be estimated from dry matter measurements using the 38% organic carbon content measured by Lickfeldt and Branham (1995). Wood and Burke (1961) measured foliage dry matter of five bluegrasses at three different clipping heights. Madison (1962) reported similar measurements for two bentgrasses. The data are summarized in Table 3, and the mean values, which exhibit a typical inverse relationship of density with height (Turgeon, 1991), can be used to determine default OC values.


View this table:
[in this window]
[in a new window]
  		Table 3. Dry matter and organic carbon contents of several turfgrasses.

 
Field Studies
Pesticide runoff data were available from five plot studies in four states. The following provides brief descriptions of the studies.

Two field experiments measured pesticide runoff from bermudagrass [Cynodon dactylon (L.) Pers.] fairways on a Cecil sandy clay loam (fine, kaolinitic, thermic Typic Kanhapludult) in Georgia. Smith and Bridges (1996) irrigated plots at 1, 2, 4, and 8 d following applications of 2,4-D, dicamba (3,6-dichloro-2-methoxybenzoic acid), and mecoprop [(RS)-2-(4-chloro-2-methylphenoxy)propionic acid]. Comparable experiments were conducted by Hong and Smith (1997) using granule and emulsifiable concentrate formulations of dithiopyr [S,S'-dimethyl 2-(difluoromethyl)-4-(2-methylpropyl)-6-(trifluoromethyl)-3,5-pyridinedicarbothioate]. Grass heights were 15 and 40 mm for the Smith and Bridges (1996) and Hong and Smith (1997) studies, respectively, and thatch was just beginning to form (A.E. Smith, personal communication, 2000). A 1-mm thatch thickness was assumed. The Cecil soil is classified as Hydrologic Group B in Ralston (1985). However, the plots are poorly drained due to a flow restricting layer, and the previous runoff model calibration by Ma et al. (1999) determined that a C classification is more appropriate. This results in CN2 = 74 for short grass, complete cover.

Evans et al. (1998) measured diazinon (O,O-diethyl O-[6-methyl-2-(1-methylethyl)-4-pyrimidinyl] phosphorothioate) runoff in Kentucky from tall fescue (Festuca arundinacea Schreb.) on a Maury silt loam (fine, mixed, semiactive, mesic Typic Paleudalf) (B soil). Plots were preirrigated at three different rates, and identical water applications of 96 mm were applied to all plots. Grass height was estimated as 76 to 102 mm (3–4 in) and no thatch was apparent (D.R. Edwards, personal communication, 2000). A grass height of 89 mm was assumed and the long grass curve number (CN2 = 58) was used.

Chlorpyrifos [O,O-diethyl O-(3,5,6-trichloro-2-pyridinyl) phosphorothioate], 2,4-D, dicamba, and mecoprop runoff from two Oklahoma irrigation experiments with 13-mm bermudagrass on a Kirkland silt loam (fine, mixed, superactive, thermic Udertic Paleustoll) (Soil D) were reported by Cole et al. (1997). Conditions for the July experiment were very dry, with no precipitation in the previous 12 d. Conversely, there were 165 mm of precipitation in the 7 d prior to the second experiment in August. The short grass curve number (CN2 = 80) was used for model runs. No thatch development was observed (J.H. Baird, personal communication, 2000).

The final runoff data set is from the Pennsylvania plot studies of Linde et al. (1995), with pesticide data given in Watschke et al. (2000). Runoff data for mecoprop and triadimefon [1-(4-chlorophenoxy)-3,3-dimethyl-1-(1H-1,2,4-triazol-1-yl)-2-butanone] were available from Watschke et al. (2000), but only the mecoprop data were used because information was insufficient to determine triadimefon application rates. Pesticide runoff was reported for two irrigation events (10 June 1992 and 24 June 1992) for a dense, thatch-producing turf (creeping bentgrass [Agrostis stolonifera L.]) and a thatch-free ryegrass (Lolium perenne L.) turf. Soil type was a Hagerstown clay (fine, mixed, semiactive, mesic Typic Hapludalf) (Soil C), and both turfs were maintained at 19 mm. Bentgrass thatch depth was measured several times, and a 1 July measurement of 1 mm was assumed most representative of the pesticide runoff events. However, this was a "compressed" sample, and it was assumed that this would be approximately equivalent to a 2-mm uncompressed thickness. The curve number for thatched short grass (CN2 = 67) was used for the bentgrass and the short grass curve number (CN2 = 74) was used for ryegrass.

Model Testing
The 52 pesticide runoff events available for testing are listed in Table 4. Data were summarized as percent of pesticide application. Most of these values were means from several plot replications. Water applications and antecedent moisture data are given in Haith and Andre (2000). The antecedent moisture limits AMC1 = 22 cm and AMC2 = 53 cm suggested by Haith and Andre were used for all events. The mean values given in Table 3 were used for foliage organic carbon levels, and the comparable value for thatch was 1120 kg/ha per mm of thatch depth, as discussed previously.


View this table:
[in this window]
[in a new window]
  		Table 4. Comparison of observed and modeled pesticide runoff events using measured and modeled runoff volumes.

 
Each of the events was simulated by the TurfPQ model for two different cases. In Case A, measured runoff values were used for the pesticide calculations, thus providing a test for the TurfPQ chemical component (Eq. [6], [11], [13]), which avoids the confounding effects of runoff errors. Case B, which uses the computed runoff values from Eq. [1] to Eq. [5], tests the complete model as it would normally be used to predict both runoff and pesticide losses.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Event Comparisons
Model results for the two cases are compared with observations for each event in Table 4. Means and coefficients of determination (R2) are also provided, the latter indicating the fraction of observed variation in pesticide runoff explained by the model. Case A results suggests that the chemical model tends to substantially overestimate pesticide runoff, with the predicted mean almost twice as large as observed. However, the model captures the differences among pesticide events well, explaining 86% of the variation in pesticide runoff. When the complete model is tested (Case B), overestimation of means is less severe, with runoff errors compensating somewhat for the chemical errors. Differences among events are still captured well, although the explained variation drops to 76%. Modeled and observed runoff volumes are compared in Table 5. The overall means are quite close (17.1 vs. 19.3 mm), but there are substantial discrepancies, particularly at the Kentucky site, where runoff was underestimated by two-thirds.


View this table:
[in this window]
[in a new window]
  		Table 5. Comparison of observed and modeled runoff means.

 
Estimated and observed pesticide runoff (y' and y, respectively) for each event are also compared in Fig. 1 for Case B. Although events generally lie above the line y' = y, most are relatively close to the line. The significant exceptions are two of the largest events, which are overestimated by more that 10% of pesticide application.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 1. Comparison of model pesticide runoff estimates with observed values—Case B.

 
The overall ability of the TurfPQ model to describe pesticide runoff events was also evaluated using the efficiency measure proposed by Nash and Sutcliffe (1970), and illustrated by Martin et al. (1993),

[14]
where my = mean observed value. The coefficient compares the model's predictive capability with an alternative model that simply uses the observed mean as a predictor. The model is an efficient predictor when E approaches one, and is no better than the mean when E approaches zero. For Case B, which evaluates the complete model (hydrology and chemistry), E = 0.273, suggesting that the model is relatively inefficient. However, this is mainly due to the two large overpredictions shown in Fig. 1. When these two event are eliminated, E increases to 0.720.

Pesticide Comparisons
Table 6 summarizes the testing results by pesticide. With the exception of diazinon for Case B, model values are higher than observations for all pesticides. Model results are most accurate for dicamba and mecoprop, which are the least strongly adsorbed chemicals. Conversely, for chlorpyrifos and dithiopyr, the most strongly adsorbed pesticides, model results exceed observations by factors of more than five and three, respectively. Comparable comparisons are made in Table 7 for the largest runoff event for each pesticide. Such large events are likely to produce the most severe water quality effects. Model performance mirrors that of the means with again the greatest accuracy seen with the weakly adsorbed pesticides.


View this table:
[in this window]
[in a new window]
  		Table 6. Comparison of observed and modeled pesticide runoff for six pesticides.

 

View this table:
[in this window]
[in a new window]
  		Table 7. Comparison of observed and modeled pesticide runoff for maximum observed event for six pesticides.

 
Sources of Error
The most troublesome of the modeling results (the underestimation of diazinon runoff and the overestimation of chlorpyrifos and dithiopyr runoff) can be similarly explained by parameter errors. The diazinon errors seem to be associated with the underestimation of runoff volumes at the Kentucky site, because the Case A results, which are based on measured runoff, are much more accurate than those for Case B. Model runoff values are very sensitive to curve number. For example, a 10% increase in the Kentucky curve number from CN2 = 58 to CN2 = 64 increases mean runoff from 2.9 to 6.7 mm and produces a mean diazinon runoff of 0.69%, which almost exactly matches the observed 0.68%.

Chlorpyrifos and dithiopyr overestimates are probably associated with the distribution of the pesticides into dissolved and adsorbed forms, which in turn depends on the Koc coefficient and the organic carbon content of the turf vegetation. Soil values for Koc may not accurately reflect adsorption to plant material. For example, if the dithiopyr Koc is increased from its soil value of 1638 to 5000 cm3/g, mean pesticide runoff decreases to 0.43%, which more closely reflects the observed 0.32%. Alternatively, indications of thatch at the Georgia site were approximate at best, and if the thatch thickness is increased from 1 mm to 3 mm, total OC increases from 3720 to 5960 kg/ha, and with the soil Koc of 1638 cm3/g, the model produces a mean dithiopyr runoff of 0.78%, substantially less than the 1.18% given in Table 6.

Similar effects on the Oklahoma chlorpyrifos estimates are seen with adjustment of Koc or thatch thickness. Doubling the Koc value to 20000 cm3/g reduces mean pesticide runoff from 2.85 to 1.46%. Increasing thatch thickness from 0 to 2 mm results in mean chlorpyrifos runoff of 1.14%.

Problems with adsorption parameter estimates also play a role in the model's better estimation of pesticide runoff when model runoff volume estimates are used (Case B) rather than the actual measured volumes (Case A). However, these differences, which are largely associated with the Georgia 2,4-D, dicamba, and mecoprop plots, are more apparent than real. For example, if the Koc value for 2,4-D is reduced from 48 to 20 cm3/g, the mean modeled pesticide runoff is reduced from 5.0 to 2.9% for Case A, which is much closer to the measured 2.2%.

Based on this sensitivity exercise, neither the runoff nor chemical model appear to have serious conceptual flaws. It is apparent that when water and pesticide runoff data are available, it would be relatively easy to calibrate the TurfPQ model by adjusting curve number and adsorption parameters. It is also apparent that more research is needed to provide better estimates for these parameters.

The errors in runoff losses for strongly adsorbed chemicals may also be a result of the simple equilibrium model used for partitioning pesticide into dissolved and adsorbed components. It is certainly possible that the assumed equilibrium is not achieved during a runoff event, and release of adsorbed pesticide is a slower process better described by a kinetic model.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The TurfPQ model is a relatively simple means of estimating pesticide runoff from turf. It consists of only eight equations and requires but four input parameters (curve number, Koc, decay rate or half-life, and OC, the organic carbon content of the turf vegetation). The model neglects processes such as volatilization and adsorption kinetics and thus will often overestimate pesticide runoff. Default values are available for model parameters, and hence TurfPQ may be used in situations where runoff and chemical loss data are unavailable for calibration.

Testing with the default parameter values for 52 pesticide runoff events in four states confirmed that TurfPQ model results consistently exceeded observed pesticide runoff, typically by a factor of two. Model overestimates were most severe with strongly adsorbed pesticides. TurfPQ captures the dynamics of the pesticide runoff events well with R2 = 0.76.

To the best of our knowledge, TurfPQ is the first pesticide runoff model developed exclusively for turf. Other models such as EPIC, GLEAMS, and OPUS, which have been applied to turf, were originally developed for agricultural crops, and are much more data intensive. Although comparisons of accuracy are difficult because TurfPQ is the only model to have been tested with independent data sets for multiple chemicals and sites, it appears that the accuracy of the uncalibrated TurfPQ meets or exceeds that of the more complex models.


   ACKNOWLEDGMENTS
 
Research described in this paper was supported, in part, by Green Section Research, U.S. Golf Association.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2001 30: 677-682. [Full Text]  



This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
D. A. Haith and F. S. Rossi
Risk Assessment of Pesticide Runoff from Turf
J. Environ. Qual., March 1, 2003; 32(2): 447 - 455.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
D. A. Haith, P.-C. Lee, J. M. Clark, G. R. Roy, M. J. Imboden, and R. R. Walden
Modeling Pesticide Volatilization from Turf
J. Environ. Qual., May 1, 2002; 31(3): 724 - 729.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Haith, D. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Haith, D. A.
Agricola
Right arrow Articles by Haith, D. A.
Related Collections
Right arrow Water Quality
Right arrow Turfgrass Pesticides
Right arrow Other Models
Right arrow Water Pollution

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome