Elsevier

Journal of Hydrology

Volume 384, Issues 1–2, 15 April 2010, Pages 164-173
Journal of Hydrology

Influence of flow concentration on parameter importance and prediction uncertainty of pesticide trapping by vegetative filter strips

https://doi.org/10.1016/j.jhydrol.2010.01.020Get rights and content

Summary

Flow concentration is a key hydrologic factor limiting the effectiveness of vegetated filter strips (VFS) in removing pesticides from surface runoff. Numerical models, such as VFSMOD-W, offer a mechanistic approach for evaluating VFS effectiveness under various hydrological conditions including concentrated flow. This research hypothesizes that the presence of concentrated flow drastically alters the importance of various hydrological, sedimentological, and pesticide input factors and the prediction uncertainty of pesticide reduction. Using data from a VFS experimental field study investigating chlorpyrifos and atrazine transport, a two-step global sensitivity and uncertainty analysis framework was used with VFSMOD-W based on (1) a screening method (Morris) and (2) a variance-based method (extended Fourier Analysis Sensitivity Test, FAST). The vertical, saturated hydraulic conductivity was consistently the most important input factor for predicting infiltration, explaining 49% of total output variance for uniform sheet flow, but only 8% for concentrated flow. Sedimentation was governed by both hydrologic (vertical, saturated hydraulic conductivity and initial and saturated water content) and sediment characteristics (average particle diameter). The vertical, saturated hydraulic conductivity was the most important input factor for atrazine or chlorpyrifos trapping under uniform sheet flow (explained more than 46% of the total output variance) and concentrated flow (although only explained 8% of the total variance in this case). The 95% confidence intervals for atrazine and chlorpyrifos reduction ranged between 43% and 78% for uniform sheet flow and decreased to between 1% and 16% under concentrated flow. Concentrated flow increased interactions among the system components, enhancing the relative importance of processes that were latent under shallow flow conditions. This complex behavior warrants the need for process-based modeling to be able to predict the performance of VFS under a wide range of specific hydrological conditions.

Introduction

Vegetation filter strips (VFS) reduce pesticide movement to water bodies by reducing runoff volumes through infiltration in the filter strip’s soil profile, through contact between dissolved phase pesticide with soil and vegetation in the filter strip, and/or by reducing flow velocities to the point where eroded sediment particles, with sorbed pesticide, can settle out of the water. However, predicting VFS effectiveness has historically been a difficult, if not impossible, task due to the variability observed in different field conditions. For example, in ten specific VFS studies, Sabbagh et al. (2009) document reported pesticide reduction of 11–100% for VFS of widths ranging from 0.5 to 20 m. Past research has attempted to develop general statistical relationships between sediment and/or chemical trapping as functions of buffer physical characteristics such as width and slope, but cannot predict strong relationships between the variables due to a lack of consideration for hydrological processes (Fox and Sabbagh, 2009).

One of the key factors influencing VFS effectiveness is concentrated as opposed to shallow overland or uniform sheet flow (Dosskey et al., 2002, Krutz et al., 2005, Blanco-Canqui et al., 2004, Blanco-Canqui et al., 2006, Fox and Sabbagh, 2009, Poletika et al., 2009, Sabbagh et al., 2009). Departure from sheet flow reduces VFS effectiveness by decreasing infiltration and sedimentation of suspended particles as the grass stems become inundated and flow velocity is undiminished (Dillaha et al., 1989). Blanco-Canqui et al. (2006) demonstrated that narrow filter strips could filter sediment and remove nutrients for interrill flow but their performance for concentrated flow was diminished even on gentle slopes of less than 5%.

Recent research has proposed that performance of VFS for pesticide trapping depends on hydrologic conditions (precipitation, infiltration and runoff) driven by the filter design (length, slope, and densities of vegetation cover) and characteristics of the incoming pollutants (sediment and pesticides) (Dosskey et al., 2002, Blanco-Canqui et al., 2004, Fox and Sabbagh, 2009, Poletika et al., 2009, Sabbagh et al., 2009). Sabbagh et al. (2009) developed and evaluated an empirical model for pesticide trapping with a foundation of hydrological, sedimentological, and chemical specific parameters:ΔP=a+b(ΔQ)+c(ΔE)+dln(Fph+1)+e(%C)where ΔP is the pesticide removal efficiency (%), a, b, c, d, and e are regression coefficients, ΔQ is the infiltration (%) defined as the difference between flow entering the VFS (i.e., inflow runon plus precipitation) minus the runoff from the VFS, ΔE is the sediment reduction (%), %C is the clay content of the sediment entering the VFS, and Fph is a phase distribution factor, defined as the ratio of pesticide mass in dissolved form to pesticide mass sorbed to sediment:Fph=QiKdEiwhere Qi and Ei are the volume of water (L) and mass of sediment (kg) entering the VFS, and Kd is the distribution coefficient defined as the product of KOC, the organic carbon sorption coefficient, and PCTOC, the percent organic carbon in the soil, divided by 100. They also proposed a procedure linking the VFS numerical model VFSMOD-W (Muñoz-Carpena et al., 1993a, Muñoz-Carpena et al., 1993b, Muñoz-Carpena et al., 1999, Muñoz-Carpena and Parsons, 2004, Muñoz-Carpena and Parsons, 2008) with the proposed empirical trapping efficiency equation that significantly improved predictions of pesticide trapping over conventional equations based solely on physical characteristics of the VFS.

VFSMOD-W, is a field-scale, mechanistic, storm-based numerical model developed to route the incoming hydrograph and sedigraph from an adjacent field through a VFS and to calculate the resulting outflow (based on the kinetic wave approximation of the Saint–Vennant’s equations for overland flow), infiltration (based on the Green–Ampt equation for unsteady rainfall), and sediment trapping efficiency based on sediment transport equations (Muñoz-Carpena et al., 1993a, Muñoz-Carpena et al., 1993b, Muñoz-Carpena et al., 1999, Muñoz-Carpena and Parsons, 2004, Muñoz-Carpena and Parsons, 2008). VFSMOD-W originated from GRASSF (Barfield et al., 1979). Muñoz-Carpena et al. (1999) improved upon GRASSF by including improved routines for flow through the filter, time-dependent infiltration, and spatial variability in surface conditions. Researchers have successfully tested the model in a variety of field experiments with good agreement between model predictions and measured values of infiltration, outflow, and trapping efficiency for particles (Muñoz-Carpena et al., 1999, Abu-Zreig, 2001, Abu-Zreig et al., 2001, Dosskey et al., 2002, Fox et al., 2005, Han et al., 2005), and phosphorus (particulate and dissolved) (Kuo, 2007, Kuo and Muñoz-Carpena, 2009). VFSMOD-W is currently used in conjunction with other watershed tools and models to develop criteria and response curves to assess buffer performance and placement at the watershed level (Yang and Weersink, 2004, Dosskey et al., 2005, Dosskey et al., 2006, Dosskey et al., 2008, Tomer et al., 2009, White and Arnold, 2009).

Poletika et al. (2009) reported a combined field/modeling study investigating the effect of runoff volume and flow concentration on removal of chlorpyrifos [O,O-diethyl O-(3,5,6-tricholoro-2-pyridyl) phosphorothioate] and atrazine [2-chloro-4-(ethylamino)-6-(isopropylamino)-s-triazine] by filter strips. The field experiments demonstrated that increased flow volume had a minor impact on removal efficiency while flow concentration reduced removal performance regardless of the drainage area ratio. Poletika et al. (2009) concluded that the lack of clear trends between flow volume and flow uniformity verified the necessity of hydrologic modeling within the VFS to capture the hydrologic conditions and response to different events, and showed that the uncalibrated VFSMOD-W was capable of predicting ΔQ (R2 = 0.79), ΔE (R2 = 0.85), and ΔP (R2 = 0.84) for uniform sheet flow and concentrated flow.

Analyses of sensitivity (Muñoz-Carpena et al., 1999, Muñoz-Carpena et al., 2007, Abu-Zreig, 2001) and uncertainty (Parsons and Muñoz-Carpena, 2001, Shirmohammadi et al., 2006, Muñoz-Carpena et al., 2007) of the VFSMOD-W model have been previously reported for numerous applications. However, the influence of flow concentration relative to input factor importance and prediction uncertainty of pesticide trapping has not been analyzed. When conducting model sensitivity analysis, often, local, “one-parameter-at-a-time” sensitivity analysis is performed by varying each input a small amount around a base value and considering all other inputs fixed. However, this approach is only valid for additive and linear output models. Instead, an alternative “global” sensitivity approach, where the entire parametric space of the model is explored simultaneously for all input factors, is needed. Global methods provide not only a ranking of input factor importance and the direct (first order) effect of the individual factors over the output, but also about their interactions (higher order) (Saltelli et al., 2004).

The objective of this study was to evaluate input factor importance and uncertainty in predicted ΔQ, ΔE, and ΔP under uniform sheet flow versus concentrated flow conditions. The research utilized modern global sensitivity and uncertainty analyses for modeling ΔP using VFSMOD-W. The analysis tools were applied to two different treatments in the field study by Poletika et al. (2009) investigating the role of uniform sheet flow versus concentrated flow on atrazine and chlorpyrifos reduction by a VFS. This research is critical to advance the role of VFS as a central component of environmental management plans related to pesticide application.

Section snippets

VFS field study

The Poletika et al. (2009) field study was conducted in western Sioux County, Iowa, with 4.6-m long smooth brome and bluegrass strips. The soil was a moderately erodible Galva silty clay loam (fine-silty, mixed, mesic, Typic Hapludoll). Slopes were uniform within the study area and ranged from 5.0% to 5.5%. Artificial runoff was metered into the VFS plots for 90 min following a simulated rainfall of 63 mm applied over 2 h. The artificial runoff contained sediment and was dosed with chlorpyrifos

Results and discussion

As proposed by Morris (1991), only input factors separated from the origin of the μ*σ plane were considered important (Fig. 1, Fig. 2). In general, the number of important input factors for predicting ΔQ (i.e., approximately four for both uniform sheet flow and concentrated flow in Fig. 1a and b) decreased considerably from the full set of 18 model inputs. The Morris results appropriately indicated that ΔQ was not dependent on any sediment or pesticide inputs. The VKS was the dominant input

Summary and conclusions

Concentrated flow provided less time for infiltration and sedimentation processes within a VFS and correspondingly less pesticide removal. Input factor importance for predicting VFS performance depended considerably on the hydrological conditions in terms of flow concentration. Attempts to rely explicitly on single input factors to predict VFS performance (i.e., hydraulic conductivity, buffer width, or slope) will fail unless one considers the flow conditions experienced by the VFS. In other

Acknowledgements

The authors acknowledge the University of Florida, High-Performance Computing Center for providing computational resources and support that have contributed to the research results reported within this paper. http://www.hpc.ufl.edu. The authors also acknowledge Zuzanna Zajac, University of Florida, for reviewing the Morris results, and Amanda K. Fox, Stillwater, OK, for reviewing an earlier version of this manuscript.

References (48)

  • H. Blanco-Canqui et al.

    Performance of grass barriers and filter strips under interrill and concentrated flow

    Journal of Environmental Quality

    (2006)
  • R.F. Carsel et al.

    Developing joint probability distributions of soil water retention characteristics

    Water Resources Research

    (1988)
  • R.I. Cukier et al.

    Study of the sensitivity of coupled reaction systems to uncertainties in rate coefficients: I. Theory

    Journal of Chemical Physics

    (1973)
  • T.A. Dillaha et al.

    Vegetative filter strips for agricultural nonpoint source pollution control

    Transactions of the ASAE

    (1989)
  • M.G. Dosskey et al.

    Assessment of concentrated flow through riparian buffers

    Journal of Soil and Water Conservation

    (2002)
  • M.G. Dosskey et al.

    Establishing conservation buffers using precision information

    Journal of Soil and Water Conservation

    (2005)
  • M.G. Dosskey et al.

    An approach for using soil surveys to guide the placement of water quality buffers

    Journal of Soil and Water Conservation

    (2006)
  • M.G. Dosskey et al.

    A design aid for determining width of filter strips

    Journal of Soil and Water Conservation

    (2008)
  • A.L. Fox et al.

    Modeling water and sediment trapping by vegetated filters using VFSMOD: comparing methods for estimating infiltration parameters. ASAE Paper No. 052118

    (2005)
  • G.A. Fox et al.

    Comment on “major factors influencing the efficacy of vegetated buffers on sediment trapping: a review and analysis”

    Journal of Environmental Quality

    (2009)
  • N. Gupta et al.

    Analysis of spatial variability in hydraulic conductivity at the field scale

    Canadian Biosystems Engineering

    (2006)
  • C.T. Haan et al.

    Design Hydrology and Sedimentology

    (1994)
  • J. Han et al.

    Suspended sediment removal by vegetative filter strip treating highway runoff

    Journal of Environmental Science and Health

    (2005)
  • M. Koda et al.

    Automatic sensitivity analysis of kinetic mechanisms

    International Journal of Chemical Kinetics

    (1979)
  • Cited by (0)

    View full text