Elsevier

Advances in Water Resources

Volume 52, February 2013, Pages 232-242
Advances in Water Resources

Catchment-scale herbicides transport: Theory and application

https://doi.org/10.1016/j.advwatres.2012.11.007Get rights and content

Abstract

This paper proposes and tests a model which couples the description of hydrologic flow and transport of herbicides at catchment scales. The model accounts for streamflow components’ age to characterize short and long term fluctuations of herbicide flux concentrations in stream waters, whose peaks exceeding a toxic threshold are key to exposure risk of aquatic ecosystems. The model is based on a travel time formulation of transport embedding a source zone that describes near surface herbicide dynamics. To this aim we generalize a recently proposed scheme for the analytical derivation of travel time distributions to the case of solutes that can be partially taken up by transpiration and undergo chemical degradation. The framework developed is evaluated by comparing modeled hydrographs and atrazine chemographs with those measured in the Aabach agricultural catchment (Switzerland). The model proves reliable in defining complex transport features shaped by the interplay of long term processes, related to the persistence of solute components in soils, and short term dynamics related to storm inter-arrivals. The effects of stochasticity in rainfall patterns and application dates on concentrations and loads in runoff are assessed via Monte Carlo simulations, highlighting the crucial role played by the first rainfall event occurring after herbicide application. A probabilistic framework for critical determinants of exposure risk to aquatic communities is defined. Modeling of herbicides circulation at catchment scale thus emerges as essential tools for ecological risk assessment.

Highlights

► Travel time formulation of transport is here extended to the description of the dynamics of solutes that are partially taken up by transpiration. ► The framework is applied to the study of atrazine circulation at catchment scale. ► Differences of residence times for water compared to those of solutes are addressed. ► Tools for ecological risk assessment related to herbicides are evaluated via a probabilistic analysis using Monte Carlo simulations.

Introduction

Catchment hydrologic response to rainfall forcings – here including fluxes of both water and solutes – involves both event (new) and non-event water stored in the catchment, possibly for rather long times, mixed in different proportions depending on the hydrologic pathways involved in runoff formation. Indeed, the mixing of water of different ages has been argued to dominate the variability of water quality in the runoff [1], [2], [3]. This stems from the fact that the chemical composition of streamflows is driven by the water residence time in the catchment which quantifies the time available for chemical, biological and physical processes to modify the original composition of inputs through gain or loss (e.g. [4], [5], [6], [7], [8]). However, some of the processes controlling the release of pre-event water from catchments are still poorly understood and rather roughly modeled, and the observational data do not suggest either simple, much less universal, behaviors [9], [10], [11], [12]. The complexity of the mixing patterns involving event and pre-event waters in catchments is partly a byproduct of the structural complexity of intertwined surface and subsurface hydrologic environments, which are typically characterized by pronounced heterogeneity. Within such contexts, the probabilistic characterization of travel times to a control surface (possibly conditional on injection times [13]) provides a robust and integrated stochastic description of how catchments retain and release water [9], [10], [12], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23].

Unerringly solute flux concentrations (i.e. the ratio of instantaneous mass flux and flow discharge) measured in catchment runoff depends on the mixing of water from different sources, ages and chemical compositions. Highly fluctuating concentrations in surface waters are observed, resulting in complex, hysteretic relations with discharge. In the case of pollutants like herbicides, whose lifetimes and mobility in surface and subsurface waters has long been studied (e.g. [8], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], fluctuating concentrations and their peaks (and duration) over a toxic threshold are crucial to assess the ecological risks to aquatic communities [35], [36], [37].

The assessment of catchment-scale mobility of herbicides is deemed a relevant endeavor because the use of crop protection agents is commonplace in modern agriculture. Therefore the problem of herbicides transported from the soils where they are applied to surface and groundwater is emerging as a critical factor for a variety of environmental hazards. In the last years great interest was placed in developing measures to minimize herbicide losses from agricultural fields to surface waters (e.g. [34]). Measures to reduce the damage due to herbicide losses range from the alternative use of products less harmful to the environment, to alternative field management strategies. Studies have focused on the parts of the catchment contributing more decisively to surface water pollution [38], [39], [40]. In particular, it has been shown that more than 80% of the total herbicide losses from small Swiss-catchments occurred during the first two rain events following application [41]. This is a recurrent pattern, owing to the intrinsic decay rates of pesticides and herbicides. Moreover, the influence of the variability of field-specific characteristics on total herbicide loads proves much larger than the difference in compound-specific properties, as indeed hydrologic factors tend to dominate the spatial distribution of losses [39], [40].

While important for management purposes, the above approaches do not address the issue of fluctuating concentrations in the streamflow. Indeed a full understanding of the transport of reactive chemicals such as herbicides adds layers of complexity to the features of the hydrologic response. This is contributed, in particular, by soil biogeochemistry that affects herbicide sorption and degradation, and by land use/cover and soil/crop management practices, i.e. crop distribution and growth, planting dates, timings and rates of pesticide applications [8]. Limited effective mobility at catchment scales may emerge [39] as it has been observed that on average only about 1% of applied atrazine mass is exported on an annual basis from catchments covering nine orders of magnitude [42]. This, among other consequences, implies that in-stream processing of, say, atrazine is negligibly small, and that much of the atrazine applied is either retained or transformed in the soils. Herbicides reach surface waters through runoff and drainage mostly during and immediately after rain events, resulting in highly time-variant river concentrations. Hence, the input of herbicides into aquatic environments often occurs in pulses rather than via smooth continuous flows, producing highly erratic exposures of aquatic life to dangerous concentrations. Traditional ecological risk assessments do not reflect this fundamental time-varying character of exposure to herbicide concentrations, mostly because toxicity experiments are based on continuous exposure of an organism to a single pollutant [43]. The fluctuating character of pollutant exposure prompts new approaches to the ecological risk [35], [36], [37], for which extended monitoring is not a cost-effective option because of the high frequency needed to describe the temporal dynamics of herbicide concentrations, and the related often unmanageable costs of herbicide analysis.

Modeling is thus seen as essentially complementary tools for any assessments of water quality. In particular, we direct our efforts towards predicting instantaneous herbicide concentrations in the hydrologic response at the closure of a catchment. This paper addresses the coupling between hydrologic and transport models through the quantification of water flow and solute transport in the general framework of the formulation of transport by travel time distributions [5], [6], [13]. To this aim we generalize a recently proposed scheme for the derivation of travel time distributions [13] to the case of solutes like herbicides, which are not completely taken up by transpiration and undergo chemical degradation. Section 2 provides the analytical derivation for a single hydrologic control volume forced by generic input and output fluxes. This general solution can be readily particularized and applied to more complex and specific hydrologic schemes comprising different control volumes in series and/or in parallel. This is the case of the application presented in Section 3 where a complete model for flow and atrazine transport in the Aabach basin (Switzerland) is developed and tested in a comparative mode with the extensive experimentation carried out therein [30], [38], [39], [41], [44]. The effects of stochasticity in rainfall patterns and application dates on concentrations and loads in runoff are assessed via Monte Carlo framework. The results of this analysis are discussed in the perspective of a new approach to ecological risk assessment (Section 4). A set of relevant conclusions close the paper (Section 5).

Section snippets

Travel time formulation of transport

Let us start by considering a control volume V (see Fig. 1), which can at first be thought of as a hillslope but it can be readily generalized to any suitable sub-unit of a more complex hydrologic scheme. The macroscopic balance of water storage S(t) depends on input I(t), outflow Q(t) and evapotranspiration ET(t). As an example, if the control volume is aimed to describe the dynamics of a superficial layer, the term I(t) represents inputs from precipitation. In the case a deeper layer is

Data

The framework developed in Section 2 is applied to model herbicide transport in a Swiss agricultural catchment. An important field measurement campaign including high frequency measurement of discharges and of the flux concentrations for several herbicides was carried out in 1999 for the basin of the Greifensee [47], [43]. The case study presented focuses on catchment-scale circulation of atrazine because this herbicide is only used for corn protection and thus has a single and more

Results and discussion

Fig. 4 shows the frequency distributions of the behavioral parameter sets. The corresponding model outputs are compared with the measured data in Fig. 5, where model results are represented by the median of all the behavioral simulations. While the agreement of the hydrological signal (panel b) is good, herbicide flux concentrations are instead reproduced less accurately. Given the complexity of the processes controlling concentration fluctuations, however, our results are deemed satisfactory.

Conclusions

In this study we have extended a theoretical framework for deriving time-variant travel time distributions to the case of solute that can be partially taken up by transpiration and undergo decay/degradation. The solution presented, which is derived under the assumption that mobilization of soil water involves randomly sampled particles from the available storage, allows the derivation of the solute mass flux flowing out of a generic hydrologic control volumes as a function of the hydrologic

Acknowledgments

The authors wish to thank two anonymous reviewers for their instrumental comments. Support from the ERC Advanced Grant RINEC 227612 and from SNF-FNS Projects 200021-124930/1 and 200021-135241 are gratefully acknowledged. The authors thank the Swiss Federal Institute for Environmental Science and Technology (EAWAG) for sharing discharge and concentration measurements.

References (53)

  • M. Brusseau et al.

    Sorption nonideality during organic contaminant transport in porous-media

    Crit Rev Environ Control

    (1989)
  • M. Flury

    Experimental evidence of transport of pesticides through field soils – a review

    J Environ Quality

    (1996)
  • B. Bird et al.

    Lightfoot, transport phenomena

    (1960)
  • A. Rinaldo et al.

    Basin scale-model of solute transport

    Water Resour Res

    (1987)
  • A. Rinaldo et al.

    On mass response functions

    Water Resour Res

    (1989)
  • S. Zanardo et al.

    Dominant controls on pesticide transport from tile to catchment scale: lessons from a minimalist model

    Water Resour. Res

    (2012)
  • J. McDonnell

    A rationale for old water discharge through macropores in a steep, humid catchment

    Water Resour Res

    (1990)
  • A. Rodhe et al.

    Transit times for water in a small till catchment from a step shift in the oxygen 18 content of the water input

    Water Resour Res

    (1996)
  • K.J. Beven

    Rainfall–runoff modelling: the primer

    (2001)
  • J.J. McDonnell et al.

    How old is streamwater? Open questions in catchment transit time conceptualization, modelling and analysis

    Hydrol Process

    (2010)
  • G. Botter et al.

    Transport in the hydrologic response: travel time distributions, soil moisture dynamics, and the old water paradox

    Water Resour Res

    (2010)
  • G. Dagan

    Flow and transport in heterogeneous porous formations

    (1987)
  • M. Stewart et al.

    Modeling base-flow soil–water residence times from deuterium concentrations

    Water Resour Res

    (1991)
  • D. Burns et al.

    Quantifying contributions to storm runoff through end-member mixing analysis and hydrologic measurements at the Panola Mountain Research Watershed (Georgia, USA)

    Hydrol Process

    (2001)
  • M. Hrachowitz et al.

    Regionalization of transit time estimates in montane catchments by integrating landscape controls

    Water Resour Res

    (2009)
  • M. Hrachowitz et al.

    Catchment transit times and landscape controls-does scale matter?

    Hydrol Process

    (2010)
  • Cited by (44)

    • A review of modeling pesticides in freshwaters: Current status, progress achieved and desirable improvements.

      2023, Environmental Pollution
      Citation Excerpt :

      Results are provided at the basin or sub-basin scale. Bertuzzo et al. (2013) defined a mathematical model to assess hydrology and herbicide transports. Weather, pesticide application rates and properties (half-life) are needed to build up the model.

    • Characterizing fast herbicide transport in a small agricultural catchment with conceptual models

      2020, Journal of Hydrology
      Citation Excerpt :

      Data to inform their effective parameters is often lacking, even at smaller scales (e.g. Marín-Benito et al., 2014) and their high computational cost tends to prohibit comprehensive assessments of their large uncertainty. Models based on residence time distributions have also been applied to characterize daily to weekly average streamflow and concentrations of herbicides (Bertuzzo et al., 2013). However, it seems challenging to reflect experimentalist knowledge about specific processes with such approaches.

    • Real-time forecasting of pesticide concentrations in soil

      2019, Science of the Total Environment
      Citation Excerpt :

      The challenge to forecast pesticide residue levels into the future therefore should incorporate the variability in the way the weather is likely to occur in the weeks and months after pesticide application. Recent approaches have attempted to quantify the role that rainfall variability has in the movement of water and chemicals through soil (McGrath et al., 2008a, 2008b; Botter et al., 2008; Harman et al., 2011; Suweis et al., 2011), from soils to streams (Zanardo et al., 2012; Bertuzzo et al., 2013) or as urban wash-off (Daly et al., 2014). The strength of these stochastic approaches is the transparent understanding of how rainfall variability propagates to retention or solute transport variability.

    View all citing articles on Scopus
    View full text