Elsevier

Advances in Water Resources

Volume 111, January 2018, Pages 58-69
Advances in Water Resources

Simulation of nitrate reduction in groundwater – An upscaling approach from small catchments to the Baltic Sea basin

https://doi.org/10.1016/j.advwatres.2017.10.024Get rights and content

Highlights

  • Spatially targeted regulation of agricultural practices can decrease N-loads.

  • We use local-scale models to simulate impact of a spatially targeted regulation.

  • We develop a methodology to upscale knowledge from local- to large-scale models.

  • The upscaling enables E-HYPE to simulate the impact on N-loads to the Baltic Sea.

Abstract

This paper describes a modeling approach proposed to simulate the impact of local-scale, spatially targeted N-mitigation measures for the Baltic Sea Basin. Spatially targeted N-regulations aim at exploiting the considerable spatial differences in the natural N-reduction taking place in groundwater and surface water. While such measures can be simulated using local-scale physically-based catchment models, use of such detailed models for the 1.8 million km2 Baltic Sea basin is not feasible due to constraints on input data and computing power. Large-scale models that are able to simulate the Baltic Sea basin, on the other hand, do not have adequate spatial resolution to simulate some of the field-scale measures. Our methodology combines knowledge and results from two local-scale physically-based MIKE SHE catchment models, the large-scale and more conceptual E-HYPE model, and auxiliary data in order to enable E-HYPE to simulate how spatially targeted regulation of agricultural practices may affect N-loads to the Baltic Sea. We conclude that the use of E-HYPE with this upscaling methodology enables the simulation of the impact on N-loads of applying a spatially targeted regulation at the Baltic Sea basin scale to the correct order-of-magnitude. The E-HYPE model together with the upscaling methodology therefore provides a sound basis for large-scale policy analysis; however, we do not expect it to be sufficiently accurate to be useful for the detailed design of local-scale measures.

Introduction

Large nutrient loads have resulted in severe environmental problems in the Baltic Sea during the past decades (Gustafsson et al., 2012). Diffuse losses of nutrients contribute about two thirds of the total nitrogen (N) and phosphorus (P) loads delivered to the sea with the majority of these originating from agricultural sources (HELCOM, 2011). In order to address the eutrophication problem, the Baltic Sea Action Plan (BSAP) was adopted in 2007, setting country specific abatement targets for N and P loads to achieve a “good status” in the Baltic Sea (Backer et al., 2010). Many abatement actions have already been implemented in the Baltic Sea countries, but the targets are still not met and more actions are required (Backer et al., 2010). This paper addresses simulation of the impacts of a novel regulation methodology in decreasing the total N loads to the sea.

The loss of N from agriculture occurs mainly as leaching of excess nitrate from the root zone. Along the flow path from the root zone via groundwater and surface water to the catchment outlet, nitrate can be naturally transformed and thereby removed by redox processes. Nitrate reduction is a microbial process occurring under anoxic conditions in the presence of an electron donor (organic matter, pyrite and Fe(II)) and occurs both in soils, groundwater and surface waters (Appelo and Postma, 2005). This removal of nitrogen by redox reactions is in this study referred to as “N-reduction” and we mainly focus on the N-reduction occurring below the root zone, i.e., in groundwater and surface waters.

Due to geological heterogeneity, N-reduction varies significantly, not only at the Baltic Sea basin scale (Wulff et al., 2014, Højberg et al., 2017) but also at the local-scale within small catchments (Hansen et al., 2014b). Present agricultural regulations to decrease the N-loss from the root zone specify the same abatement requirements for all areas without considering local variations in N-reductions in the groundwater or surface water. This is obviously not very cost-effective (Jacobsen and Hansen, 2016), and it has led to the idea of implementing a spatially targeted regulation focussing on decreasing N-loss from areas with low natural N-reduction, instead of a uniform regulation where all areas have to decrease by the same amount regardless of the natural reduction capacity in the area (Hansen et al., 2017). A spatially targeted regulation can consist of a number of different regulatory measures implemented at or in connection to areas with a low natural N-reduction. These regulatory measures can either be implemented at the source by decreasing N-leaching from the root zone on areas with low N-reduction, e.g. by changing the crop or the management practice on the field, or they can be implemented where water from areas with low N-reduction is discharging into surface waters, e.g. in the form of constructed wetlands that can remove some of the discharging nitrogen (Langergraber et al., 2011).

The effectiveness of regulatory measures for decreasing N-loads can be assessed using hydrological and nutrient flux models (Vagstad et al., 2009). To exploit the local (field and farm-scale) variability in N-reduction, however, requires simulations at scales resolving the field and farm-scale variability. This is extremely data demanding and requires detailed process descriptions within the hydrological models (Hansen et al., 2014b, Karlsson et al., 2016). Modeling of nitrate processes at large scales, such as the Baltic Sea basin, has been carried out by Arheimer et al. (2012) and Wulff et al. (2014) to assess the impacts of international regulations that are applied uniformly across larger regions or even the entire basin. These large-scale models operate with computational units of several hundred km2, making it impossible to explicitly simulate the impacts of local-scale spatially targeted measures, because they cannot spatially resolve the measures and often have inadequate process descriptions, in particular for describing the spatial variability in groundwater processes. Therefore, there is a need to combine the knowledge and results achieved by local- and large-scale modeling in an upscaling procedure.

Upscaling can be done in a variety of ways (Bloschl and Sivapalan, 1995, Refsgaard et al., 1999, Vereecken et al., 2007). The most common upscaling approach in distributed hydrological modeling is the effective parameter approach, which assumes that process equations and system data originating from smaller scales are applicable at a larger scale and that effective parameters exist that can reproduce the mean behavior of the system observed at the larger scale. This assumption is often justifiable (Refsgaard, 1997, Henriksen et al., 2003), while it in some cases has to be rejected (Beven, 1995). Another approach used in distributed models is the distribution function approach, where the statistical distributions, but not the geo-referenced locations, of system data and parameters are represented in the model (Andersen et al., 2001, Herbst and Diekkruger, 2002). The use of hydrological response units in semi-distributed models like HYPE (Arheimer et al., 2012, Stromqvist et al., 2012) can be seen as a kind of statistical distribution function approach, although the continuous properties here are replaced by categorical data such as soil type and land use.

A fundamentally different upscaling approach is the dynamic upscaling approach, where model results from a model with a finer resolution of the computational units are utilized somehow in a large-scale model with coarser resolution. There are only a few examples of this upscaling approach (Bronstert et al., 2007, Hansen et al., 2008). In the study by Bronstert et al. (2007), a HBV model was set up for the Rhine basin. For three meso-scale catchments within the Rhine basin, small-scale models were set up using the distributed physically-based model WASIM-ETH, and simulated stream discharge from these models was used in the calibration of the large-scale HBV model. The effect of land use changes was simulated with the small-scale models, and these results were then used as input to the HBV model in order to simulate land use changes for the entire Rhine basin.

To be able to assess the potential impacts of implementing a spatially targeted regulation of agricultural practices on the N-loads to the Baltic Sea, it will be necessary to use a large-scale model for the entire Baltic Sea Basin and to develop an upscaling methodology to account for the impacts that cannot be handled explicitly in existing large-scale models. The objectives of the present study were i) to introduce a methodology to upscale knowledge from a local-scale model to a large-scale model simulating water and N flows for the entire Baltic Sea basin with a focus on N-reduction in groundwater; and ii) to test it on predicting impacts of a spatially targeted regulation to decrease the N-load to the sea for two study areas. We use the dynamic upscaling approach in a manner similar to Bronstert et al. (2007), extending this approach from only considering river flow to also including N-fluxes and N-reduction. In this paper we thus report the development and testing of the upscaling approach, whereas the full application on the entire Baltic Sea basin will be reported in a second paper.

Section snippets

Local study areas

Two Danish catchments were used to develop the upscaling methodology. The 85 km2 Norsminde catchment (upstream part of the 101 km2 Norsminde Fjord catchment) is located on the east coast of Jutland, and the 486 km2 Odense catchment (upstream of Kratholm) is located on the island of Funen (Fig. 1). Both catchments discharge into fjords, which are ecologically sensitive with respect to nitrogen, and a significant decrease of the N-loads to the fjords is required to obtain good ecological status (

Results

In this section we show first the results from the impact scenario for Norsminde and Odense using the local-scale models (Section 3.1). Based on these impact results the upscaling relationships used in step 3 of the upscaling approach were developed (Section 3.2). Finally, after developing the upscaling relationships, the upscaling approach was tested on the two study areas Norsminde and Odense (Section 3.3).

Upscaling approach – principles

In this paper we have presented a methodology to upscale knowledge on the impact of a spatially targeted regulation of agricultural practices from two local-scale physically based MIKE SHE catchment models to the large-scale and more conceptual E-HYPE model. The future goal is to then use the large-scale model to simulate changes for the entire Baltic Sea basin. We used the local-scale models to develop a relationship between the impact of spatially targeted regulation in a catchment and the

Acknowledgments

This work was carried out in the BONUS Soils2Sea project, which received funding from BONUS, the joint Baltic Sea research and development programme (Art 185), funded jointly from EU's Seventh Programme for research, technological development and demonstration and from Innovation Fund Denmark, The Swedish Environmental Protection Agency, The Polish National Centre for Research and Development, The German Ministry for Education and Research, and The Russian Foundation for Basic Researches

References (45)

  • I.B. Karlsson et al.

    Combined effects of climate models, hydrological model structures and land use scenarios on hydrological impacts of climate change

    J. Hydrol.

    (2016)
  • J.C. Refsgaard

    Parameterisation, calibration and validation of distributed hydrological models

    J. Hydrol.

    (1997)
  • J.C. Refsgaard et al.

    Large scale modelling of groundwater contamination from nitrate leaching

    J. Hydrol.

    (1999)
  • J.C. Refsgaard et al.

    Uncertainty in the environmental modelling process – a framework and guidance

    Environ. Model. Softw.

    (2007)
  • S. Stisen et al.

    Model parameter analysis using remotely sensed pattern information in a multi-constraint framework

    J. Hydrol.

    (2011)
  • H.E. Andersen et al.

    Identifying hot spots of agricultural nitrogen loss within the Baltic Sea drainage basin

    Water Air Soil Pollut.

    (2016)
  • C.A.J. Appelo et al.

    Geochemistry, Groundwater and Pollution

    (2005)
  • B. Arheimer et al.

    Climate change impact on riverine nutrient load and land-based remedial measures of the Baltic Sea action plan

    Ambio

    (2012)
  • J.G. Arnold et al.

    Automated base-flow separation and recession analysis techniques

    Ground Water

    (1995)
  • K. Beven

    Linking parameters across scales - Subgrid parameterizations and scale-dependent hydrological models

    Hydrol. Process.

    (1995)
  • G. Bloschl et al.

    SCALE issues in hydrological modeling – a review

    Hydrol. Process.

    (1995)
  • A. Bronstert et al.

    Multi-scale modelling of land-use change and river training effects on floods in the Rhine basin

    River Res. Appl.

    (2007)
  • Cited by (17)

    • Impacts of land use, climate change and hydrological model structure on nitrate fluxes: Magnitudes and uncertainties

      2022, Science of the Total Environment
      Citation Excerpt :

      As it is not possible to measure leaching data on catchment scale, no observation data exist to validate this number directly, except indirectly through crop yield data. However, the results are close to the 37.5 kg N/ha/y found by the Danish National Nitrate Model (Højberg et al., 2017; Højberg et al., 2015), and the 41.8 kg N/ha/y and 35.1 kg N/ha/y found by Hansen et al. (2018) in Odense using both HYPE and a NLES-MIKE SHE model system, respectively. Hence, the values estimated in the current study accord with other estimates.

    • Numerical groundwater flow and nitrate transport assessment in alluvial aquifer of Varaždin region, NW Croatia

      2022, Journal of Hydrology: Regional Studies
      Citation Excerpt :

      Groundwater flow and solute transport modeling has become an essential tool for studying spatio-temporal distribution of nitrate in groundwater. The modeling framework most commonly relies on either utilizing lumped models, e.g. LPMs in Hajhamad and Almasri (2009), E-HYPE in Hansen et al. (2018), BICHE in Surdyk et al. (2021), or spatially distributed models, e.g. integrating the MODFLOW code for the simulation of the groundwater flow (McDonald and Harbaugh, 1988) and MT3DMS code for the simulation of nitrate transport (Zheng and Wang, 1999). Numerous regional studies have been conducted by combining MODFLOW and MT3DMS codes, as the problem with nitrate contamination of groundwater occurs worldwide.

    • Downscaling a national hydrological model to subgrid scale

      2021, Journal of Hydrology
      Citation Excerpt :

      Other studies suggested that models generally only have predictive capability at spatial scales larger than the model grid scale (Wood et al., 1988, Refsgaard et al., 2016). Consequently, it is commonly assumed that accurate simulation of local hydrogeological responses requires a high resolution, local scale model whose parameters are optimised to the processes and observations at that scale (e.g., Hansen et al. 2018). In the following we use the term downscaling for a process, whereby a coarse scale model is used to construct a model with a finer spatial resolution.

    • Reducing uncertainty of estimated nitrogen load reductions to aquatic systems through spatially targeting agricultural mitigation measures using groundwater nitrogen reduction

      2018, Journal of Environmental Management
      Citation Excerpt :

      Leaching of nitrogen (N) in the form of nitrate (NO3−) from agricultural land is a significant environmental issue in many parts of the world, since this severely affects the quality of groundwater and surface waters (Hashemi et al., 2016). In Denmark, the N-load from agriculture to surface waters caused the eutrophication of marine environments that is one of the major problems (Højberg et al., 2017; Hansen et al., 2018) in water resource management. By imposing national regulations on agricultural land and nutrient management, the N-load has been halved over the last two decades (Dalgaard et al., 2014; Jacobsen et al., 2017).

    View all citing articles on Scopus
    View full text