Analysing nitrate losses from an artificially drained lowland catchment (North-Eastern Germany) with a mixing model

Abstract

Tile drainage shortens the residence time of water in the soil and may therefore aggravate the diffuse pollution of adjacent surface water bodies. To assess the environmental impacts of tile drainage on surface water bodies, it is important to assess how the drainage discharge and its solute signal translate from the frequently studied plot scale to the catchment scale. We used results from the automated hydrograph separation method ‘recursive digital filter’ in combination with a two-component mixing model to quantify the role of the different flow components and flow paths for the nitrate-nitrogen losses at three different scales – collector drain outlet (4.2 ha), ditch (179 ha) and brook catchment (15.5 km²) – in a pleistocene lowland area in North-Eastern Germany. Measured and modelled NO₃-N concentrations of three 6-months winter seasons agreed reasonably well. At the tile drainage plot, the fast flow component was responsible for 63–91% of the total simulated nitrate-nitrogen losses. The stated ranges were derived from all accepted model runs. This flow component was interpreted as a fast component bearing nitrate from the nutrient-enriched topsoil. Tile drainage itself delivered 89–95% of the total nitrate losses in the ditch catchment. In the brook catchment, at most 25% of the area was responsible for 54–85% of the NO₃-N losses. Although the mixing model is limited by the assumption of constant component concentrations and conservative behaviour of the solutes, it has shown to be a useful tool for hydrochemical studies. Overall, the results emphasise the importance of tile drainage for the catchments’ hydrochemistry and its environmental impact on the larger scale. Consequently, it will be difficult to significantly reduce diffuse pollution in an artificially drained lowland landscape on the catchment scale without addressing the issue of tile drainage. As a next step for model validation, other solutes such as sulphate and chloride could be added to reduce the uncertainty, and grassland should be explicitly included into the mixing model.

Introduction

Tile drainage is a common agricultural practice which is used to improve moisture and aeration conditions especially in lowland areas, but shortens the residence time of water in the soil and therefore aggravates diffuse pollution of adjacent surface water bodies with nutrients and pesticides (e.g. David et al., 1997, Heppell and Chapman, 2006, Tomer et al., 2003). For example, Behrendt and Bachor (1998) estimated that 47% of the nitrogen and 12% of the phosphorus emissions from the federal state Mecklenburg-Vorpommern (North-Eastern Germany) to the Baltic Sea originated from tile drainage.

However, given that the area contributing to the tile drain discharge can be defined and an impermeable layer underlies the tile drains, tile-drained fields can be considered as large lysimeters and are therefore ideal for the quantification of solute transport at the plot and field scale (Lennartz et al., 1999, Richard and Steenhuis, 1988). Numerous investigations (e.g. Heppell and Chapman, 2006, Kohler et al., 2003, Lennartz et al., 1999, Stamm et al., 1998) on tile drainage have been conducted at the field scale, from where the diffuse pollution originates. However, there are only a few catchment scale studies on this subject (Tomer et al., 2003) although this is frequently the relevant management scale, as, for example, for the European Water Framework Directive (European Parliament and European Council, 2000). To assess the environmental impacts of tile drainage on surface water bodies, it is therefore a matter of particular interest how the tile drain discharge and its solute signal translate from the field to higher scales.

Frequently, tile drainage is accompanied by flow anomalies causing a further acceleration of water and solute fluxes (Kohler et al., 2003, Lennartz et al., 1999). This combination of tile drainage and preferential flow may therefore create a ‘short circuit’ between contaminated topsoils and receiving water bodies (Laubel et al., 1999, Stamm et al., 1998). In catchment hydrology, hydrograph separation is carried out to split the total streamflow into a baseflow component and a surface runoff component. When transferring hydrograph separation methods to tile drain discharge data, it is generally assumed that the fast component originates from ‘preferential flow’ (Everts and Kanwar, 1990, Richard and Steenhuis, 1988), from ‘event water’ (Heppell and Chapman, 2006) or, more generally, from a ‘rapid flow component’ (Laubel et al., 1999), while the baseflow is thought to correspond to matrix flow. To our knowledge, only manual linear baseflow separation methods (Laubel et al., 1999), often combined with mixing models (Everts and Kanwar, 1990, Kohler et al., 2003, Richard and Steenhuis, 1988), have been applied for tile drain discharge data so far. For larger catchments, in contrast, automated methods are widespread (e.g. Arnold et al., 1995, Müller et al., 2003), which also do not have a real physical basis, but have a better reproducibility and are more objective than manual methods (Chapman, 1999, Müller et al., 2003). Another approach to split the hydrograph into different components or origins is the use of isotopic and tracer data for mixing models (Heppell and Chapman, 2006, Joerin et al., 2002, Ladouche et al., 2001, Uhlenbrook and Hoeg, 2003). A simple, yet frequently used method is the end-member mixing approach (EMMA, Hooper et al., 1990), which assumes that the solute concentrations in a stream are the result of the mixing of two or more constant sources (‘end members’) (e.g. Durand and Torres, 1996, Soulsby et al., 2003, Wade et al., 1999).

In this paper, we aim to quantify the role of different flow components and flow paths for the NO₃-N losses at the different spatial scales using data of a hierarchical monitoring program. This program was set up in an agriculturally dominated lowland area in North-Eastern Germany, and high nutrient concentrations have been measured at a collector drain outlet, in ditches and in a brook (Tiemeyer et al., 2006). For this purpose, we utilise the automated hydrograph separation method ‘recursive digital filter’ (Nathan and McMahon, 1990) in combination with a simple two-component mixing model.