Elsevier

Water Research

Volume 43, Issue 6, April 2009, Pages 1788-1800
Water Research

Water and phosphorus mass balance of Lake Tegel and Schlachtensee – A modelling approach

https://doi.org/10.1016/j.watres.2009.01.007Get rights and content

Abstract

Management models for aquatic systems can be used to determine which measures in the watershed or in the water body have been effective and/or which one should be used in future. The newly developed management models presented in the following for Lake Tegel and Schlachtensee are empirical and lake specific. The values for the unknown factors are estimated by an iterative process using optimisation routines and sensitivity analysis methods. The resulting models describe the water and phosphorus balance of each lake. The Lake Tegel water balance model calculates the unknown water inflow from the River Havel depending on the other main in- and outflows with very good validation results. The phosphorus models of both lakes quantify mixing of the upper and lower water body as well as sedimentation and release from the sediment as functions of measured variables. For Lake Tegel, management scenarios were run indicating effective management interventions. For Lake Schlachtensee, the phosphorus model captured the variations in the hypolimnion well but produced poorer results for the epilimnion because of unknown external phosphorus loads. For these the model indicated possible sources and magnitudes.

Introduction

A central question in lake restoration planning is whether external measures alone are sufficient to achieve restoration targets or whether internal measures are necessary as well. Assessing the external and internal P loads requires complete water and phosphorus (P) mass balances, but often, not all of the necessary data can be measured. A management model for a specific lake can help to fill data gaps in the water and nutrient balance. It is useful both for planning the measures, and for following the lake's restoration response, i.e. to assess which measures prove effective and which need to be changed or fine-tuned.

This paper describes the development and application of models for two lakes in Berlin which were subject to major restoration measures. The management models developed for Lake Tegel and Schlachtensee are simple mass balance models, empirical and lake specific. In contrast, most lake models are either dynamic (time dependent) process models which usually need to be calibrated for a given lake, e.g. Janse and van Liere, 1995, Reynolds et al., 2001, or they are empirical models which can be generally applied, e.g. Vollenweider, 1976, OECD, 1982, Jensen et al., 2006. Simple static empirical models require the lake to be in steady state and cannot describe the transient phase after reduction of nutrient loading in combination with high internal load. They tend to underestimate the P content of a lake, particularly where internal loading prevails over a long time period (Sas, 1989, Søndergaard et al., 2003). Simple two-box models which combine a water phase and a sediment phase with sedimentation and sediment release (Nürnberg and LaZerte, 2004) mainly operate with 1-year time steps and cannot describe the seasonal variations of the TP-concentration in a deep, stratified lake (Jensen et al., 2006). Complex dynamic-process models can model seasonal changes of stratified lakes, but require in-depth investigations to quantify process parameters (Jørgensen and Mitsch, 1983). There are also some general process models (e.g. Janse, 2005, Håkanson and Boulin, 2002) which have a great value to predict long-term means because their predictive quality is higher than general empirical models and they can be applied without extensive calibration of parameters. However, they also poorly represent the seasonal and annual variability of measured values (Bryn and Håkanson, 2007), possibly because particularly P release from sediments is very difficult to describe with general parameter values (Mieleitner and Reichert, 2006). This drawback is particularly significant for lakes on the verge of moving from eutrophic, cyanobacteria-dominated, to mesotrophic with more diverse phytoplankton, because in these situations, seasonal patterns of P-maxima are critically important for determining phytoplankton patterns.

Empirical and dynamic lake-specific models (termed “management models” in the following) can combine the advantages of both approaches. They need less parameters than lake-specific process models, since they do not describe processes in detail, and they can model seasonal changes in a stratified lake with a multi-box approach. Their uncertainty is less than that of both empirical models and of general process models since they are calibrated for the lake in question. Such management models can help to understand the key factors driving the water and phosphorus balances better than static empirical models can, as they can be used to run scenarios to estimate the effects of lake management measures – both in retrospective and for planning further measures. While empirical lake-specific models are a simplified alternative to more complex lake-specific process models, both can also be used in combination: if values for a process parameter (e.g. sedimentation rates) are not available, a management model could be used instead of a lake-specific process model. If input data are not available (e.g. the P load from a certain source), a management model can quantify the missing data empirically using information about the drivers of the missing input variable. The key limit of lake-specific management models, however, is that in contrast to general models, they are not transferable to other settings. Like all lake-specific models their adaptation depends on good data sets in adequate temporal scale. The longer the time record for a lake, the more accurate is the evaluation of the unknown parameters, of the natural variability and of the effects of former measures for the lake's management.

The management models presented here were developed

  • 1.

    to establish complete water and phosphorus mass balances over the past 20–25 years for Lake Tegel and Schlachtensee;

  • 2.

    to quantify and understand the processes and key factors driving the phosphorus balance of each of these two lakes;

  • 3.

    to identify effective further management options for further reduction of in-lake P concentrations.

Modelling was necessary to assess influences that were postulated to be important, but which are not amenable to direct measurement. Unknowns in the water balance of Lake Tegel were the River Havel inflow and the proportion of water exfiltrating through the abstraction of drinking water by bank infiltration. For Schlachtensee the unknowns were the inflow from the storm water overflows and diffusive surface water inflow. The latter may be a relatively important carrier of phosphorus. For both lakes, a main modelling target was to differentiate the external and the internal phosphorus load by establishing a phosphorus balance and using it to estimate sedimentation and release of phosphorus from the sediment, as these processes are very difficult to measure in situ.

Both lakes are situated in western Berlin, close to the River Havel. In response to intensive eutrophication problems, phosphorus elimination plants (German abbreviation: OWA) were installed at their main inflows in the early 1980s, resulting in substantial decrease of total P concentrations (from several hundred μg L−1 P before restoration to a few tens of μg L−1 after restoration) as well as of phytoplankton biomass (Schauser and Chorus, 2007). However, trophic recovery patterns and mechanisms differed between both lakes, with Schlachtensee showing the more pronounced response (see below). Understanding the causes of these differences in response patterns required differentiation between external and internal phosphorus sources.

For Lake Tegel in the north-west of Berlin, the main inflow is from the east (Nordgraben and Tegeler Fliess). At the confluence of Nordgraben and Tegeler Fliess, the OWA Tegel went into operation in 1985 together with a lake pipeline through which water is pumped from the River Havel into the OWA to maintain a minimum discharge of the OWA (set at rates between 1.7 and 3 m s−1 during different restoration phases). Additionally, at its western shore the lake is multiply connected to the River Havel, which still carries a high nutrient load. Thus, the entire upstream Havel watershed is part of Lake Tegel's watershed. Since Lake Tegel is one of the most important drinking water resources of Berlin, most of its shoreline is surrounded by wells for abstracting bank infiltrate together with varying and undefined shares of groundwater. The local water works also take some water straight from the lake and pump it into shallow basins for artificial groundwater recharge. Since 1979, aerators have been installed in the lake to keep the hypolimnion oxic. They have been operated in a variety of modes over the last decades (Schauser and Chorus, 2007). Furthermore, the ammonium and nitrate concentrations have changed considerably, since before 1992 high ammonium loads reached the lake via Tegeler Fliess. In 1992, the wastewater treatment plant upstream established a nitrification step, shifting the ammonium load to a nitrate load. After 2000, denitrification basins were taken into operation, thus the complete nitrogen load decreased.

Schlachtensee's main inflow is a pipe through which River Havel water has been pumped into the lake since 1913 to keep the lake from desiccating. This became necessary after water works went into operation nearby, abstracting groundwater and lowering the groundwater table by several meters. The water of Schlachtensee flows into Krumme Lanke, which is the next lake in a chain of lakes connected by channels or ditches (“Grunewaldseenkette”). Some water from Schlachtensee is also pumped into Waldsee. The pumping of nutrient rich River Havel water into the lake chain has lead to severe eutrophication of Schlachtensee and the downstream lakes in the chain since the 1970s. Therefore an OWA was established in 1981 to extract the phosphorus from the Havel water before it flows into Schlachtensee. Additionally, phosphorus rich water was withdrawn from the hypolimnion of Schlachtensee during a couple of weeks at the end of each summer stratification period 1981–1996. Schlachtensee is cut into a valley, which is the direct watershed of Schlachtensee (37.3 ha). The lake is surrounded by woods, a suburban housing area and some park. The lake and its shoreline are intensively used for recreation. It is also used as recipient for storm water overflows from the suburb.

The principal physical parameters characterizing Lake Tegel and Schlachtensee are given in Table 1. Because of the morphological complexity of the shallow interaction zone between the River Havel and Lake Tegel we consider only the main basin of Lake Tegel. Both lakes are seasonally stratified. From the analysis of the temperature profiles, the boundary between the upper (epilimnion) and lower compartment (hypolimnion) was set at 6 m depth for Schlachtensee and at 8 m depth for Lake Tegel.

Section snippets

Methods

Mass balance equations of water, chloride (Cl) and P on a monthly basis were established for both lakes. By using Cl as a conservative tracer (as it is neither taken up by organisms nor removed by precipitation) the complete water mass balance was calculated for both lakes as time integrated mass balance using ModelMaker® Vs. 4 (Family Genetix). Unknown parameters were estimated with an iterative process of calibration, optimisation and sensitivity analysis, using iterative numerical methods of

Lake Tegel: water mass balance

We calculated the Cl and water balances using a one-box model approach (Fig. 1A), because the similarity of Cl concentrations in 0.5, 7 and 14 m depth did not indicate any Cl accumulation or dilution in the epilimnion or in the hypolimnion.

The fluxes (F) of Cl were calculated as product of the concentration (C) of Cl and the concomitant water flow (Q). Because of the hydrogeological setting of Lake Tegel, groundwater inflow into the lake is very unlikely (Pekdeger, 2004). The fraction of lake

Discussion

Modelling the water balance proved successful for both lakes, as indicated by the very close fit between Cl concentrations predicted by the model to those actually measured for the years used for validation. This filled an important data gap, as for both lakes not all inflows and outflows can be measured, and provided the basis for the lake-specific phosphorus budget models. For both lakes, model results confirmed that oligotrophication is the result of the reduction of the total external load,

Conclusions

  • 1.

    The modelling techniques proved useful for filling the gaps that otherwise precluded the establishment of P balances, an important basis for lake restoration decisions. A key advantage of these validated, lake-specific models is that they require only few parameters which could be derived from known data.

  • 2.

    For Lake Tegel, the chief data gap filled by our model was one major hydrological inflow, for which Cl and P concentrations are known. Thus, we could build a model describing the lake's P

Acknowledgements

The financial support to the study, a research project of Kompetenzzentrum Wasser Berlin, was provided by Veolia Water and the Interreg IIIC project LakePromo. The Berliner Wasserbetriebe, including the OWA Tegel and Beelitzhof, and the Senat of Berlin supported this study by conducting the analyses for phosphorus and provision of data. Many years of sampling and laboratory analyses by Antje Köhler, Hans Ulrich Wolf, Elke Pawlitzky, Gertrud Schlag, Christa Kopplin, Ingrid Klinkmueller, Thomas

References (28)

  • B. Hölting

    Hydrogeologie. Einführung in die Allgemeine und Angewandte Hydrogeologie

    (1992)
  • Janse, J.H., 2005. Model studies on the eutrophication of shallow lakes and ditches. PhD thesis, Wageningen University,...
  • J.P. Jensen et al.

    An empirical model describing the seasonal dynamics of phosphorus in 16 shallow eutrophic lakes after external loading reduction

    Limnol. Oceanogr.

    (2006)
  • Cited by (31)

    • A modelling framework to track phosphorus sources of the drinking water intakes in a large eutrophic lake

      2022, Journal of Hydrology
      Citation Excerpt :

      However, the modelling framework considered various processes including sediment–water interface, water transfer from Wangyu River, water exchange from central Lake Taihu. Compared with the existing studies on tracking P sources at watershed scale (Table S4), lake scale model tracking phosphorus sources mainly focused on phosphorus mass balance (Schauser and Chorus, 2009). Our modelling framework not only focused on tracking phosphorus sources in lake scale, but also paid more attention on these areas (e.g., drinking water intakes) should be highly concerned in the emergency management of nutrients in Lake Taihu.

    • Modelling the effects of nutrient loads reduction and testing the N and P control paradigm in a German shallow lake

      2015, Ecological Engineering
      Citation Excerpt :

      While the bulk P release is caused by the continuous mineralisation taking place in sediments, sharp summer release peaks emerge during stratification periods and are triggered by desorption during times of higher mineralisation activity and FeS precipitation. Increased phosphorus release from sediments during summer stratification was observed by Schauser and Chorus (2009) in the lake Tegel, a branch of Havel River upstream of our studied section. In our studied section, summer measurements of P release from sediments in 2012 gave values as high as 25 mg P m−2 d−1 while external loads at the same time were below 15 mg P m−2 d−1 (Lindim et al., 2013).

    • Seeking a compromise between pharmaceutical pollution and phosphorus load: Management strategies for Lake Tegel, Berlin

      2012, Water Research
      Citation Excerpt :

      The modeling efficiency (indicator of goodness of fit) for the period 1995–2010 is EF = 0.48, which is comparable to that of the empirical box model by Schauser and Chorus (2009), EF = 0.51 for the validation period 1986–1990. For a better comparison, we calculated the EF values of both models for the overlapping period 1995–2002 and obtained values of 0.59 (this study) and 0.63 (Schauser and Chorus, 2009). The slightly higher modeling efficiency of the simpler box model can be attributed to the fact that the box model was fitted to the chloride data for the period 1991–2002.

    View all citing articles on Scopus
    View full text