A water quality model to support water management for the Bay of Gdansk J. van Gils," W. Robakiewicz,* M. Robakiewicz/ M. Bielecka-Kieloch/ L.

Transcription

1 A water quality model to support water management for the Bay of Gdansk J. van Gils," W. Robakiewicz,* M. Robakiewicz/ M. Bielecka-Kieloch/ L. Postma" ^ D# Netherlands Jli/^a^Zzc^ f 0 Boz ^77, 2GOO ME, & PoKs/t Academy o/ 5"c2e?%ceg, /mafzwe o/ 7, & ABSTRACT The present paper deals with a mathematical water quality model that is able to support policy making for the Bay of Gdansk, with respect to eutrophication and oxygen problems. A hydrodynamic model computes the transport coefficients for the water quality model, which contains all relevant state variables and all relevant physical, chemical and biological processes, including nutrient cycles and primary production by two types of algae. The model performs a dynamic simulation of the water quality. The model has been calibrated and is being applied for the computation of waste reduction scenarios. INTRODUCTION The Bay of Gdansk (fig.l) suffers from various water quality problems. The beaches impose a danger to public health because of bacterial pollution [1]. The increased input of nutrients during the last decades causes a eutrophication problem, with increasing algae concentrations and decreasing oxygen concentrations in the deep parts where a persistent stratification exists. The biggest source of pollution is the Vistula river (fig. 1), which leads through different industrial zones in Poland, many of them operating with old fashioned and highly polluting technologies. Besides the Vistula also local sources in the region around Gdansk and Gdynia contribute to the pollution problem. The study presented herein was executed from August 1991 till December 1991 as a case study in a technology transfer project. The project was sponsored by the government of The Netherlands in the framework of the Cooperation Programme for Eastern Europe (PSO).

2 192 Water Pollution OVERVIEW OF METHODS Advection diffusion equation An integral water quality model of the Bay of Gdansk has been set up that comprises the transport of dissolved and suspended constituents and various physical, chemical and biological processes. In order to deal with oxygen and eutrophication problems the model contains the following state variables: salinity (for the verification of the transport of dissolved substances), carbonaceous waste (BOD) and dissolved oxygen, several forms of nitrogen, phosphorus and silicon (nutrients) and two types of phytoplankton. The basic model equation is the well-known advection diffusion equation [2]: Figure 1: Geographic position of the project area = V ( D VC - CK ) + S, V = (,, dt dx dy dz (1) On the right hand side the source term S represents waste loads and various physical, chemical and biological processes. Physical processes: vertical mixing The transport coefficients in the water quality model will be discussed later. For the vertical diffusion coefficient, we derive the wind induced vertical diffusion in well-mixed systems from relations given by Fisher [3], Wu [4] and Munck-Anderson [5]: Ku,H «= 7.10-* ( Ri vo.75 )' (2) Biochemical processes The water quality model includes a number of water quality processes that contribute to the oxygen budget and the nutrient cycles. All these are incorporated in the model equation by means of the source and sink term S in

3 Water Pollution 193 eq. (1). The formulations used are conventional and have been reported by the author [6]. They include the mineralization of BOD and dead algae and the related consumption of oxygen, the settling of dead algae and the nitrification and denitrification reactions. The presence of algae plays a key role in the oxygen budget and the nutrient cycles. The model distinguishes between diatoms and all other non-diatom species. It has been reported by the author [6]. The model includes respiration, mortality and primary production. The latter is governed by the availability of nutrients and light. MODEL IMPLEMENTATION Numerical solution of the governing equations The water quality modelling package DELWAQ [7] forms the basis for the water quality model for the Bay of Gdansk. The DELWAQ package solves the advection diffusion equation numerically by means of some 10 different finite volume methods. The package allows the user to adapt the source term that Figure 2: Model schematization/oxygen deficit in the lower layer after the calibration year contains the water quality processes to the requirements of the problem under consideration. Spatial scales and model schematization Considering the subject of the case study, the eutrophication problem, we are interested in the larger spatial scales of the Bay as a whole. The bay has therefore been divided horizontally into 71 computational elements, which can reflect concentration gradients on the scale of the bay (fig.2). The depth of the study area ranges from 1 m in Puck Bay south of Hel

4 194 Water Pollution Peninsula to 100 m in the Gdansk Deep (central part of the northern model boundary). The available water quality measurements reveal an almost isohaline layer with a thickness of m. During the summer a thermocline exists at approximately 40 m. The measurements show that in general the water quality in this layer is uniform. Therefore, the water quality model features two layers, with the interface at a depth of 40m. Temporal scales, hydrodynamic coefficients, residence time The Bay of Gdansk has a large residence time (several months) and since the study focuses on the eutrophication problem, for which the relevant time scale is the changing of the seasons, the water quality computations have been based on the average flowfield. This average flowfield has been computed by a hydrodynamic model [8]. The model was driven by a characteristic Vistula river flow (800 nf/s) and by the wind velocity and direction which represent the average wind shear stress (2.4 m SW). We applied the 2D vertically averaged model to the top 40 m layer of the water column, being the top layer of the water quality model. The dispersion coefficient has been calibrated so that the model would predict the correct salinity in the top 40 m layer (fig.4a). The result was a uniform value of 200 mvs. Based on the calibrated salinity model, it was possible to compute the residence time of the bay: the total (average) fresh water content of the bay divided by the fresh water flow yields a value of about 100 days. Boundary conditions and point sources The implementation of the model requires information concerning the concentrations of all state variables at the Baltic model boundary as well as information concerning the pollution loads. We distinguish three different sources: river inputs (mainly Vistula river), inputs from cities, treatment works /istula I I Gdansk region KXSN Atmosphere Figure 3: Relative contribution of different pollution sources

5 Water Pollution 195 and industries and atmospheric deposition. Figure 3 shows the relative contribution of all three types of loads to the total input of BOD, nitrogen and phosphorus in From this figure it is clear that the river Vistula is the main source of pollution to the Bay of Gdansk. MODEL CALIBRATION Measurements, initial conditions, model parameters The water quality model has been calibrated on water quality measurements in the Bay from 7 cruises during The comparison was done for 6 locations, for averaged concentrations over the top 40 m layer. (The measurements did not show any consistent sign of stratification within this layer.) The simulation was done over a period of one year. The initial conditions were derived from a steady state computation with the winter pollution loads and without any biochemical process acting. The latter assumption is realistic because the winter temperature and therefore the biochemical activity are low. The model parameters were estimated from literature and modified within the range that literature provides. It was not always possible to reproduce the measurements exactly, but the levels of the concentrations of all state variables could be reproduced. For parameters with a clear seasonal variation also this variation was reproduced adequately. Figure 4 shows some typical calibration results: a: chloride (g/1), b: BOD; (mg/1), c: organic nitrogen (mg/1), d: nitrates (mg/1), e: dissolved oxygen (mg/1), f: chlorophyl-a (/xg/1). Algae dynamics No measurements were available for the biomass of algae or the primary production. However, the general characteristics of the algae model were in agreement with the literature on this subject ([9]). Algae grow in the model from the beginning of April until October (fig.4f). There is a bloom of diatoms in April and May and a second bloom of nondiatoms in June and July. The computed annual primary production varies from 125 (Baltic boundary) to 350 (mouth Vistula) g.mtyeart Oxygen budget in the lower layer Part of the dead organic matter in the upper layer settles to the subhalocline layer in the deep parts of the bay. This material is mineralized by biochemical activity, a process which consumes oxygen. Due to this process, the oxygen concentration of the lower layer tends to decrease. The oxygen content of the lower layer is replenished by vertical transport and by the input of salt North Sea water through the Danish Straits. The input of salt North Sea water with a high oxygen content through the Danish Straits occurs irregularly during storm periods. Different inflow events may be separated by periods of several years without any significant inflow. Such events are clearly visible in the oxygen concentrations in deep parts of the Baltic, e.g. the Gotland Deep [10]. The occurrence of such events can not be

6 196 Water Pollution Chloride (mg/l) ouuu - * y *** 400o- $ * * 300o- BOD-5 (mg/l) ) - u 1 * \ * I 1^s i I 0.5 * Q! i ( )0 C Organic nitrogen (mg/l) rt 15 Nitrates (mgn/l) $ /^\ / \ * J \ C JO C)issolved oxygen (mg/l) $ 8- * A^ * * 0.2! [ f\ I / \ 0,2 s \ 0.1!, i /, \ 0, 0.0! \ \ 1 \, ^^JWL--'*, )hlorophyl-a (ug/l) 25 r ^\ 10- \ **/ * \^^/ A / \ i I W \ I \ \ i \^ 1 \^ \ ' \ Figure 4: Calibration results element 14; a: chloride (g/1), b: BODy (mg/l), c: organic nitrogen (mg/l), d: nitrates (mg/l), e: dissolved oxygen (mg/l), f: chlorophyl-a (jwg/1).

7 Water Pollution 197 predicted in the present model set up. Figure 2 shows the computed undersaturation of oxygen after the calibration year, starting from a saturated initial state, and assuming that no inflow of North Sea water occurs. The deficit amounts up to 0.15 g.m'^.year* after one year. This number is an indication towards the sensitivity of the system for oxygen problems in the deeper layer. MODEL APPLICATION The model has been used to compute the effects of measures by the water management authorities. As an example we present the following imaginative measure: all sources within the Gdansk region are reduced 100%. This measure was evaluated in order to demonstrate the dominating effect that the water quality of the river Vistula has on the water quality in the Bay. Table 1 shows the results. Effect of measure on: Element 19 Element 62 Element 69 Total nitrogen -4% -2% -1% Total phosphorus -21% 48% 43% Annual primary prod. 43% -20% 43% Oxygen deficit lower layer (after 1 year) (shallow) 49% 43% Table 1: result of 100% reduction of pollution sources in the Gdansk region CONCLUSIONS The analysis of the water quality in the Bay of Gdansk by means of an integrated approach taking into account hydrodynamics, transport physics and biochemical processes has produced satisfactory results. The available information about waste loads, together with generally accepted modelling techniques of transport and water quality processes, proved sufficient to reproduce the water quality of the upper layers of the Bay of Gdansk with a mathematical model. The level of eutrophication and oxygen problems predicted by the model, agrees with the available information on this subject. In its present form the model may be used to analyze the effect of waste load scenarios and changes in the meteorological and hydrological conditions on the oxygen and eutrophication problems in the Bay of Gdansk. Tentative computations indicate that the pollution supplied by the Vistula river is the main cause of water quality problems in the bay. The contribution of regional sources seems to be of secondary importance.

8 198 Water Pollution The present project has proven the feasibility of water quality modelling to support water management in the Gdansk region. At this moment the results of the present study form the basis for a full water quality management analysis. This follow up project involves the national, regional and local water management authorities in Poland as well as several scientific institutes in Poland. The follow up project will set a first step towards investment plans for the remediation of the Bay of Gdansk and the Baltic. REFERENCES [1] Bielecka-Kieloch, M. and Robakiewicz, M., 'Bacteria pollution in the Polish coastal zone', in Water Pollution 93 (to be published), Proceedings of the Second International Conference on Water Pollution, Milan, [2] Crank, J., The mathematics of diffusion, Clarendon Press, Oxford, [3] Fisher, H.B. e.a., Mixing in Inland and Coastal Waters, Academic Press, New York, [4] Wu, J., 'Wind stress and surface roughness at air-sea interface' Journal of Geophysical research 74, pp , [5] Munck, W.H., and Anderson, E.R., 'Note on the theory of the thermocline' Journal of Marine Research 1, p.276, [6] Van Gils, J.A.G., Ouboter, M.O. and De Rooij, N.M., 'Modelling of water and sediment quality in the Scheldt Estuary', submitted to the Journal of Aquatic Ecology, [7] Van Gils, J.A.G., Vos, R.J. and Postma, L., DELWAQ water quality program, Definition document, DELFT HYDRAULICS, [8] Robakiewicz, W., Okrieslenie warunkow pradowych na Zatoce Gdanskiej przy pomocy modelu matematycwego, Polska Akademia Nauk, In sty tut Budownictwa Wodnego, Gdansk, 1992 (in Polish). [9] Voipio, A. ed., The Baltic Sea, Elsevier oceanographic series 30, Amsterdam, [10] Gerlach, S.A., Nitrogen, phosphorus, plankton and oxygen deficiency in the German Bight and in the Kieler Bucht (project Eutrophication of the North Sea and the Baltic Sea, Institut fur Meereskunde, Universitat Kiel, LIST OF MATHEMATICAL SYMBOLS C concentration (g/m*) t time (s) D dispersion coefficient (mvs) u* friction velocity (m/s) H water depth (m) V velocity (m/s) K constant of Von Karman (0.4) W,Q wind speed at 10 m (m/s) Ri Richardson number (-) x,y,z spatial coordinates (m) S source term (g/m^s)