Water quality modeling in Berze River catchment.
Nutrient leakage models have not been applied in Latvia in order to estimate water quality and pollution source apportionment. Therefore, in co-operation with the Swedish University of Agricultural Sciences (personal conversation with M. Wallin, A. Gustafson, M. Larsson) water quality monitoring and modeling framework was developed for Berze River (Fig. 1). To establish an empirical link between river headwaters and main stem of the river (Smith 2003), a multiscale monitoring approach was proposed.
This included water quality and load measurements at three different monitoring scales, i.e. drainage field, small catchment and medium sized river. These measurements are already being carried out as part of the proposed catchment measurement program (Kyllmar et al. 2006) and partly through the state financed monitoring programs.
Additional data are collected in connection with reporting to the EU Nitrates Directive. Simulation results are used to identify this river's catchment specific nutrients (nitrogen and phosphorus) pollution load distribution (Povilaitis 2008), and retention rates (Kneis et al. 2006). This could be useful for water protection measures regarding the EU Water Framework Directive river basin management approach.
1. Materials and methods
The Berze Piver catchmont (Fig. 1) was selected as the study area for water quality modeling in Latvia. The Berze River is a part of Lielupe Piver basin district that is one of four river basin districts designated according to the Water Framework Directive.
Most of the Berze river is located within vulnerable zone according to the EU Nitrates Directive.
Starting the year 2005 water samples at 15 Berze River subcatchments were collected on monthly and seasonal basis (Fig. 1), to characterize the water quality (Vadas et al. 2007) of eiver stages (Subcatchment ID 2, 3, 6, 9, 12, 15), major tributaries (ID 4, 8, 10, 13, 14) and various types of land use impacts, e.g. agrtculture (ID 14.) tile drained area (No. 15), Dobele city (IDS 12), forest (ID 10, 11), lake (ID 5) and peat bog (ID 1), Annenieku hydro-power plant (ID 6), management of organic manure from animal husbandry (ID 7).
To emphasize the modeling period, additionally, two time series (2000-2005) both for toral nitrogen and phosphorus concentrations before anV aiser Dobele city were atfVeV from Latfian Environment, Geology anV Meteorology (Centre (LEGMC) monitoring. One meteorological station Dobele is locateV within Berze River catchment.
[FIGURE 1 OMITTED]
There are available data for 2 LEGMC hydrological gauging stations Berze-Balozi and Berze-Biksti with daily discharge measurements since 1951 (Table 1). Unfortunately, nowadays station Berze-Biksti is closed, but still existing data sets are available for model calibration of this catchment.
Long-term agricultural run-off monitoring data (1994-2010) collected by Latvia University of Agriculture (LLU) in the monitoring station "Berze" including measurements in small agricultural catchment (3.68 [km.sup.2]) and drainage field (76.6 ha) are representative for agricultural production levels and trends (Klavins et nl. 2001; Klavins, Kokorite 2002) as well as type-specific concentrations for arable land.
2. Description of study area
Berze River is situated at the central part of Latvia and is the tributary of Svete River that inflows into Lielupe River and then into Gulf of Riga. The length of the Berze is 109 km (slope 108 m per 109 km) and the river catchment covers an area of ~872 [km.sup.2].
Berze River starts in drained meadows in Southern part of Eastern-Courland highland (~120m abovethe sea level) with slightly hilly surroundings and steep banks. In the middle part of basin there is a hydro-power plant "Annenieki" with reservoir that could influence nutrient retention.
Then Berze flows through Dobele city. In the downstream part of Berze River (land level ~10 m above the sea level) large sub-surface drainage systems are constructed.
[FIGURE 2 OMITTED]
Last 6.5 km before the inlet into Svete River the riverbed of Berze is straightened and dams system of polder separates river from surrounded drainage area. Middle and downstream part of Berze River catchment is typical for Zemgale region plains with highly intensive agricultural land in the catchment. Normal year water balance: precipitation 630 mm, run-off 200 mm and evaporation 430 mm (Table 1).
3. Model description
The dynamic FyrisNP model calculates source apportioned gross and net transport of nitrogen and phosphorus in rivers and lakes (Hansson et al. 2008). The main scope of the FyrisNP model (Fig. 2) is to assess the effects of different nutrient reduction measures on the catchment scale. The time step for the model is in the majority of applications one month and the spatial resolution is on the sub-catchment level.
[FIGURE 3 OMITTED]
Retention, i.e. losses of nutrients in rivers and lakes through sedimentation, up-take by plants and denitrification, is calculated as a function of water temperature, potential nitrogen concentration and lake area, and stream area. The model is calibrated with regard to two retention parameters, [k.sub.vs] (retention parameter, m/year) and c0 (temperature parameter, dimension less), using time series on measured nitrogen and phosphorus concentrations in subcatchments (Hansson et al. 2008). In order to evaluate the fit of simulated to measured values the model efficiency E, and the correlation coefficient r are used.
The definition of the FyrisNP model efficiency:
E = 1 -[[n.summation over (i=1)[([[THETA].sub.obs,i]-[[THETA].sub.sim,i]).sup.2]]/ [[n.summation over (i-1)][([THETA].sub.obs,i]-[[bar.[THETA]].sub.obs]).sup.2]],
where: n is the number of observations, and [[bar.[THETA]].sub.obs] is the mean value of all observations. The [THETA] symbolizes whatever time-series are compared. In the FyrisNP model, [[THETA].sub.obs] and [[THETA].sub.sim] are the observed and modelled concentrations, respectively.
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
Data used for calibrating and running the FyrisNP (Hansson et al. 2008) model can be divided into time-dependent data, e.g. time series on observed nitrogen and phosphorus concentration, water temperature, runoff and point source discharges, and timeindependent data, e.g. land-use information (Table 2) according to CORINE Land Cover 2000 (CLC 2000), lake area and stream length and width.
[FIGURE 6 OMITTED]
4. Results and discussion
The conceptual FyrisNP model (Hansson et al. 2008) was chosen to identify the impact of the sources of pollution with total nitrogen (N) and phosphorus (P) in the Berze River. The modeling encompasses the time period from 2000-2010. There is relatively high mire (ID 1) and the forest (ID 10) background Ntot average concentrations, 1.83 and 2.08 mg [l.sup.-1], respectively, but nitrate nitrogen concentrations in these subcatchments (0.50 and 1.1 mg [l.sup.-1] are among the lowest in Berze River catchment (Fig. 3).
Observed water quality data (Fig. 3) show significant differences between average concentrations (Vuorenma et al. 2002) of natural background (bogs, forests) and anthropogenic impacted areas (Dobele city and agricultural land). It was also found that N[O.sub.3]-N ratio against [N.sub.tot] is higher in agricultural lands (70-85%) compared to forests (~53%) or bogs (27%). Similar ratios are typical also for P[O.sub.4]-P against [P.sub.tot,] i.e. agricultural lands (70%-81%), forests (45-55%) or bogs (40%).
One of the FyrisNP model tasks is to determine the type-specific pollution concentrations for load calculations. For this purpose, in Berze River, subcatchments were preselected with significant mire (bogs), forest and agricultural land share information on diffuse pollution concentrations (Stalnacke et al. 2003). Seasonal fluctuation of total nitrogen concentrations shows higher values during winter and spring periods (Fig. 4) whereas total phosphorus concentrations are high in summer period compared to winter season (Fig. 5).
[FIGURE 7 OMITTED]
There are no mountains in Berze River catchment and thus it was possible to improve the FyrisNP model calibration process (Hansson et al. 2008). Data needed for mountain monthly type-specific concentrations were replaced by arable land type-specific monthly concentrations. The type-specific concentrations of arable land were derived from long-term agricultural run-off monitoring data ([N.sub.tot] 7.4 mg [l.sup.-1] and [P.sub.tot] 0.165 mg [l.sup.-1]) provided by Latvia University of Agriculture. Afterwards mountain pollution loads were referred to arable land.
[FIGURE 8 OMITTED]
After the calibration for the period 2000-2010 (132 months), the model efficiency coefficient for nitrogen was E =0.498, fairly good, and the correlation coefficient was r =0.71, but for the phosphorus model E =0.28 and r =0.60 (Figs 6 and 7).
To estimate mean retention (Table 3) for each subcatchment model output results for gross and net contribution is taken into account:
Retention [%] = 100 (Gross-Net) [Gross.sup.-1].
Internal gross contribution (before retention) and net contribution (after retention) is given for entire period of 11 years. Originally projected significant retention of total nitrogen and phosphorus in Annenieku HPP reservoir (ID 6) has not been confirmed even the model results show highest nitrogen and phosphorus retention rate in lake subcatchment (id 5) 41.3% and 69.1%, respectively. This could be explained by fast water turnover in reservoir (Vassiljev, Stalnacke 2005).
[FIGURE 9 OMITTED]
Load compilation for both total nitrogen and total phosphorus is based on gross contribution (Table 3): Load [kg [ha.sup.-1] [year.sup.-1]] = Gross * [Area.sup.-1][Years.sup.-1].
Nowadays the nutrient loading from diffuse sources is the major source of anthropogenic nutrients in many areas since water protection measures have been applied to point sources (Povilaitis 2008). Agriculture is the main source of diffuse loads. Diffuse source impact of the each subcatchments is calculated as percentage distribution of leakage using the weighted average which is calculated by multiplying each subcatchment area with an average concentration (type-specific) and the annual runoff volume divided by the total nitrogen or phosphorus loads (tons [year.sup.-1] or kg [year.sup.-1]
[FIGURE 10 OMITTED]
[FIGURE 11 OMITTED]
Namely, nitrogen and phosphorus releases from each subcatchment (Monaghan rt nl. 2007) area divided by the total nutrient loads per year, resulting in the proportional distribution:
[Impact.sub.i], [%] = 100 ([A.sub.i] R [Conc.sub.(i)]) [L.sup.-1.sub. t0t]
[A.sub.i]--type-specific area for diffuse sources, ha;
R--annual run-off in Berze River, mm;
[Conc.sub.(i)]--type-specific concentration, mg [l.sup.-1];
[L.sub.tot]--total annual load of nitrogen or phosphorus from the subcatchment, kg.
In order to implement river basin management plans both total loads (Figs 8 and 9) and diffuse source apportionment (Figs 10 and 11) estimations should be given to decision makers. For example, total loads per each subcatchment give an impression of priorities where the pollution potential is the highest (ID 14) and which subcatchments to treat first, while diffuse source apportionment pie-diagrams (Figs 10 and 11) show the background (forest, bogs) and anthropogenic (agricultural land, arable land and pasture) pollution apportionment (Pieterse et al. 2003).
Assuming that background pollution is rather nonsense to treat then the plan of action should be set only for agricultural, arable and pasture dominant subcatchments (Figs 10 and 11) with regard of appropriate pollution reduction measures. If more stringent measures are not taken to reduce emissions from agriculture, the improvement of a water quality may turn out to be too small to achieve good status in water bodies. After the proper model calibration it will be possible to assess future climate scenarios of water quality with a variety of contributing to pollution or cutting measures, as well as the impact of climate variability.
1. Accurate and precise model calibration requires hydro-chemical database that covers the period of observation with various hydro-meteorological conditions for more than 5 years.
2. The FyrisNP model calibration needs to be improved--model efficiency for nitrogen is E = 0.498, and the correlation coefficient is r =0.71, but for the phosphorus model E = 0.28 and r = 0.60.
3. It was found that N[O.sub.3]-N ratio against [N.sub.tot] as well as P[O.sub.4]-P ratio against [P.sub.tot] is higher in agricultural lands compared to forests or bogs.
4. The model results show highest nitrogen and phosphorus annual retention coefficients in lake subcatchment (ID 5)-41.3% and 69.1%, respectively, but the lowest in Dobele city (ID 12) subcatchment-2.9% and 7.3%, respectively.
5. The output results of pollution source apportionment on the subcatchment basis could help river basin management decision makers to point out the catchments for agricultural mitigation measures.
Caption: Fig. 1. Location of Berze River and water quality monitoring network
Caption: Fig. 2. Structure of FyrisNP model inputs and outputs (Hansson rt ql. 2008)
Caption: Fig. 3. Average N and P concentrations [mg l-1] in Berze River subcatchments (2005-2010)
Caption: Fig. 4. Total nitrogen type-specific concentrations for diffuse source pollution
Caption: Fig. 5. Total phosphorus type-specific concentrations for diffuse source pollution
Caption: Fig. 6. Modeled and observed total nitrogen concentrations in FyrisNP model
Caption: Fig. 7. Modeled and observed total phosphorus concentrations in FyrisNP model
Caption: Fig. 8. Ntot loads and source apportionment in Berze River subcatchments (2000-2010)
Caption: Fig. 9. Ptot loads and source apportionment in Berze River subcatchments (2000-2010)
Caption: Fig. 10. Ntot diffuse source apportionment in Berze River subcatchments (2000-2010)
Caption: Fig. 11. Ptot diffuse source apportionment in Berze River subcatchments (2000-2010)
The authors gratefully acknowledge the funding from ESF Project "Establishment of interdisciplinary scientist group and modeling system for groundwater research" (2009/0212/ 1DP/184.108.40.206.0/09/APIA/VIAA/060EF7) and EU BONUS program RECOCA (Reduction of Baltic Sea Nutrient Inputs and Cost Allocation within the Baltic Sea Catchment).
Submitted 16 May 1012; accepted 11 Nov. 2012
Hansson, K.; Wallin, M.; Djodjic, F.; Lindgren, G. 2008. The FyrisNP Model Version 3.1--A Tool for Catchment-Scale Modelling of Source Apportioned Gross and Net Transport of Nitrogen and Phosphorus in Rivers. Technical description. ISSN 1403-977X. SLU, Uppsala. 17 p.
Klavins, M.; Kokorite, I. 2002. Long term changes of river water discharge in Latvia, Boreal Environment Research 32: 567-591.
Klavins, M.; Rodionov, V.; Kokorite, I.; Klavina, I.; Apsite, E. 2001. Long-term and seasonal changes in chemical composition of surface waters in Latvia, Environmental Monitoring and Assessment 66: 233-251.
Kneis, D.; Knoesche, R.; Bronstert, A. 2006. Analysis and simulation of nutrient retention and management for a lowland river-lake system, Hldrologl and Earth Slstem Sciences 10: 575-588. http://dx.doi.org/10.5194/hess-10-575-2006
Kyllmar, K.; Carlsson, C.; Gustafson, A.; Ulen, B.; Johnsson, H. 2006. Nutrient discharge from small agricultural cachments in Sweden. Characterisation and trends, Agriculture, Ecoslstems & Environment 115: 15-26. http://dx.doi.org/10.1016/j.agee.2005.12.004
Monaghan, R. M.; Wilcock, R. J.; Smith, L. C.; Tikkisetty, B.; Thorrold, B. S.; Costall, D. 2007. Linkages between land management activities and water quality in an intensively farmed catchment in southern New Zealand, Agriculture, Ecosystems & Environment 118: 211-222. http://dx.doi. org/10.1016/j.agee.2006.05.016
Pieterse, N. M.; Bleuten, W.; Jorgensen, S. E. 2003. Contribution of point sources and diffuse sources to nitrogen and phosphorus loads in lowland river tributaries, Journal of Hydrology 271: 213-225. http://dx.doi.org/10.1016/S0022-1694(02)00350-5
Povilaitis, A. 2008. Source apportionment and retention of nutrients and organic matter in the Merkys river basin in southern Lithuania, Journal of Environmental Engineering and Landscape Management 16: 195-204. http://dx.doi.org/10.3846/1648-6897.2008.16.195-204
Smith, V. H. 2003. Eutrophication of freshwater and coastal marine ecosystems, a global problem, Environmental Science and Pollution Research 10: 126-139. http://dx.doi.org/10.1065/espr2002.12.142
Stalnacke, P.; Grimvall, A.; Libiseller, C.; Laznik, M.; Kokorite, I. 2003. Trends in nutrient concentrations in Latvian rivers and the response to the dramatic change in agriculture, Journal of Hydrology 283: 184-205. http://dx.doi.org/10.1016/S0022-1694(03)00266-X
Vadas, P. A.; Gburek, W. J.; Sharpley, A. N.; Kleinman, P. J. A.; Moore, Jr. P. A.; Cabrera, M. L.; Harmel, R. D. 2007. A model for phosphorus transformation and runoff loss for surface-applied manures, Journal of Environmental Quality 36: 324-332. http://dx.doi.org/10.2134/jeq2006.0213
Vassiljev, A.; Stalnacke, P. 2005. Statistical modelling of riverine nutrient sources and retention in the Lake Peipsi drainage basin, Water Science and Technology. 51: 309-317.
Vuorenma, S.; Rekolainen, S.; Lepisto, A.; Kenttamies, K.; Kauppila, P. 2002. Losses of nitrogen and phosphorus from agricultural and forest areas in Finland during 1980 and 1990, Environmental Monitoring and Assessment 76: 213-248. http://dx.doi.org/10.1023/A:1015584014417
Kaspars ABRAMENKO. Ms. Sci. Eng., Lecturer in Department of Environmental Engineering and Water Management of Latvia University of Agriculture (LLU). Master of Science in (environmental engineering) LLU, 2002. PhD thesis "Agricultural run-off monitoring in Latvia". Publications: author/co-author of 12 scientific papers. Research interests: nutrient leakage, agricultural run-off monitoring, diffuse source pollution in Latvia.
Ainis LAGZDINS. Dr. Sci. Eng., Assistant Professor in Department of Environmental Engineering and Water Management of Latvia University of Agriculture (LLU). Doctor of Science (environmental engineering) LLU, 2012. PhD thesis "Analysis of the water quality concerning nutrients in the agricultural runoff". Publications: author/coauthor of ~15 scientific papers. Research interests: water quality classes, agro-climatic regions, water quality and quantity modeling.
Arturs VEINBERGS. Ms. Sci. Eng., Assistant in Department of Environmental Engineering and Water Management of Latvia University of Agriculture (LLU). Master of Science (environmental engineering) LLU, 2010. Master thesis "Influence of Groundwater Fluctuations on Nitrate Nitrogen Runoff in Open Channels". Publications: author/coauthor of 8 scientific papers. Research interests: water quality modeling, groundwater monitoring in Latvia.
Department of Environmental Engineering and Water Management, Latvia University of Agriculture, 19 Akademijas str., LV-3001 Jelgava, Latvia
Corresponding author: Kaspars Abramenko
Table 1. Main characteristics of Berze River hydrology Average River Gauging station Observation Area Q, [m.sup.3] period [km.sup.2] [s.sup.-1] Berze Balozi 1951-2010 872 5.04 Berze Biksti 1951-1994 275 2.46 Average River q, l[s.sup.-1] [Q.sub.max], 1%, summer winter [km.sup.2] [m.sup.3] [s.sup.-1] Berze 8.06 92.6 1.21 2.65 Berze 8.10 41.8 0.65 1.40 Table 2. Land use in Berze River subcatchments (CLC 2000) Arable Mixed Subcatchment land Pasture agricultural ID ([km.sup.2]) ([km.sup.2]) land ([km.sup.2]) 1 0.41 0.11 0.24 2 9.21 8.91 11.19 3 22.97 21.05 15.65 4 5.67 7.29 4.19 5 1.11 3.11 2.15 6 2.21 1.00 0.27 7 14.58 6.85 3.04 8 32.13 12.44 7.57 9 28.40 12.38 14.14 10 6.11 4.64 8.90 11 2.78 0.29 1.67 12 1.71 1.58 2.22 13 20.18 4.95 11.75 14 53.90 5.02 11.43 15 26.49 7.22 17.63 Total in basin 227.85 96.85 112.04 Proportion, % 26.13 11.11 12.85 Urban Subcatchment Forest Clearcuts areas ID ([km.sup.2]) ([km.sup.2]) ([km.sup.2]) 1 4.82 0.15 -- 2 38.45 1.19 -- 3 58.34 1.80 -- 4 37.14 1.15 -- 5 15.88 0.49 -- 6 0.38 0.01 -- 7 17.88 0.55 0.03 8 44.71 1.38 0.63 9 46.30 1.43 0.91 10 30.40 0.94 -- 11 15.39 0.48 -- 12 2.54 0.08 4.25 13 49.75 1.54 -- 14 21.59 0.67 0.46 15 10.23 0.32 1.03 Total in basin 393.81 12.18 7.31 Proportion, % 45.16 1.40 0.84 Lake Stream Mire, Subcatchment area area wetlands ID ([km.sup.2]) ([km.sup.2]) ([km.sup.2]) 1 0.03 0.00 3.57 2 0.18 0.15 -- 3 0.62 0.45 0.27 4 0.37 0.07 1.35 5 4.81 0.01 0.34 6 0.27 0.04 -- 7 0.15 0.08 -- 8 0.45 0.19 1.44 9 1.02 0.32 0.69 10 1.35 0.06 0.59 11 -- 0.01 -- 12 0.10 0.33 -- 13 0.79 0.10 0.42 14 0.46 0.14 -- 15 0.03 0.74 -- Total in basin 10.64 2.70 8.67 Proportion, % 1.22 0.31 0.99 Subcatchment Total area ID ([km.sup.2]) 1 9.32 2 69.28 3 121.16 4 57.22 5 27.90 6 4.19 7 43.16 8 100.94 9 105.59 10 53.00 11 20.62 12 12.81 13 89.49 14 93.68 15 63.69 Total in basin 872.05 Proportion, % 100 Table 3. Pollution load and retention in Berze River subcatchments (2000-2010) Total Nitrogen Gross Net Load Mean Subcatchment contribution contribution kg [ha.sup.-1] retention ID (kg) (kg) year % 1 50562 44104 4.93 12.8 2 492138 419035 6.46 14.9 3 922488 801997 6.92 13.1 4 381253 299550 6.06 21.4 5 176065 103409 5.74 41.3 6 46313 44285 10.05 4.4 7 410500 339296 8.65 17.3 8 881331 750963 7.94 14.8 9 880060 809197 7.58 8.1 10 363109 247584 6.23 31.8 11 142165 136998 6.27 3.6 12 160041 155436 11.36 2.9 13 698509 531515 7.10 23.9 14 1059232 859919 10.28 18.8 15 646733 623024 9.23 3.7 Total Phosphorus Gross Net Load Mean Subcatchment contribution contribution kg [ha.sup.-1] retention ID (kg) (kg) year % 1 1015 726 0.10 28.5 2 5538 3880 0.07 29.9 3 16373 11807 0.12 27.9 4 9857 6061 0.16 38.5 5 2726 842 0.09 69.1 6 981 889 0.21 9.3 7 5949 4115 0.13 30.8 8 20725 15540 0.19 25.0 9 17119 14363 0.15 16.1 10 5808 2103 0.10 63.8 11 2431 2204 0.11 9.4 12 24845 23042 1.76 7.3 13 12196 6800 0.12 44.2 14 21990 14042 0.21 36.1 15 22259 20282 0.32 8.9
|Printer friendly Cite/link Email Feedback|
|Author:||Abramenko, Kaspars; Lagzdins, Ainis; Veinbergs, Arturs|
|Publication:||Journal of Environmental Engineering and Landscape Management|
|Date:||Dec 1, 2013|
|Previous Article:||Influence of hydropower dams on the state of macroinvertebrates assemblages in the Virvyte River, Lithuania.|
|Next Article:||Aplinkos apsaugos katedrai 20 metu.|