- Email: [email protected]
Source apportionment of nutrients in Estonian rivers
Desalination 226 (2008) 222–230
Source apportionment of nutrients in Estonian rivers Anatoli Vassiljeva*, Irina Blinovab, Peeter Ennetc a
Tallinn University of Technology, Ehitajate tee 5, Tallinn 19086, Estonia Tel. +372 6202551; Fax +372 6202550; email: [email protected] b National Institute of Chemical Physics and Biophysics, Akadeemia tee 23, Tallinn 12618, Estonia c Estonian Environment Information Centre, Mustamäe tee 33, Tallinn 10616, Estonia Received 18 December 2006; revised accepted 2 February 2007
Abstract Eutrophication caused by excess nutrient loads is the main problem for Estonian surface waters. The assessment of the type of human activity on the catchment area that may cause an impact on the status of a water body is needed for successful implementation of the Water Framework Directive. The lack of necessary information often makes it difficult to perform this task. The simple export coefficients approach has been used in this investigation for evaluation of the impact of different sources of nutrients on the water quality in Estonian rivers. Attention has mainly been concentrated on the diffuse pollution sources (natural and anthropogenic). Nonlinear regression was used for the estimation of the export coefficients and retention of nutrients. Results revealed that the export coefficients vary within a wide range of limits even in such a small country as Estonia. Soil type, topography, climatic conditions and water flow are the main natural factors influencing export coefficients. It was demonstrated that the use of export coefficients estimated for other regions may lead to wrong conclusions about the impact of different diffuse sources on the water body status. Keywords: Diffuse sources; Export coefficients; MESAW; Nutrients; Water Framework Directive
1. Introduction The European Union Water Framework Directive (WFD) [1] requires assessment of the pressures from human activity, which, combined with the information on the sensitivity of the receiving water body to the pressures, will identify
*Corresponding author.
those water bodies at risk of failing to meet the environmental objectives of the Directive. The size typology given in the Water Framework Directive implies that the status of all rivers with catchment areas greater than 10 km2 must be assessed [2]. It means that the impact of different pollution sources on the water quality status of the river must be estimated for hundreds of rivers in Estonia. Lack of hydrochemical data
Presented at the 10 th IWA International Specialized Conference on Diffuse Pollution and Sustainable Basin Management, Istanbul, Turkey, 18–22 September 2006. 0011-9164/06/$– See front matter © 2008 Published by Elsevier B.V. doi:10.1016/j.desal.2007.02.108
A. Vassiljev et al. / Desalination 226 (2008) 222–230
for most of the rivers requires a methodology for the evaluation of the potential danger of pollution sources for the rivers without measurements. The purpose of the paper was to examine a simple approach to accomplish the assessment of unmonitored water bodies. Eutrophication is one of the most serious problems related to the pollution of surface waters in Estonia. The nutrient concentrations exceed permitted level in many rivers. In order to effectively manage nutrient pollution reductions, it is important to estimate the influence of different nutrient sources on water quality in rivers. The Guidance Document No. 3 [3], which states a common strategy for the implementation of the WFD, proposes to use analysis of the pressures and impacts before elaboration of water protection measures. It is demonstrated in the paper that such analysis requires different methods for estimation of point and diffuse sources. The impact of point sources may be calculated on the basis of information on pressures and river characteristics. This approach is very simple, though some questions can arise in the performance of the task. The situation is more complex with diffuse pollution. Sophisticated models need much detailed information on watershed, which is not available at present and will not be in the foreseeable future. Therefore, the large number of water bodies without hydrochemical monitoring practically excludes the possibility of using sophisticated models. According to the WFD guidance [3], the evaluation of diffuse sources may be done also with the help of the export coefficients (EC) of nutrients. This simple approach is based on the idea that the nutrient load exported from a catchment is the sum of the losses from individual sources and on the assumption that, for a given climate, specific land-use will yield characteristic quantities of nitrogen and phosphorus to a receiving water body [4]. Despite its simplicity, the export coefficient model was successfully tested in the modelling of long-term nitrate losses from catchments
223
Table 1 Empirical data on nitrogen and phosphorus export coefficients (kg/ha/y) Nitrogen
Agriculture Urban Forest
Phosphorus
Min
Max
Min
Max
2.1 1.5 1.37
53.2 38.5 7.32
0.08 0.19 0.01
5.4 6.23 0.83
(e.g., [5] and [6]). However, Smith et al. [7] note the variability of export coefficients for the same land-use classes. Large differences in export coefficients for same land-use categories have been mentioned also in Arheimer and Brandt [8], Baginska et al. [9], Haggard et al. [10], Pieterse et al. [11]. Table 1 contains empirical export coefficients for different types of land-use [12]. It is evident that differences between minimal and maximal values are very large. Large differences between values of export coefficients available from published studies prevent direct use of them. Hence, export coefficients must be estimated (or at least controlled) for each region on the basis of measurements. In this paper, simple statistical methods have been used for the estimation of export coefficients for Estonian rivers. 2. Description of the method As was mentioned above, the problem is that we have to evaluate the potential impact of different pollution sources on the status of hundreds of water bodies without hydrochemical monitoring data. The method used for such evaluation is described below in more detail. The key stages of evaluation of potential impact as laid down in the WFD are [3]: • Identifying driving forces and pressures; • Identifying the significant pressures; • Assessing the impacts; • Evaluating the likelihood of failing to meet the objective (good status by the year 2015).
224
A. Vassiljev et al. / Desalination 226 (2008) 222–230
2.1. Identifying driving forces and pressures The national database of point sources has been used in order to identify point sources. The database contains information on coordinates of point sources, the list of pollutants and pollution load from each source. The coordinates give the possibility to use Geographical Information System (GIS) for locating of point sources in the catchment area. The sources of diffuse pollution have been evaluated using maps which contain information on the land-use, number of animals, etc. GIS analyses/queries were used for processing of this information.
2.2. Identifying the significant pressures The WFD requires that only significant pressures must be identified. However, cumulative influence of the large number of small sources, which are not significant individually, may be very important, especially for small rivers. Therefore, all sources from the national databases were used regardless of their size.
2.3. Assessing the impacts The next step was to estimate the influence of the identified pressures on the river status. According to the Estonian classification of the water quality of rivers, the water body status is good if concentration in 90% of the analyses is
lower than 3 mg/L for total nitrogen and 0.08 mg/L for total phosphorus [13]. First of all, it is necessary to stress the differences in impact of point and diffuse sources on concentrations in rivers. The impact of point sources is highest at the lowest river water flow when dilution is minimal. Even a relatively small annual phosphorus load may lead to high concentrations in river water in such conditions. Fig. 1(a), for example, illustrates the dependence of phosphorus concentrations on river flow for one Estonian river. It is evident that the highest concentrations of phosphorus are typical for low river flow. It means that some source of phosphorus is present which is diluted by river water flow (the higher the river water flow, the lower is concentration). Such behaviour of concentrations is typical for point sources. Other authors also demonstrate the primary importance of dilution of point source loads for the concentration of phosphorus in rivers [14,15]. Thus, the influence of point sources is maximal at low flow. On the contrary, the influence of diffuse sources often increases with river water flow. Fig. 1(b) illustrates the situation that is typical for nitrogen. Concentrations of nitrogen in river water increase with increasing flow. It means that diffuse sources are responsible for the highest concentrations. The more water comes to the catchment, the higher is the washout of pollutants from the catchment. Accordingly, different approaches must be used for analysis 7
0.30
N-tot, mg/L
P-tot, mg/L
0.25 0.20 0.15 0.10 0.05
4 3 2 1 0
0 0 (a)
6 5
5
10
15
River flow, m3/s
20
25
0 (b)
5
10
15
20
25
River flow, m3/s
Fig. 1. Concentrations of total phosphorus (a) and total nitrogen (b) vs. river flow in the Keila River in 2000–2003.
A. Vassiljev et al. / Desalination 226 (2008) 222–230
of the impact of point and diffuse sources of nutrients. Therefore, the comparison of the annual loads of point sources with loads from diffuse sources is not applicable. For example, in the Keila River the contribution of point sources to annual load is only 10%, but they are responsible for the highest concentrations of phosphorus in the river at low flow (Fig. 1(a)). To improve the situation, firstly decrease in the point source loads is necessary in spite of the fact that the contribution of diffuse sources to annual phosphorus load is almost 90%. On the contrary, in the case of nitrogen, the reduction of pressure of diffuse sources is necessary as they are responsible for the highest concentrations of nitrogen (Fig. 1(b)). 2.3.1. Assessment of the impact of point sources A simple mixing model is used to evaluate the impact of point sources as it is recommended in the Guidance [3]. The calculation of the influence of point sources on water quality is practically the calculation of the dilution of wastewater by the simple formula
Q1c1 + Q2 c2 (1) , Q1 + Q2 where c is the concentration in river after point source; c1 is the concentration in river before point source; c2 is the concentration in wastewater; Q1 is the water discharge in river before point source; Q2 is the discharge of wastewater. The concentration in river before point source c1 was estimated using upstream sources, concentration in base flow and retention. c=
2.3.2. Assessment of the impact of diffuse sources The situation is more complex for diffuse sources because the pressures and cumulative reaction of them is known only for limited number of Estonian rivers. The main diffuse sources
225
of nutrients are agriculture, urban areas, bogs, forest, and deposition from atmosphere. Assessing the impacts of these sources on a water body requires some quantitative information to describe the state of the water body itself, and the pressures acting on it. In many situations a simple approach may be suitable for assessing the impact of a pressure. The diffuse model itself can be simple, for example nutrient loss can be based on export coefficients that represent the human activity within the catchment area [3]. But, as it was mentioned in introduction, export coefficients estimated by correlation for different geographical conditions greatly vary. Thus, the use of export coefficients obtained from other regions may lead to unreliable results. Irvine et al. [16] state that even for similar catchments differences in export coefficients may be greater than a thousand percent. Moreover, some authors assert that the annual export of nutrients for the same catchment is not always comparable because of variation in annual runoff [10,17,18]. Consequently, each region and each year needs individual estimation of export coefficients. There are a large number of successful attempts to find export coefficients for nutrients by statistical methods (e.g. Grizetti et al. [19] showed quite high efficiency of statistical model). Some of them include the estimation of retention of nutrients in rivers and lakes as well. Practically all of them use correlation (linear or non-linear). The quality of available data is usually a more significant issue than the type of calculation method. In this paper, a statistical method proposed by Grimvall and Stålnacke [20] is used. The advantage of the method is that the export coefficients and retention of nutrients are evaluated simultaneously. The MESAW software for this method was developed in Estonia by Vassiljev [17] in order to simplify the use of this method. The software was developed in the MS Excel environment using the Visual Basic language for programming. This model approach uses nonlinear regression for the simultaneous estimation
226
A. Vassiljev et al. / Desalination 226 (2008) 222–230
of export coefficients for different specified land-use or soil categories and retention coefficients for pollutants in a watershed. The basic principles and major steps in the procedure are as follows: • estimation of riverine loads for the fixed time period at each monitoring site; • sub-division of the entire drainage basin into sub-basins, defined by the monitoring sites for water quality and their upstream–downstream relationships; • derivation of statistics on e.g. land-use, number of animals, soil type, lake area, point source emissions and other relevant data for each sub-basin; • using a general non-linear regression expression with loads at each sub-basin as dependent response variable and sub-basin characteristics as explanatory variables. For animals, the export coefficient expresses the proportion of wastes voided by animals, which will subsequently be exported from livestock housing and grazing land in the catchment to the drainage network [21]. The load from animals has been calculated on the basis of N and P, which come to environment with wastes of one animal per day [22]. The load at the outlet of an arbitrary subbasin is estimated in the MESAW model by the following general expression: n
R =1−
1 1 + par * fact
(3)
where par is the parameter estimated by the MESAW; fact is the empirical function based on the input data available (sub-basin area for retention in streams and relationship between lake area and lake catchment area for retention in lakes are used in this study). 2.4. Evaluating the likelihood of failing to meet the objective On the basis of the impact assessment of point and diffuse sources the likelihood of failing to meet the good status has been evaluated for all waterbodies. 3. Results and discussions
Li = ∑ (1 − R j ,i ) L j + (1 − R)( LU i + LA i ) j =1
+ (1 − R) Pi + (1 − R) Di + e i
Pi is point source discharges to waters in the sub-basin i; Di is atmospheric deposition on surface waters in the sub-basin i; R is retention in the sub-basin i; ei is the statistical error term. Annual loads have been calculated as the sum of daily nutrient fluxes. Daily fluxes were calculated as the product of water discharge and nutrient concentration for that day. Nutrient concentrations for the days without measurements were calculated by linear interpolation between measured concentrations. Retention is estimated in the MESAW by the formula
(2)
where Li is the load at the outlet of sub-basin i; Lj is the load at the outlet of the nearest upstream sub-basin j; Rj,i is retention on the way from the outlet of sub-basin j to the outlet of sub-basin i; n is the number of sub-basins located nearest upstream; LUi is losses from different types of areas to water in the sub-basin i; LAi is the load from animals in the sub-basin i;
As mentioned above, different methods have been used for point and diffuse nutrient sources to assess the influence of these sources on the state of water bodies. 3.1. Point sources Calculations revealed that about 15% of the evaluated water bodies are at risk just because of the influence of point sources. It should be mentioned that river water flow is needed for
A. Vassiljev et al. / Desalination 226 (2008) 222–230
227
Table 2 Relationships for calculation of export coefficients (EC, kg/ha/y) of nitrogen and phosphorus from different land-use categories (r is depth of runoff in mm/y)
Agricultural land Forest
Nitrogen
R2
Phosphorus
R2
EC = 0.0305r + 3.12 EC = 0.0095r + 0.56
0.80 0.62
EC = 0.0016r – 0.05 EC = 0.00012r + 0.022
0.61 0.53
calculations of the dilution of wastewater by formula (1). Since for the majority of evaluated rivers the hydrological monitoring is absent, we assumed that water discharge in a water body is proportional to its catchment area in our calculations. This approach can be successfully used at the first stage of assessment for the identification of water bodies at risk. 3.2. Diffuse sources The results of statistical methods depend on the accuracy of data. For example, it is impossible to include the evaluation of export coefficient for land-use category of a small area because its influence on the concentration in the river is lower than the precision of load estimation. It must be mentioned that only relatively large rivers are monitored in Estonia. The prevailing land-use categories for monitored rivers are arable land and forests. Therefore, only export coefficients for arable land, forest, and livestock have been estimated by correlations. Other landuse categories cover a low percentage of catchment areas and as a result their contributions to total load are negligible. The data set for the 25 monitored river basins, which included 10 years of measurements, have been used for the estimation of the export coefficients. The analysis of data showed that export coefficients depend on the annual depth of runoff (the total runoff from a drainage basin divided by its area) and vary from year to year. Positive correlation has been found between the export coefficients of nutrients and the depth of
runoff (Table 2). The results showed in Table 2 have been obtained for the annual depth of runoff in the range of 119–380 mm/year and for usual topography and combination of soils. Unfortunately we had a detailed animals database only for one year and therefore could not obtain the dependencies of export coefficients on water flow. At the present stage of assessment the values obtained for one year (with the 200 mm depth of runoff) are used: 19% of nitrogen load from animals and 1.7% of phosphorus load goes to the river. Moreover, it is necessary to take into account the decreasing of nutrient content in river water due to the different processes occurring in streams and lakes (retention of nutrients). The retention of nutrients has been evaluated separately for lakes and for the river system. The retention in lakes is needed because some river systems contain lakes and it is necessary to know the retention of nutrients from the sources located upstream of the lakes. The retention of nutrients in lakes obtained by the MESAW model is as follows
R = 1−
1 1 + 15 * ( Fl / Flc )
(4)
where Fl is the lake area; Flc is the catchment area of the lake. The results for retention in streams obtained by the MESAW for nutrients may be expressed by the formula
R = 1−
1 1 + 0.000169 * F
(5)
228
A. Vassiljev et al. / Desalination 226 (2008) 222–230
where F is the sub-catchment area. Sub-catchment areas F in formula (5) were in the limits of 10–500 km2. It must be pointed out that the use of simple statistical models hides many aspects of system complexity that might usefully be discovered using a more sophisticated approach [6]. In particular, calculations using the MESAW model showed also that export coefficients are 1.5–2 times higher for phosphorus if clay soils prevail in the watershed or for watersheds with hilly landscape (e.g. sub-basins located in southeast Estonia). Export coefficients for nitrogen were 1.5–2 times lower for such watersheds. But detailed description of dependencies on topography and type of soil needs much more measurements for statistical analysis. Usually export coefficients for phosphorus increase with increasing sub-basin surface slope, especially in winter when the soil is frozen [23]. Grizzetti et al. [24] indicate the increase of nitrogen export with slope but the calculations showed inverse dependence. It can be said that the dependencies of nutrients export on topography exist but the quantification of these dependences is practically impossible by statistical methods due to lack of information. The same situation is valid for dependence on types of soil. It is clear that clay soils increase losses of phosphorus and decrease losses of nitrogen but obtaining of dependence in numerical expression needs much more data. More sophisticated models could help to solve this problem (e.g. SWAT [25] or AVGWL [26]) if more data will be available in future. However, the approach described in this paper can be used in the current situation. The export coefficients obtained have been used for evaluating the likelihood of failing to meet the WFD objective for water bodies. Calculations showed that diffuse sources (arable land and livestock) are responsible for the pollution of about 40% of the water bodies, which are at risk of failing to meet the good status requirements by the year 2015.
For some waterbodies, for which our calculations reveal poor status, water quality was studied in 2006. The investigation showed that the real situation in controlled rivers is very close to the calculated. 4. Conclusions • The scheme proposed in this paper, which was elaborated in compliance with the Guidance [3], for the evaluation of the influence of diffuse and point sources on river status may be successfully used for creating of the preliminary list of unmonitored waterbodies at risk. This list will allow us to select the top priority waterbodies for further investigations. • This scheme includes different approaches for the estimation of the impact of point and diffuse sources on the hydrochemical status of water bodies. Our results showed that simple methods (the mixing model for point sources and export coefficients for diffuse sources) could be used successfully for the implementation of the WFD. • The analysis of the 10-year data of 25 Estonian rivers showed that export coefficients vary from year to year depending on water flow. The equations for the calculation of export coefficients for Estonian rivers for different types of land-use and different hydrological conditions are proposed in the paper. • Our calculations reveal that the impact of diffuse sources on the status of rivers is much higher than the impact of point sources for nitrogen. Both the diffuse and point sources are significant in terms of phosphorus. Acknowledgements This work was financially supported by Estonian Ministry of Education and Research (Grant No. 0142514s03) and by the Estonian Science Foundation (Grant No. 6199).
A. Vassiljev et al. / Desalination 226 (2008) 222–230
References [1] European Union (2000): Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Communities in the field of water policy, Official Journal of the European Communities, L 327/1, 22.12.2000. [2] WFD CIS Guidance Document No. 7, Monitoring under the Water Framework Directive, Published by the Directorate General Environment of the European Commission, Brussels, Jan 2003. http:// www.europa.eu.int/comm/environment/water/waterframework/guidance_documents.html [3] WFD CIS Guidance Document No. 3, Analysis of Pressures and Impacts, Published by the Directorate General Environment of the European Commission, Brussels, Dec 2002. http://www.europa. eu.int/comm/environment/water/water-framework/ guidance_documents.html [4] T.H. Robinson, A. Leydecker, A.A. Keller and J.M. Melack, Steps towards modeling nutrient export in coastal Californian streams with a Mediterranean climate, Agric. Water Manage., 77 (2005) 144–158. [5] P.J. Johnes and T.P. Burt, Nitrate in surface waters, in: T.P. Burt (Ed.), Nitrate: Processes, Patterns and Management, Wiley, Chichester, 1993, pp. 269–317. [6] T.P. Burt, Integrated management of sensitive catchment systems, Catena, 42 (2001) 275–290. [7] R.V. Smith, C. Jordan and J.A. Annett, A phosphorus budget for Northern Ireland: inputs to inland and coastal waters, J. Hydrol., 304 (2005) 193–202. [8] B. Arheimer and M. Brandt, Watershed modelling of nonpoint nitrogen losses from arable land to the Swedish coast in 1985 and 1994, Ecol. Eng., 14 (2000) 389–404. [9] B. Baginska, T. Pritchard and M. Krogh, Roles of land-use resolution and unit-area load rates in assessment of diffuse nutrient emissions, J. Environ. Manage., 69 (2003) 39–46. [10] B.E. Haggard, P.A. Moore Jr., I. Chaubey and E.H. Stanley, Nitrogen and phosphorus concentrations and export from an Ozark Plateau catchment in the United States, Biosyst. Eng., 86 (1) (2003) 75–85. [11] N.M. Pieterse, W. Bleuten and S.E. Jørgensen, Contribution of point sources and diffuse sources
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22] [23]
[24]
229
to nitrogen and phosphorus loads in lowland river tributaries, J. Hydrol., 271 (2003) 213–225. J.D. Wickham and G. Timothy, Wade watershed level risk assessment of nitrogen and phosphorus export, Comput. Electron. Agric., 37 (2002) 15–24. Regulation of the Minister of the Environment, No. 33, Classes of surface water bodies, values of quality indicators corresponding to water classes and procedure of designating of water classes (in Estonian), 22 June 2001. https://www.riigiteata ja.ee/ert/act.jsp?id=85753 W.A. House and F.H. Denison, Nutrient dynamics in a lowland stream impacted by sewage effluent – Great Ouse, England., Sci. Total Environ., 205 (1) (1997) 25–49. C.P. Mainstone and W. Parr, Phosphorus in rivers – ecology and management, Sci. Total Environ., 282–283 (2002) 25–47. K. Irvine, I. Donohue, A. Wemaere, D. Styles and W. Hobbs, Connecting land-use with water quality: sources, sinks and time-bombs, IWA Diffuse Pollution Conference Proceedings, Dublin, 2003. A. Vassiljev and P. Stålnacke, Statistical modelling of riverine nutrient sources and retention in the Lake Peipsi drainage basin, Water Sci. Technol., 51 (3–4) (2005) 309–317. P.M. Haygarth and S.C. Jarvis, Soil derived phosphorus in surface runoff from grazed grassland lysimeters, War. Res., 31 (1) (1997) 140–148. B. Grizzetti, F. Bouraoui, G. de Marsily and G. Bidoglio, A statistical method for source apportionment of riverine nitrogen loads, J. Hydrol., 304 (2005) 302–315. A. Grimvall and P. Stålnacke, Statistical methods for source apportionment of riverine loads of pollutants, Environmetrics, 7 (1996) 201–213. P.J. Johnes, Evaluation and management of the impact of land-use change on the nitrogen and phosphorus load delivered to surface waters: the export coefficient modelling approach, J. Hydrol., 183 (1996) 323–349. A. Maastik, Water protection in agriculture (in Estonian), Tallinn, Valgus, 1984. A. Vasilyev, Modeling wash-off and leaching of pollutants by spring-time flow, J. Hydrol., 159 (1994) 215–222. B. Grizzetti, F. Bouraoui and G. De Marsily, Modelling nitrogen pressure in river basins: a comparison between a statistical approach and the
230
A. Vassiljev et al. / Desalination 226 (2008) 222–230
physically-based SWAT model, Phys. Chem. Earth, 30 (2005) 508–517. [25] J.G. Arnold, R. Srinivasan, R.S. Muttiah and J.R. Williams, Large area hydrologic modeling and assessment; part I, model development, J. Am. Water Res. Assoc., 34 (1) (1998) 73–89.
[26] B.M. Evans, D.W. Lehning, K.J. Corradini, G.W. Petersen, E. Nizeyimana, J.M. Hamlett, P.D. Robillard and R.L. Day, A comprehensive gis-based modeling approach for predicting nutrient loads in watersheds, J. Spatial Hydrol., 2 (2) (2002) 1–18.
Source apportionment of nutrients in Estonian rivers Anatoli Vassiljeva*, Irina Blinovab, Peeter Ennetc a
Tallinn University of Technology, Ehitajate tee 5, Tallinn 19086, Estonia Tel. +372 6202551; Fax +372 6202550; email: [email protected] b National Institute of Chemical Physics and Biophysics, Akadeemia tee 23, Tallinn 12618, Estonia c Estonian Environment Information Centre, Mustamäe tee 33, Tallinn 10616, Estonia Received 18 December 2006; revised accepted 2 February 2007
Abstract Eutrophication caused by excess nutrient loads is the main problem for Estonian surface waters. The assessment of the type of human activity on the catchment area that may cause an impact on the status of a water body is needed for successful implementation of the Water Framework Directive. The lack of necessary information often makes it difficult to perform this task. The simple export coefficients approach has been used in this investigation for evaluation of the impact of different sources of nutrients on the water quality in Estonian rivers. Attention has mainly been concentrated on the diffuse pollution sources (natural and anthropogenic). Nonlinear regression was used for the estimation of the export coefficients and retention of nutrients. Results revealed that the export coefficients vary within a wide range of limits even in such a small country as Estonia. Soil type, topography, climatic conditions and water flow are the main natural factors influencing export coefficients. It was demonstrated that the use of export coefficients estimated for other regions may lead to wrong conclusions about the impact of different diffuse sources on the water body status. Keywords: Diffuse sources; Export coefficients; MESAW; Nutrients; Water Framework Directive
1. Introduction The European Union Water Framework Directive (WFD) [1] requires assessment of the pressures from human activity, which, combined with the information on the sensitivity of the receiving water body to the pressures, will identify
*Corresponding author.
those water bodies at risk of failing to meet the environmental objectives of the Directive. The size typology given in the Water Framework Directive implies that the status of all rivers with catchment areas greater than 10 km2 must be assessed [2]. It means that the impact of different pollution sources on the water quality status of the river must be estimated for hundreds of rivers in Estonia. Lack of hydrochemical data
Presented at the 10 th IWA International Specialized Conference on Diffuse Pollution and Sustainable Basin Management, Istanbul, Turkey, 18–22 September 2006. 0011-9164/06/$– See front matter © 2008 Published by Elsevier B.V. doi:10.1016/j.desal.2007.02.108
A. Vassiljev et al. / Desalination 226 (2008) 222–230
for most of the rivers requires a methodology for the evaluation of the potential danger of pollution sources for the rivers without measurements. The purpose of the paper was to examine a simple approach to accomplish the assessment of unmonitored water bodies. Eutrophication is one of the most serious problems related to the pollution of surface waters in Estonia. The nutrient concentrations exceed permitted level in many rivers. In order to effectively manage nutrient pollution reductions, it is important to estimate the influence of different nutrient sources on water quality in rivers. The Guidance Document No. 3 [3], which states a common strategy for the implementation of the WFD, proposes to use analysis of the pressures and impacts before elaboration of water protection measures. It is demonstrated in the paper that such analysis requires different methods for estimation of point and diffuse sources. The impact of point sources may be calculated on the basis of information on pressures and river characteristics. This approach is very simple, though some questions can arise in the performance of the task. The situation is more complex with diffuse pollution. Sophisticated models need much detailed information on watershed, which is not available at present and will not be in the foreseeable future. Therefore, the large number of water bodies without hydrochemical monitoring practically excludes the possibility of using sophisticated models. According to the WFD guidance [3], the evaluation of diffuse sources may be done also with the help of the export coefficients (EC) of nutrients. This simple approach is based on the idea that the nutrient load exported from a catchment is the sum of the losses from individual sources and on the assumption that, for a given climate, specific land-use will yield characteristic quantities of nitrogen and phosphorus to a receiving water body [4]. Despite its simplicity, the export coefficient model was successfully tested in the modelling of long-term nitrate losses from catchments
223
Table 1 Empirical data on nitrogen and phosphorus export coefficients (kg/ha/y) Nitrogen
Agriculture Urban Forest
Phosphorus
Min
Max
Min
Max
2.1 1.5 1.37
53.2 38.5 7.32
0.08 0.19 0.01
5.4 6.23 0.83
(e.g., [5] and [6]). However, Smith et al. [7] note the variability of export coefficients for the same land-use classes. Large differences in export coefficients for same land-use categories have been mentioned also in Arheimer and Brandt [8], Baginska et al. [9], Haggard et al. [10], Pieterse et al. [11]. Table 1 contains empirical export coefficients for different types of land-use [12]. It is evident that differences between minimal and maximal values are very large. Large differences between values of export coefficients available from published studies prevent direct use of them. Hence, export coefficients must be estimated (or at least controlled) for each region on the basis of measurements. In this paper, simple statistical methods have been used for the estimation of export coefficients for Estonian rivers. 2. Description of the method As was mentioned above, the problem is that we have to evaluate the potential impact of different pollution sources on the status of hundreds of water bodies without hydrochemical monitoring data. The method used for such evaluation is described below in more detail. The key stages of evaluation of potential impact as laid down in the WFD are [3]: • Identifying driving forces and pressures; • Identifying the significant pressures; • Assessing the impacts; • Evaluating the likelihood of failing to meet the objective (good status by the year 2015).
224
A. Vassiljev et al. / Desalination 226 (2008) 222–230
2.1. Identifying driving forces and pressures The national database of point sources has been used in order to identify point sources. The database contains information on coordinates of point sources, the list of pollutants and pollution load from each source. The coordinates give the possibility to use Geographical Information System (GIS) for locating of point sources in the catchment area. The sources of diffuse pollution have been evaluated using maps which contain information on the land-use, number of animals, etc. GIS analyses/queries were used for processing of this information.
2.2. Identifying the significant pressures The WFD requires that only significant pressures must be identified. However, cumulative influence of the large number of small sources, which are not significant individually, may be very important, especially for small rivers. Therefore, all sources from the national databases were used regardless of their size.
2.3. Assessing the impacts The next step was to estimate the influence of the identified pressures on the river status. According to the Estonian classification of the water quality of rivers, the water body status is good if concentration in 90% of the analyses is
lower than 3 mg/L for total nitrogen and 0.08 mg/L for total phosphorus [13]. First of all, it is necessary to stress the differences in impact of point and diffuse sources on concentrations in rivers. The impact of point sources is highest at the lowest river water flow when dilution is minimal. Even a relatively small annual phosphorus load may lead to high concentrations in river water in such conditions. Fig. 1(a), for example, illustrates the dependence of phosphorus concentrations on river flow for one Estonian river. It is evident that the highest concentrations of phosphorus are typical for low river flow. It means that some source of phosphorus is present which is diluted by river water flow (the higher the river water flow, the lower is concentration). Such behaviour of concentrations is typical for point sources. Other authors also demonstrate the primary importance of dilution of point source loads for the concentration of phosphorus in rivers [14,15]. Thus, the influence of point sources is maximal at low flow. On the contrary, the influence of diffuse sources often increases with river water flow. Fig. 1(b) illustrates the situation that is typical for nitrogen. Concentrations of nitrogen in river water increase with increasing flow. It means that diffuse sources are responsible for the highest concentrations. The more water comes to the catchment, the higher is the washout of pollutants from the catchment. Accordingly, different approaches must be used for analysis 7
0.30
N-tot, mg/L
P-tot, mg/L
0.25 0.20 0.15 0.10 0.05
4 3 2 1 0
0 0 (a)
6 5
5
10
15
River flow, m3/s
20
25
0 (b)
5
10
15
20
25
River flow, m3/s
Fig. 1. Concentrations of total phosphorus (a) and total nitrogen (b) vs. river flow in the Keila River in 2000–2003.
A. Vassiljev et al. / Desalination 226 (2008) 222–230
of the impact of point and diffuse sources of nutrients. Therefore, the comparison of the annual loads of point sources with loads from diffuse sources is not applicable. For example, in the Keila River the contribution of point sources to annual load is only 10%, but they are responsible for the highest concentrations of phosphorus in the river at low flow (Fig. 1(a)). To improve the situation, firstly decrease in the point source loads is necessary in spite of the fact that the contribution of diffuse sources to annual phosphorus load is almost 90%. On the contrary, in the case of nitrogen, the reduction of pressure of diffuse sources is necessary as they are responsible for the highest concentrations of nitrogen (Fig. 1(b)). 2.3.1. Assessment of the impact of point sources A simple mixing model is used to evaluate the impact of point sources as it is recommended in the Guidance [3]. The calculation of the influence of point sources on water quality is practically the calculation of the dilution of wastewater by the simple formula
Q1c1 + Q2 c2 (1) , Q1 + Q2 where c is the concentration in river after point source; c1 is the concentration in river before point source; c2 is the concentration in wastewater; Q1 is the water discharge in river before point source; Q2 is the discharge of wastewater. The concentration in river before point source c1 was estimated using upstream sources, concentration in base flow and retention. c=
2.3.2. Assessment of the impact of diffuse sources The situation is more complex for diffuse sources because the pressures and cumulative reaction of them is known only for limited number of Estonian rivers. The main diffuse sources
225
of nutrients are agriculture, urban areas, bogs, forest, and deposition from atmosphere. Assessing the impacts of these sources on a water body requires some quantitative information to describe the state of the water body itself, and the pressures acting on it. In many situations a simple approach may be suitable for assessing the impact of a pressure. The diffuse model itself can be simple, for example nutrient loss can be based on export coefficients that represent the human activity within the catchment area [3]. But, as it was mentioned in introduction, export coefficients estimated by correlation for different geographical conditions greatly vary. Thus, the use of export coefficients obtained from other regions may lead to unreliable results. Irvine et al. [16] state that even for similar catchments differences in export coefficients may be greater than a thousand percent. Moreover, some authors assert that the annual export of nutrients for the same catchment is not always comparable because of variation in annual runoff [10,17,18]. Consequently, each region and each year needs individual estimation of export coefficients. There are a large number of successful attempts to find export coefficients for nutrients by statistical methods (e.g. Grizetti et al. [19] showed quite high efficiency of statistical model). Some of them include the estimation of retention of nutrients in rivers and lakes as well. Practically all of them use correlation (linear or non-linear). The quality of available data is usually a more significant issue than the type of calculation method. In this paper, a statistical method proposed by Grimvall and Stålnacke [20] is used. The advantage of the method is that the export coefficients and retention of nutrients are evaluated simultaneously. The MESAW software for this method was developed in Estonia by Vassiljev [17] in order to simplify the use of this method. The software was developed in the MS Excel environment using the Visual Basic language for programming. This model approach uses nonlinear regression for the simultaneous estimation
226
A. Vassiljev et al. / Desalination 226 (2008) 222–230
of export coefficients for different specified land-use or soil categories and retention coefficients for pollutants in a watershed. The basic principles and major steps in the procedure are as follows: • estimation of riverine loads for the fixed time period at each monitoring site; • sub-division of the entire drainage basin into sub-basins, defined by the monitoring sites for water quality and their upstream–downstream relationships; • derivation of statistics on e.g. land-use, number of animals, soil type, lake area, point source emissions and other relevant data for each sub-basin; • using a general non-linear regression expression with loads at each sub-basin as dependent response variable and sub-basin characteristics as explanatory variables. For animals, the export coefficient expresses the proportion of wastes voided by animals, which will subsequently be exported from livestock housing and grazing land in the catchment to the drainage network [21]. The load from animals has been calculated on the basis of N and P, which come to environment with wastes of one animal per day [22]. The load at the outlet of an arbitrary subbasin is estimated in the MESAW model by the following general expression: n
R =1−
1 1 + par * fact
(3)
where par is the parameter estimated by the MESAW; fact is the empirical function based on the input data available (sub-basin area for retention in streams and relationship between lake area and lake catchment area for retention in lakes are used in this study). 2.4. Evaluating the likelihood of failing to meet the objective On the basis of the impact assessment of point and diffuse sources the likelihood of failing to meet the good status has been evaluated for all waterbodies. 3. Results and discussions
Li = ∑ (1 − R j ,i ) L j + (1 − R)( LU i + LA i ) j =1
+ (1 − R) Pi + (1 − R) Di + e i
Pi is point source discharges to waters in the sub-basin i; Di is atmospheric deposition on surface waters in the sub-basin i; R is retention in the sub-basin i; ei is the statistical error term. Annual loads have been calculated as the sum of daily nutrient fluxes. Daily fluxes were calculated as the product of water discharge and nutrient concentration for that day. Nutrient concentrations for the days without measurements were calculated by linear interpolation between measured concentrations. Retention is estimated in the MESAW by the formula
(2)
where Li is the load at the outlet of sub-basin i; Lj is the load at the outlet of the nearest upstream sub-basin j; Rj,i is retention on the way from the outlet of sub-basin j to the outlet of sub-basin i; n is the number of sub-basins located nearest upstream; LUi is losses from different types of areas to water in the sub-basin i; LAi is the load from animals in the sub-basin i;
As mentioned above, different methods have been used for point and diffuse nutrient sources to assess the influence of these sources on the state of water bodies. 3.1. Point sources Calculations revealed that about 15% of the evaluated water bodies are at risk just because of the influence of point sources. It should be mentioned that river water flow is needed for
A. Vassiljev et al. / Desalination 226 (2008) 222–230
227
Table 2 Relationships for calculation of export coefficients (EC, kg/ha/y) of nitrogen and phosphorus from different land-use categories (r is depth of runoff in mm/y)
Agricultural land Forest
Nitrogen
R2
Phosphorus
R2
EC = 0.0305r + 3.12 EC = 0.0095r + 0.56
0.80 0.62
EC = 0.0016r – 0.05 EC = 0.00012r + 0.022
0.61 0.53
calculations of the dilution of wastewater by formula (1). Since for the majority of evaluated rivers the hydrological monitoring is absent, we assumed that water discharge in a water body is proportional to its catchment area in our calculations. This approach can be successfully used at the first stage of assessment for the identification of water bodies at risk. 3.2. Diffuse sources The results of statistical methods depend on the accuracy of data. For example, it is impossible to include the evaluation of export coefficient for land-use category of a small area because its influence on the concentration in the river is lower than the precision of load estimation. It must be mentioned that only relatively large rivers are monitored in Estonia. The prevailing land-use categories for monitored rivers are arable land and forests. Therefore, only export coefficients for arable land, forest, and livestock have been estimated by correlations. Other landuse categories cover a low percentage of catchment areas and as a result their contributions to total load are negligible. The data set for the 25 monitored river basins, which included 10 years of measurements, have been used for the estimation of the export coefficients. The analysis of data showed that export coefficients depend on the annual depth of runoff (the total runoff from a drainage basin divided by its area) and vary from year to year. Positive correlation has been found between the export coefficients of nutrients and the depth of
runoff (Table 2). The results showed in Table 2 have been obtained for the annual depth of runoff in the range of 119–380 mm/year and for usual topography and combination of soils. Unfortunately we had a detailed animals database only for one year and therefore could not obtain the dependencies of export coefficients on water flow. At the present stage of assessment the values obtained for one year (with the 200 mm depth of runoff) are used: 19% of nitrogen load from animals and 1.7% of phosphorus load goes to the river. Moreover, it is necessary to take into account the decreasing of nutrient content in river water due to the different processes occurring in streams and lakes (retention of nutrients). The retention of nutrients has been evaluated separately for lakes and for the river system. The retention in lakes is needed because some river systems contain lakes and it is necessary to know the retention of nutrients from the sources located upstream of the lakes. The retention of nutrients in lakes obtained by the MESAW model is as follows
R = 1−
1 1 + 15 * ( Fl / Flc )
(4)
where Fl is the lake area; Flc is the catchment area of the lake. The results for retention in streams obtained by the MESAW for nutrients may be expressed by the formula
R = 1−
1 1 + 0.000169 * F
(5)
228
A. Vassiljev et al. / Desalination 226 (2008) 222–230
where F is the sub-catchment area. Sub-catchment areas F in formula (5) were in the limits of 10–500 km2. It must be pointed out that the use of simple statistical models hides many aspects of system complexity that might usefully be discovered using a more sophisticated approach [6]. In particular, calculations using the MESAW model showed also that export coefficients are 1.5–2 times higher for phosphorus if clay soils prevail in the watershed or for watersheds with hilly landscape (e.g. sub-basins located in southeast Estonia). Export coefficients for nitrogen were 1.5–2 times lower for such watersheds. But detailed description of dependencies on topography and type of soil needs much more measurements for statistical analysis. Usually export coefficients for phosphorus increase with increasing sub-basin surface slope, especially in winter when the soil is frozen [23]. Grizzetti et al. [24] indicate the increase of nitrogen export with slope but the calculations showed inverse dependence. It can be said that the dependencies of nutrients export on topography exist but the quantification of these dependences is practically impossible by statistical methods due to lack of information. The same situation is valid for dependence on types of soil. It is clear that clay soils increase losses of phosphorus and decrease losses of nitrogen but obtaining of dependence in numerical expression needs much more data. More sophisticated models could help to solve this problem (e.g. SWAT [25] or AVGWL [26]) if more data will be available in future. However, the approach described in this paper can be used in the current situation. The export coefficients obtained have been used for evaluating the likelihood of failing to meet the WFD objective for water bodies. Calculations showed that diffuse sources (arable land and livestock) are responsible for the pollution of about 40% of the water bodies, which are at risk of failing to meet the good status requirements by the year 2015.
For some waterbodies, for which our calculations reveal poor status, water quality was studied in 2006. The investigation showed that the real situation in controlled rivers is very close to the calculated. 4. Conclusions • The scheme proposed in this paper, which was elaborated in compliance with the Guidance [3], for the evaluation of the influence of diffuse and point sources on river status may be successfully used for creating of the preliminary list of unmonitored waterbodies at risk. This list will allow us to select the top priority waterbodies for further investigations. • This scheme includes different approaches for the estimation of the impact of point and diffuse sources on the hydrochemical status of water bodies. Our results showed that simple methods (the mixing model for point sources and export coefficients for diffuse sources) could be used successfully for the implementation of the WFD. • The analysis of the 10-year data of 25 Estonian rivers showed that export coefficients vary from year to year depending on water flow. The equations for the calculation of export coefficients for Estonian rivers for different types of land-use and different hydrological conditions are proposed in the paper. • Our calculations reveal that the impact of diffuse sources on the status of rivers is much higher than the impact of point sources for nitrogen. Both the diffuse and point sources are significant in terms of phosphorus. Acknowledgements This work was financially supported by Estonian Ministry of Education and Research (Grant No. 0142514s03) and by the Estonian Science Foundation (Grant No. 6199).
A. Vassiljev et al. / Desalination 226 (2008) 222–230
References [1] European Union (2000): Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Communities in the field of water policy, Official Journal of the European Communities, L 327/1, 22.12.2000. [2] WFD CIS Guidance Document No. 7, Monitoring under the Water Framework Directive, Published by the Directorate General Environment of the European Commission, Brussels, Jan 2003. http:// www.europa.eu.int/comm/environment/water/waterframework/guidance_documents.html [3] WFD CIS Guidance Document No. 3, Analysis of Pressures and Impacts, Published by the Directorate General Environment of the European Commission, Brussels, Dec 2002. http://www.europa. eu.int/comm/environment/water/water-framework/ guidance_documents.html [4] T.H. Robinson, A. Leydecker, A.A. Keller and J.M. Melack, Steps towards modeling nutrient export in coastal Californian streams with a Mediterranean climate, Agric. Water Manage., 77 (2005) 144–158. [5] P.J. Johnes and T.P. Burt, Nitrate in surface waters, in: T.P. Burt (Ed.), Nitrate: Processes, Patterns and Management, Wiley, Chichester, 1993, pp. 269–317. [6] T.P. Burt, Integrated management of sensitive catchment systems, Catena, 42 (2001) 275–290. [7] R.V. Smith, C. Jordan and J.A. Annett, A phosphorus budget for Northern Ireland: inputs to inland and coastal waters, J. Hydrol., 304 (2005) 193–202. [8] B. Arheimer and M. Brandt, Watershed modelling of nonpoint nitrogen losses from arable land to the Swedish coast in 1985 and 1994, Ecol. Eng., 14 (2000) 389–404. [9] B. Baginska, T. Pritchard and M. Krogh, Roles of land-use resolution and unit-area load rates in assessment of diffuse nutrient emissions, J. Environ. Manage., 69 (2003) 39–46. [10] B.E. Haggard, P.A. Moore Jr., I. Chaubey and E.H. Stanley, Nitrogen and phosphorus concentrations and export from an Ozark Plateau catchment in the United States, Biosyst. Eng., 86 (1) (2003) 75–85. [11] N.M. Pieterse, W. Bleuten and S.E. Jørgensen, Contribution of point sources and diffuse sources
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22] [23]
[24]
229
to nitrogen and phosphorus loads in lowland river tributaries, J. Hydrol., 271 (2003) 213–225. J.D. Wickham and G. Timothy, Wade watershed level risk assessment of nitrogen and phosphorus export, Comput. Electron. Agric., 37 (2002) 15–24. Regulation of the Minister of the Environment, No. 33, Classes of surface water bodies, values of quality indicators corresponding to water classes and procedure of designating of water classes (in Estonian), 22 June 2001. https://www.riigiteata ja.ee/ert/act.jsp?id=85753 W.A. House and F.H. Denison, Nutrient dynamics in a lowland stream impacted by sewage effluent – Great Ouse, England., Sci. Total Environ., 205 (1) (1997) 25–49. C.P. Mainstone and W. Parr, Phosphorus in rivers – ecology and management, Sci. Total Environ., 282–283 (2002) 25–47. K. Irvine, I. Donohue, A. Wemaere, D. Styles and W. Hobbs, Connecting land-use with water quality: sources, sinks and time-bombs, IWA Diffuse Pollution Conference Proceedings, Dublin, 2003. A. Vassiljev and P. Stålnacke, Statistical modelling of riverine nutrient sources and retention in the Lake Peipsi drainage basin, Water Sci. Technol., 51 (3–4) (2005) 309–317. P.M. Haygarth and S.C. Jarvis, Soil derived phosphorus in surface runoff from grazed grassland lysimeters, War. Res., 31 (1) (1997) 140–148. B. Grizzetti, F. Bouraoui, G. de Marsily and G. Bidoglio, A statistical method for source apportionment of riverine nitrogen loads, J. Hydrol., 304 (2005) 302–315. A. Grimvall and P. Stålnacke, Statistical methods for source apportionment of riverine loads of pollutants, Environmetrics, 7 (1996) 201–213. P.J. Johnes, Evaluation and management of the impact of land-use change on the nitrogen and phosphorus load delivered to surface waters: the export coefficient modelling approach, J. Hydrol., 183 (1996) 323–349. A. Maastik, Water protection in agriculture (in Estonian), Tallinn, Valgus, 1984. A. Vasilyev, Modeling wash-off and leaching of pollutants by spring-time flow, J. Hydrol., 159 (1994) 215–222. B. Grizzetti, F. Bouraoui and G. De Marsily, Modelling nitrogen pressure in river basins: a comparison between a statistical approach and the
230
A. Vassiljev et al. / Desalination 226 (2008) 222–230
physically-based SWAT model, Phys. Chem. Earth, 30 (2005) 508–517. [25] J.G. Arnold, R. Srinivasan, R.S. Muttiah and J.R. Williams, Large area hydrologic modeling and assessment; part I, model development, J. Am. Water Res. Assoc., 34 (1) (1998) 73–89.
[26] B.M. Evans, D.W. Lehning, K.J. Corradini, G.W. Petersen, E. Nizeyimana, J.M. Hamlett, P.D. Robillard and R.L. Day, A comprehensive gis-based modeling approach for predicting nutrient loads in watersheds, J. Spatial Hydrol., 2 (2) (2002) 1–18.