described in this paper is only valid provided that the in-situ mixture toxicity is non-specific and additive if present.
1 Introduction
It is important to perform systematic monitoring studies. The Minis-try of Environment in Denmark has so far used the following defini-tion of monitoring activities: A systematic and repeatedly collecdefini-tion of data, analysis and assessment over time of a given set of informa-tion based on a pre-designed survey. The objective is to be able to establish time trends, areas of concern and possible cause-relations.
According to the terminology used within the Water Framework Di-rective, this definition corresponds to operational monitoring. Such monitoring activities are rather costly and thus limited compared with the large number of chemicals. It is therefore very important to combine monitoring activities with data interpretation in order to gain maximum knowledge from the existing data and to set up plans for future activities.
Two goals needs to be approached, i.e., 1) the monitoring data needs to be informative and 2) the resources spent on the monitoring activ-ity need to be minimised.
1.1 Informative monitoring
The occurrence of chemicals within the natural environment is a function of
1)emission patterns and sources,
2)the inherent environmental parameters characteristic for different catchment areas and
3)the physico-chemical properties of the individual chemicals.
However, such interpretation will typically be associated with a rather large degree of uncertainty as the transport and process pa-rameters are difficult to quantify under full-scale and realistic condi-tions. Thus, robust methods are needed for data handling, which can focus on the most general information about the fate and occurrence of the chemicals. Methods based on process-oriented compartment/
transport models or on metric statistics will often be difficult to apply due to the inherent assumptions. Therefore, more robust ordinal (ranking) methods may be a good alternative or merely methods to support cause-relations and optimisation of future activities. Conse-quently this investigation will introduce aspects of the partial order theory for data interpretation of monitoring data. The first attempt of this study was to follow the DPSIR-concept, i.e. try to relate different parameters/indicators to
♦ Driving forces (D)
♦ Pressure (P)
♦ States (S)
♦ Impacts (I)
♦ Reactions (R).
Figure 1 shows the elements of the DPSIR-concept of which the Pres-sure-State relation is central for environmental management. Focus of the present study is on the Pressure and State. The State of the envi-ronment may be expressed in many ways, however, this investigation focus on State in relation to pesticide contamination of surface waters.
Reactions
Impacts Driving forces Pressures States
•Agriculture
•Industry
•Energy
•Transport
•Householding
•Emissions
•Land use
•Use of resources
•Transport
•Risky Technologies
•Physical
•Chemical
•Ecological
•Nature and Environment
•Man and Society
Prioritisations Laying down
environmental objectives Environmental
policy Sector specific
policies
Figure 1. The Driving forces, Pressures, States and Impacts relationship which form the basis for environmental management, i.e. Reactions, having focus on the usage of pesticide in the agriculture.
The main Driving Force in relation to surface water contamination by pesticides is agriculture. The Pressure caused by the use of pesticides is quantified by data on 1) emission patterns and sources. In relation to chemical contamination, the State may be directly related to the Pressure data (Thomsen et al., 2003). In addition, State is expected to be a function of 2) the inherent environmental parameters character-istic for different catchment areas and 3) the physico-chemical prop-erties of the individual chemicals.
In the preprocessing of monitoring data it turned out that there was a significant difference in the variance within and between sampling stations. However, for many of the sampling stations there was no significant difference in the mean concentration within and between catchment areas. Detection limits vary between the different samp-ling locations. When the maximum reported detection limits for each specific pesticide are used as a requirement for further data analysis, 204 out of 392 mean values at national scale are rejected. For more detailed information concerning the monitoring data please refer to Thomsen et al. 2003.
Here we present a methodological concept, which is able to identify redundant sampling stations and at the same time is able to cover all pesticides at minimum cost, i.e. sampling locations. We may assume that the objective is to fulfil the generation goal by the year 2020.
Therefore, the highest risk catchment areas, still covering all priority pesticides, should be monitored covering short-term state and the
development of these areas should be controlled every six year within each six-year programme period.
The task is now to select representative sampling stations covering the highest risk areas and all priority pesticides, which is to form the basis for operative monitoring within the next programme period.
1.2 Minimising the demand of resources
One way to optimise the monitoring activities is to identify the re-dundant stations and/or substances in order to be able to implement new compounds within the monitoring programme under the given economic resources. In this way it is possible to perform the best cov-erage of the compounds and data that serve management decision needs regarding control monitoring. In this paper preliminary data analysis will identify the potential of using partial order theory in order to remove redundant stations and substances. Basically, if a substance is registered in a large number of stations in similar con-centration levels resources seem to be lost when the only purpose with the monitoring is to control the possible occurrence of that sub-stance. Because in that case the substance only have to be identified once. Similarly, when one substance is always found in lower con-centration than another substance the most important task is to re-gister the highest concentration level substance. As long as the high-est concentration substance is registered in low concentration values, i.e. control monitoring level, there is no reason to seek for the other lower concentration substance. Given that a fairly simple relation between the usage, e.g. dose, of a given pesticide in a given catch-ment area, and the occurrence in the recipient surface water exists (Sørensen et al. 2003), a tool which takes into account the above con-siderations may form the basis for a dynamic and progressive moni-toring program.
A more detailed derivation of guidelines for designing monitoring activities is outside the scope of this proceeding. The focus of this paper is to disclose redundancy in existing monitoring data using partial order theory.