Optimized selection of river sampling sites
Introduction
World-wide, the managers of catchment water quality monitoring programs are responsible for considerable expenditures of funds and effort and the selection of river sampling sites ranks highly among their tasks. The optimum selection of sampling sites is related to the objective of the program, whether it is, for example, trend detection, regulatory enforcement, or estimation of pollutant loadings. Sampling programs are often required, however, to fulfil several roles or may have constraints, which make the manager's choice more difficult. Few tools are available to assist managers when designing the spatial distribution of sampling sites.
This paper describes a method which can assist managers with their spatial design. It can be used to provide a justifiable, best choice design for single objective programs; or as will be demonstrated, provide a best compromise design for hybrid monitoring programs and those with constraints.
The method uses a geographical information system (GIS) and graph theory to first derive a mathematical representation of a river network as a matrix. The matrix can contain any cumulative data for locations within the catchment, such as upstream catchment area, pollutant loading or flow volume. An optimization algorithm, known as simulated annealing, is then applied to the matrix to obtain the best sampling site configuration. If constraints are applied to the availability of possible sampling sites, the method provides a mechanism for finding and justifying the best among a limited number of alternatives. In this context, the “best” configuration is taken to mean that which can provide the most information for the least expenditure of effort and resources. The method can be used for any chosen number of sampling sites.
A larger number of sites will always provide more information than a smaller number when both are optimally placed. Therefore, in practice, the number of sampling sites will usually be determined by the budgetary constraints of the program. The method can be adapted, however, to assist in assessing the relative cost-effectiveness of providing more sampling sites.
The placement of sampling sites can be considered on three levels (Sanders et al., 1983):
- 1.
the macrolocation of sites is concerned with the selection of which reach, or link, of a river should hold a sampling station,
- 2.
the microlocation of sites is concerned with where in a selected reach a station should be placed,
- 3.
representative location determines where and how a representative sample may be taken at the chosen site.
This method is concerned only with the macrolocation of sampling sites. It has been assumed that sites will be placed at the downstream end of a reach, immediately before the next confluence.
Section snippets
Previous work
Although there has been much work done on other aspects of sampling program design (Dixon and Chiswell, 1996), this discussion is limited to those which have direct relevance to the method presented in this publication.
Sanders et al. (1983)published a book which has become a standard reference for monitoring program design. They stated that location is probably the most critical design factor and that macrolocation is a function of the specific objectives of the sampling agency. Monitoring
Assumptions
The method assumes that maximum mixing of confluent waters occurs at the end of the immediate downstream reach and there is no backwash upstream into that reach, e.g. from tidal effects. The first case study on the Logan and Albert rivers, presented below, involves a catchment with a tidal section, but this has been ignored to demonstrate how the method may be used with single outlet catchments. Account may be taken of tidal effects by treating the tidal section as a separate hydrological unit.
Summary
This paper has presented a method for the optimization of sampling site placement, which is applicable to a variety of monitoring goals. The method can be adapted to accommodate particular local constraints and produce a sampling plan which is justifiable and defensible on the grounds of monitoring program goals. The method also performed better than the one other published algorithmic method.
Acknowledgements
The authors are grateful to the Queensland Department of Natural Resources, which provided financial and logistical support for this study.
References (11)
- et al.
Review of aquatic monitoring program design
Water Res.
(1996) - et al.
Simulated annealing: an introduction
Statistica Neerlandica
(1989) - Dixon W. (1997) The spatial optimization of water quality monitoring programs by simulated annealing. Ph.D. thesis,...
- Dixon W., Smyth K. G. and Chiswell B. (1996) Topologically optimum monitoring of rivers by approximation algorithms....
- et al.
Regional water quality: evaluation of data for assessing conditions and trends
Environ. Sci. Technol.
(1990)
Cited by (40)
A comprehensive review on the design and optimization of surface water quality monitoring networks
2020, Environmental Modelling and SoftwareCitation Excerpt :River topology-based methods are amongst the earliest WQMN design methods proposed in the literature. The Sanders approach is a typical example (Dixon et al., 1999; Sanders and Adrian, 1978; Sanders et al., 1983). It named after Emeritus Professor Thomas G. Sanders of Colorado State University, who published a book in 1983 that was long a standard reference for monitoring programme design (Sanders et al., 1983).
Efficient method for optimal placing of water quality monitoring stations for an ungauged basin
2014, Journal of Environmental ManagementCitation Excerpt :No global optimum is known for this case. This particular case was studied by Dixon et al. (1999) and Ouyang et al. (2008). Dixon et al. (1999) used the simulated annealing method to determine the solution for minimum-E, whereas Ouyang et al. (2008) used a type of multiple GA for minimum-F criterion.
Spatial Linear Models for Environmental Data
2024, Spatial Linear Models for Environmental DataThe impact of agricultural transformation on water quality in a data-scarce, dryland landscape—a case study in the Bot River, South Africa
2023, Environmental Monitoring and Assessment