The use of Hugin® to develop Bayesian networks as an aid to integrated water resource planning

doi:10.1016/j.envsoft.2003.12.021

Environmental Modelling & Software

Volume 20, Issue 2, February 2005, Pages 231-242

https://doi.org/10.1016/j.envsoft.2003.12.021 Get rights and content

Abstract

Integrated management is the key to the sustainable development of Europe's water resources. This means that decisions need to be taken in the light of not only environmental considerations, but also their economic, social, and political impacts; it also requires the active participation of stakeholders in the decision making process. The problem is to find a practical way to achieve these aims. One approach is to use Bayesian networks (Bns): networks allow a range of different factors to be linked together, based on probabilistic dependencies, and at the same time provide a framework within which the contributions of stakeholders can be taken into account. A further strength is that Bns explicitly include the element of uncertainty related to any strategy or decision. The links are based on whatever data are available. This may be an extensive data set, output from a model or, in the absence of data, can be based on expert opinion. Networks are being developed for four catchments in Europe as part of the MERIT project; these are in the UK, Denmark, Italy and Spain. In each case stakeholder groups are contributing to the design of the networks that are used as a focus for the consultation process. As an example, the application to water management of a UK basin is discussed.

Section snippets

Integrated water resource management: essential features

It is now widely recognised that sustainable, equitable and efficient management of water resources, particularly at the catchment scale can best be achieved through an Integrated Water Resource Management (IWRM) approach (Mariño and Simonavic, 2001). The European Union has adopted the concept as an integral part of their Water Framework Directive (WFD), which came into force in October 2000. The concept has been discussed at length in many publications and a number of definitions are

Stakeholder participation

Stakeholder participation is central to an integrated management approach; the opinions of people affected by decisions have to be taken into account. Without this grass roots involvement, the successful implementation of any strategy faces difficulty for the reason that people may perceive it to be irrelevant.

Bns offer one way to introduce the contribution of stakeholders to the decision making process. Networks provide a framework within which the opinions of stakeholders can be fed. It is

Examples of network development

MERIT is developing networks in four countries; the UK, Demark, Italy and Spain. In each case they are being constructed in consultation with relevant stakeholder groups, using a range of consultation techniques. The issues being addressed vary from country to country.

•
UK, Loddon Catchment and Portsmouth area: Two networks are focused on;
- -
  the management of domestic water demand,
- -
  the impact of abstraction from a spring on the bird population of coastal flats.
•
Denmark, Havelse Catchment: a network is

Conclusion

The use of Bayesian networks as a focus for stakeholder consultation and as a tool to aid decision makers offers a number of advantages.

•
The way in which the network can be built up with the involvement of stakeholder groups and the graphical nature of the presentation makes it a useful tool for decision makers keen to encourage transparency and stakeholder participation in the decision making process. Encouraging stakeholders to contribute to the design and development of the network fosters a

Acknowledgements

We would like to thank the European Union for their financial contribution to the MERIT project and in particular Dr. Panos Balabanis who has provided much encouragement and support. Thanks also to our fellow participants from Birmingham, Aalborg, La Mancha and Milan Universities and from GEUS, CESI and the Regione Abruzzo. Without their invaluable contributions this paper would not have been possible. Finally special thanks to all those stakeholders, who have tolerated our consultations with

References (20)

A. Dixon et al.
Water saving potential of domestic water reuse systems using greywater and rainwater in combination
Water Science Technology
(1999)
P.S. Mikkelsen et al.
Collected rainfall as a water source in Danish households—what is the potential and what are the costs?
Water Science Technology
(1999)
D. Ames et al.
A Bayesian decision network engine for internet-based stakeholder decision-making
M. Borsuk et al.
Stakeholder values and scientific modeling in the Neuse River watershed
Group Decision and Negotiation
(2001)
E. Charniak
Bayesian networks without tears
Artificial Intelligence Magazine
(1991)
European Union, 2002. Guidance on public participation in relation to the Water Framework Directive. Active...
A.J. Gray et al.
Evaluating Methods for Public Participation. R&D Technical Summary E2-030/TS. ISBN No 1 85705 840 2
(2002)
D. Heckerman et al.
Real world applications of Bayesian networks
Communications of the ACM
(1995)
R.A. Howard et al.
Influence diagrams
F.V. Jensen
An Introduction to Bayesian Networks
(1996)

There are more references available in the full text version of this article.

Cited by (212)

Bayesian networks supporting management practices: A multifaceted perspective based on the literature
2024, International Journal of Information Management Data Insights
Bayesian network is a probabilistic graphical model within machine learning that supports decision-making under conditions of uncertainty in different domains. Although the scientific literature has increasingly addressed the implementation of Bayesian networks to support management practices (BNM), a thorough review is currently lacking. This bibliometric review investigates the transformative potential of Bayesian networks in reshaping decision-making paradigms across multidisciplinary domains. The knowledge set findings reveal a predominant focus on risk management within the Engineering domain; the scientific openings involve significant progress in both theoretical frameworks and practical applications across Computer Science, Engineering, Medicine, and Environmental Science; and research trends indicate a progressive BNM within Engineering and Medicine, contrasting with a decline in innovative studies related to Computer Science. This study acts as a catalyst, propelling inventive BNM applications and fostering interdisciplinary advancements. It lays a foundation for pioneering BNM strategies.
Risk assessment of inter-basin water transfer plans through integration of Fault Tree Analysis and Bayesian Network modelling approaches
2024, Journal of Environmental Management
Inter-basin water transfer projects are a common method used to balance water resources and meet regional demand, particularly in the drinking water supply sector. The potential failure risk associated with inter-basin water transfer projects was examined using Fault Tree Analysis (FTA) and Fuzzy Fault Tree Analysis (FFTA) methodologies in this study. Additionally, the conversion of Fault Tree models into Bayesian Network (BN) and Fuzzy Bayesian Network (FBN) models was explored. Ten basic events were identified as factors that could affect the success of inter-basin water transfer plans, including socio-political, environmental, water resource, economic, and technical criteria. Fault Tree and Fuzzy Fault Tree models were utilized to conduct a risk analysis, which was then converted into crisp and fuzzy FTA-BN through an integrated approach. This approach was applied to evaluate inter-basin water transfer scenarios from the Great Karun basin to the Central Iran Plateau. The superior scenario among eight water transfer scenarios was found to be water transfer from the Behesht-Abad Basin to Isfahan province and from the Khersan Basin to Kerman and Yazd provinces, with a failure risk of 0.649 and 0.601 respectively, based on the crisp and fuzzy integrated models. Basic events were ranked based on their contribution to the occurrence of the top event using two FIM and BI indices in the Fault Tree model and two indices of MI and SI in the Bayesian Network. Furthermore, after considering the correlation between basic events and risk factors, the risk obtained by crisp and fuzzy integrated models was found to increase to 0.811 and 0.789 respectively. The results of this study demonstrate that an integrated approach can assist decision-makers and stakeholders in evaluating inter-basin water transfer projects.
A Bayesian network-GIS probabilistic model for addressing human disturbance risk to ecological conservation redline areas
2023, Journal of Environmental Management
Population growth and associated ecological space occupation are posing great risks to regional ecological security and social stability. In China, “Ecological Conservation Redline” (ECR) that prohibited urbanization and industrial construction has been proposed as a national policy to resolve spatial mismatches and management contradictions. However, unfriendly human disturbance activities (e.g., cultivation, mining, and infrastructure construction) still exist within the ECR, posing a great threat to ecological stability and safety. In this article, a Bayesian network (BN)-GIS probabilistic model is proposed to spatially and quantitatively address the human disturbance risk to the ECR at the regional scale. The Bayesian models integrate multiple human activities, ecological receptors of the ECR, and their exposure relationships for calculating the human disturbance risk. The case learning method geographic information systems (GIS) is then introduced to train BN models based on the spatial attribute of variables to evaluate the spatial distribution and correlation of risks. This approach was applied to the human disturbance risk assessment for the ECR that was delineated in 2018 in Jiangsu Province, China. The results indicated that most of the ECRs were at a low or medium human disturbance risk level, while some drinking water sources and forest parks in Lianyungang City possessed the highest risk. The sensitivity analysis result showed the ECR vulnerability, especially for cropland, that contributed most to the human disturbance risk. This spatially probabilistic method can not only enhance model's prediction precision, but also help decision-makers to determine how to establish priorities for policy design and conservation interventions. Overall, it presents a foundation for later ECR adjustments as well as for human disturbance risk supervision and management at the regional scale.
Participatory Bayesian Network modeling of climate change risks and adaptation regarding water supply: Integration of multi-model ensemble hazard estimates and local expert knowledge
2023, Environmental Modelling and Software
Local climate change risk assessments (LCCRAs) are best supported by a quantitative integration of physical hazards, exposures and vulnerabilities that includes the characterization of uncertainties. We propose to use Bayesian Networks (BNs) for this task and show how to integrate freely-available output of multiple global hydrological models (GHMs) into BNs, in order to probabilistically assess risks for water supply. Projected relative changes in hydrological variables computed by three GHMs driven by the output of four global climate models were processed using MATLAB, taking into account local information on water availability and use. A roadmap to set up BNs and apply probability distributions of risk levels under historic and future climate and water use was co-developed with experts from the Maghreb (Tunisia, Algeria, Morocco) who positively evaluated the BN application for LCCRAs. We conclude that the presented approach is suitable for application in the many LCCRAs necessary for successful adaptation to climate change world-wide.
Overloading risk assessment of water environment-water resources carrying capacity based on a novel Bayesian methodology
2023, Journal of Hydrology
Water environment-water resources carrying capacity (WERCC) is a sustainability indicator, rendering it significant to measure the water systems covering the water environment and water resources. Currently, water environment carrying capacity (WECC) accounting and assessment mostly tend to the deterministic results of a single element under a design hydrological condition. However, the WERCC, as well as its components WECC and water resources carrying capacity (WRCC), is a random variable with a certain probability distribution. To account for that, we established a Bayesian network-based WERCC overloading risk assessment method system. It can calculate the water system carrying state and its corresponding probability after determining the WECC and WRCC probability distributions. Among them, the probability distributions of the WECC and WRCC were determined by the Bayesian formula, and the spatiotemporal law, respectively. Additionally, based on the above quantitative uncertainty analysis in the basin, a Bayesian network-based WERCC overloading risk assessment path was built by modeling the causal chain of the WERCC internal impact mechanism, and the risk assessment results were obtained by the software Netica. Taking the North Canal Basin in China as an example, the results of WERCC overloading risk zoning showed that CU6 and CU9 were severe risk areas; CU2, CU7, CU8, and CU10 had moderate risks; and the remaining 40% control units were slight or mild risks. According to the overloading risk characteristics, the uncertain quantitative assessment has provided a scientific basis for ensuring the sustainable development of the North Canal from the perspective of reducing the water environment pressure (WEP) and improving the WERCC.
A causal structure-based multiple-criteria decision framework for evaluating the water-related ecosystem service tradeoffs in a desert oasis region
2022, Journal of Hydrology: Regional Studies
Qira oasis, a typical catchment alluvial fan composed by agriculture and natural shrubs at the south margin of Taklimakan Desert, China
Water management modeling involves causal interaction effects of various water-related ecosystem services (ESs) under multiple-criteria water management alternatives, measuring expected benefits of water-related ESs. Many multi-criteria decision tools have been developed to mitigate some of major challenges of ESs tradeoffs. However, few formal ES trade-off frameworks have focused on the multiple-criteria alternatives assessment and indifference point identifications of water-related ES. This paper proposes a causal structure-based multiple-criteria decision framework to model water-related ESs, and to detect the points where the decision makers would be indifferent between two alternatives and to compare with the optimum recommendation value using Bayesian networks (BNs) with analytic hierarchy process (AHP).
The study confirms that the proposed causal structure-based multiple-criteria decision framework is a promising approach to modeling possible climate, irrigation, and water policy scenarios and to examining influences of those scenarios on water-related ESs. The framework can be used to effectively recommend optimum water management alternatives and to identify the indifference points by combining BNs into AHP under stakeholder participation. The framework also provides a qualitative and quantitative assessment to reduce the conflicts and uncertainties of multiple-criteria weights in diagnosing water-related ES trade-offs, owing to different stakeholder preferences.

View all citing articles on Scopus

^☆: This work is being carried out as part of MERIT (Management of the Environment and Resources using Integrated Techniques), an EU Framework V funded project No. EVK1-CT-2000-00085. Web site: http://www.merit-eu.net.

View full text

The use of Hugin® to develop Bayesian networks as an aid to integrated water resource planning☆

Abstract

Section snippets

Integrated water resource management: essential features

Stakeholder participation

Examples of network development

Conclusion

Acknowledgements

Water Science Technology

Water Science Technology

A Bayesian decision network engine for internet-based stakeholder decision-making

Stakeholder values and scientific modeling in the Neuse River watershed

Group Decision and Negotiation

Bayesian networks without tears

Artificial Intelligence Magazine

Evaluating Methods for Public Participation. R&D Technical Summary E2-030/TS. ISBN No 1 85705 840 2

Real world applications of Bayesian networks

Communications of the ACM

Influence diagrams

An Introduction to Bayesian Networks

The use of Hugin^® to develop Bayesian networks as an aid to integrated water resource planning☆