Design of a data-driven environmental decision support system and testing of stakeholder data-collection

The aims of this paper are to present the requirements and top level design of a decision support system that facilitates the exchange of environmental information between local level and higher levels of government, as well as to assess the possibility to include the local individual in the decision making process. The design of a tool for data collection and exchange of available data also aims to predict impacts of small-scale locally oriented actions by the local administration and residents on incomes and biodiversity, monitor results of the decisions that follow such prediction and inform central policy assessors to enable appropriate tuning of regulatory and fiscal incentives. The potential of data gathering for use in a DSS was tested by case studies across Europe. The main challenges for implementing effective environmental decision support are now more socio-economic than technical, requiring also a more local-orientated attitude of researchers and government. 2014 Elsevier Ltd. All rights reserved. 1. Decision support systems background and concepts Decision Support Systems (DSSs) are computerized systems which are based on two main pillars. Information Systems Science contributes to the planning and the application of DSSs with the supply of the necessary tools, materials and software, while the Sciences of Operational Research and Management provide the general theoretical frameworks for the analysis of various decisions. Other disciplines are also used to various extents in DSSs, including Systems Science, Artificial Intelligence, Cognitive Science and Psychology (Eom, 2008). Thus, modern DSSs are truly interdisciplinary. Indeed, Alter (2004) correctly states that contemporary DSS has developed into an umbrella term spanning a broad range of systems and functional support capabilities. Arnott and Pervan (2008) analyze in depth the academic field of decision support systems in an exhaustive literature review; Holsapple and Whinston (1996) and Holsapple (2008) provide the basic structure of a DSS, while Manos et al. (2010a) presented a simpler, yet more concise and model-driven description of a DSS architecture. Liu et al. (2010) review the current research efforts with regard to integrated DSS and Power (2001, 2008) identifies five ail.com, [email protected] All rights reserved. generic DSS types, as follows: model-driven DSS, data-driven DSS, knowledge-driven DSS, document-driven DSS and communicationsdriven DSS. According to this scheme, model-driven DSSs emphasize access to (and manipulation of) deterministic, optimization and/or simulation models and use limited amounts of data, which differentiates them from the data-driven DSSs that are capable of utilizing huge datawarehouses. A project to develop the top level design for a Transactional Environmental Support System (TESS, www.tessproject.eu) was funded under the European Commission’s 7th Framework Programme (FP7), as a system to synthesize mainly the first two of these DSS categories, using deterministic, stochastic and simulation models in various risk analysis scenarios that may also require large sets of geo-spatial data. The project ran from 2008 to 2011 (Kenward et al., 2013a). DSSs often attempt to offer solutions in modern managerial environments which are full of redundant and complex information, in which rapidly evolving situations engage a number of individuals in the decision making process e very often on an international level. In these circumstances, DSSs projects have been known to fail (Arnott and Dodson, 2008), even at the stage of requirements analysis and initial development. That is why, especially in ambitious and complex projects like TESS, which need to involve state of the art web technologies (Bhargava et al., 2007; Zahedi et al., 2008), careful planning is an essential prerequisite, especially in order to check the feasibility of the system design and to ensure that the final users will actually use and promote it. J. Papathanasiou, R. Kenward / Environmental Modelling & Software 55 (2014) 92e106 93 Although technology for environmental DSS in both rural (Benson, 1995) and urban (Culshaw et al., 2006) conditions is longestablished, a major lesson from these previous projects was the need to build a system that fits the requirements of users, by working with them throughout the design process. 2. Current status of environmental research in the EU Topics like sustainable development, farm regional planning, climate change, waste management, food supply chain management, environmental protection and biodiversity conservation plus a number of other relevant issues are becoming major focal points of international research. All are interconnected; a major imbalance in one tends to affect the others andmost can benefit from adaptive management. For both these reasons, a critical factor stressed by the Environmental European Agency report (Schutyser and Condé, 2009) is that continually updated datasets are needed. With our entire economy underpinned by ecosystem services, and biodiversity an important component in the ability of the ecosystems to deliver much needed services, how will appropriate datasets be obtained and updated at a regular basis? Who will fulfill the task and with what funds? The Natura 2000 network of protected areas is a cornerstone of nature conservation policy in the European Union, covering many areas that are being enlarged through updates and expansion of EU political borders (Maiorano et al., 2007). In addition, directives for Environmental Impact Assessment (EIA), complemented by Strategic Environmental Assessment (SEA) have been defined and introduced by the EU as a requisite for projects and programmes having a significant effect on the environment. But biodiversity at local level is still declining at alarming rates across Europe (e.g. Thomas et al., 2004), despite measures like the growth in number of nationally designated protected areas in 39 European countries (Schutyser and Condé, 2009). The European target of halting the loss of biodiversity by 2010 has slipped away (Dimas, 2009) and moved a decade ahead, to 2020 (EU COM, 2011). In many circumstances, a regulatory framework is simply not enough (Manou and Papathanasiou, 2009) because a myriad small and locally based land-use decisions outside protected areas summate to change the environment. The resulting habitat degradation and loss is often not immediately perceivable, as Kuussaari et al. (2009) explainwith the notion of ‘extinction debt’. DSS design in TESS was aimed specifically at these small and locally based decisions. The EU has funded much environmental research related to TESS, including a project on Governance and Ecosystem Management for the CONservation of BIOdiversity (www.gemconbio.eu, Manos and Papathanasiou, 2008) that lay foundations for the TESS project. GEMCONBIO brought together 12 partners from Greece, Sweden, UK, Germany, Belgium, Hungary, Romania, Iran, Indonesia, and Bolivia during 2006e2008 to explore the interactions of governance processes and institutions with sustainable development objectives and conservation of biodiversity across more than 30 thematic and geographic case studies. A worrying finding was that where biodiversity diminishes, local people may to lose interest in the natural environment, as shown by fewer people engaging in wildlife-related activities in the most urbanized parts of Europe (Kenward and Sharp, 2008). However, the strongest positive associations with conservation and sustainable use of biodiversity were for knowledge leadership and adaptive management (Kenward et al., 2011), which are quintessential characteristics of a DSS. Other EU-funded projects relevant to data collection for biodiversity policy implementation e and therefore also directly relevant to the TESS e are ALARM, SCALES and EU BON. ALARM (Assessing LArge scale Risks for biodiversity with tested Methods, www.alarmproject.net), aimed inter alia to establish socioeconomic risk indicators related to the drivers of biodiversity pressures as a tool to support long-term mitigation policies. The SCALES project (Securing the Conservation of biodiversity across Administrative Levels and spatial, temporal, and Ecological Scales, www.scales-project.net) has as a general objective to provide the most appropriate assessment tools and policy instruments to foster the capacity for biodiversity conservation across spatial and temporal scales and to disseminate them to awide range of users, while EU BON (European Biodiversity Observation Network, www.eubon. eu) focuses on the delivery of near-real-time relevant data, both from on-ground observation and remote sensing, to the various stakeholders and end users ranging from local to global levels. A relevant COST (European Cooperation in Science and Technology, www.cost.eu) action was also launched in 2011, called HarmBio (Harmonizing global Biodiversity modeling, www.harmbio.eu), aiming to harmonize current biodiversity models and datasets in order to improve the reliability of future projections of biodiversity change (e.g. under various policy options which may be used to assist environmental decision making). The EEA (European Environmental Agency, www.eea.europa.eu) has launched the BISE (Biodiversity Information System for Europe, http://biodiversity. europa.eu) initiative for bringing together biodiversity datasets (albeit without analytic capabilities) and the Eye on Earth system (www.eyeonearth.org) that focuses on GIS data. On a global scale UNEP (United Nations Environment Programme) is working in parallel with the EU initiatives on the Global Environment Outlook, (GEO, www.unep.org/geo) and The Economics of Ecosystems and Biodiversity (www.teebweb.org). 3. Environmental decision making De Marchi et al. (2012) provide a survey of formal methods available to help policy makers improve their decisions, while Moran et al. (2006) have worked in the analysis, implementation and assessment of public policies. Tsoukias et al. (2013) suggest a framework to support the use of analytics in the policy cycle e not only for environmental issues e and conceptualise it as “Policy Analytics”. They correctly identify the need to use tangible and intangible public resources during the decision making process, the engagement of many diverse stakeholders with different and often conflicting interests, and the long time horizon needed for today’s policy cycle. The role of stakeholders can often be complicated but their participation throughout will generally produce better decisions, as they are the ones who will bear the consequences of these same decisions (Voinov and Bousquet, 2010). Laniak et al. (2013) also introduce the concept of Integrated Environmental Modeling and using their own words this is ‘inspired by modern environmental problems, decisions, and policies and enabled by transdisciplinary science and computer capabilities that allow the environment to be considered in a holistic way’. It is in the above context that environmental decision makers need robust DSS tools; indeed, a resent advice paper prepared by the LERU biodiversity working group (League of European Research Universities, De Meester et al., 2010) recommends investing in interoperable databases using adopted standards as well as tools to use these data. Such DSSs combine environmental modeling techniques and IS technology in a fast-developing field; Jakeman et al. (2008), followed by Manos et al. (2010b) and Andreopoulou et al. (2011) all edited books on agricultural and other environmental decision support systems. Recently, McIntosh et al. (2011) identified the key research challenges for the development and adoption of Environmental DSSs and provided some recommendations for addressing them. J. Papathanasiou, R. Kenward / Environmental Modelling & Software 55 (2014) 92e106 94 The development of an environmental DSS which uses high volumes of diverse data frommultiple sources, must be considered in terms of complex systems (Sànchez-Marrè et al., 2008). Outcomes predicted by such DSS are liable to affect disadvantaged parts of society, such as the rural population, and to be influenced by many climatic demographic and socioeconomic variables which are difficult to predict. Addressing a variety of environmental problems requires cooperation between different groups, stakeholders, NGOs, policy makers and the local population. Trade-offs between system simplicity and scientifically concrete results may depend on quality of available data, which in turn depend on interrelated trust, confidentiality and uncertainty. Quality of data from citizen science or other local knowledge can be improved at input, for example by image recognition (Kumar et al., 2012) or during processing, for instance by application of fuzzy logic (Giordano and Liersch, 2012). Such systems require continuous evaluation by the end-user, for which the MIS (Management Information Systems) discipline provides tools, with perceived effectiveness by the enduser coming to focus especially in participatory planning (Inman et al., 2011). 4. The TESS project ambition and scope TESS brought together 14 partners from 10 different European countries. Its aim was the requirements analysis and top level design of a decision support system to facilitate the integration of local knowledge into policy making, while at the same time guiding and encouraging local activities that restore and maintain biodiversity and ecosystem services. The vision was to enlighten, encourage and empower local communities to support biodiversity restoration across Europe, through an internet system that could unify all available knowledge to guide decisions for the benefit of biodiversity and livelihoods (Kenward et al., 2009). Considering that a core democratic maxim is that those affected by a decision should also participate directly in the decision making process (Smith, 1982), TESS sought to include the local individuals in the entire environmental policy cycle, and not only for data gathering activities as in previous participatory approaches (Jankowski, 2009; McCall, 2003; Elwood, 2008; Hessela et al., 2009; McCall and Dunn, 2012). TESS assessed also the possibility of models being applied to the available data for non experts, to improve their immediate decisions outside the higher level policy cycle. In order to do that, TESS first listed and analyzed government information requirements at national and intermediate levels (Sharp et al., 2013), identified local information needs (Hodder et al., 2013) and quantified flows between sources and recipients of knowledge (Perrella et al., 2013). The project then developed a database of environmental models suitable for bio-socio-economic predictions (Ivask et al., 2013), and analyzed for gaps between the current supply of models and the forecasting required. From about 2400models in the database, 198 were suitable for use by scientists at fine scale (field, pond, garden), but only ten of these were suitable for use by untrained stakeholders and only four could be considered easy for them to use (Kenward et al., 2013b). Of the ten locally-usable model-based routines, eight operated in English, one also in French and two in Hungarian. This created substantial deficiencies in documentation as a further factor hindering wide use of models. Beyond the analysis of information supply and demand, TESS also investigated governance effects that might benefit from a comprehensive DSS, and hence motivate its development. A survey of national government and local practices, in 26 of the 27 EU member states plus Norway, Switzerland, Turkey and Ukraine (Ewald et al., 2013), identified factors associated with effective application of formal environmental assessments (EIA þ SEA), together with priority areas for internet-based decision support and local monitoring to benefit livelihoods and biodiversity (Beja et al., 2013). Most of the factors which associated with prevalence of assessments, which in turn associated with low rates of urban sprawl, were attitudes, consultation and participation in activities assessed at local level. As other EU research projects were focusing on recommendations for high-level policy, TESS focused its final stages on a top level system design for DSS delivery at local level, and on case studies to test what local knowledge could be supplied in exchange. We hypothesized that the system should be built to handle spatial data, partly because a Pan-European survey found about half the states to be digitally enabled for GIS in local authorities (Kenward et al., 2013c), with maps also used routinely for landmanagement by a diversity of interests. A crucial consideration is that maps at local level aggregate and scale up to much wider coverage. Studies of participatory GIS that can be applied to public decisions indicate that the proportion of people willing to participate gets smaller as the spatial scale of decision increases from local to the regional and national level (Kingston et al., 2000); this too made it appropriate for TESS to work at local level. Following the case studies presented in the next section, the TESS team organized a number of workshops to assess the lessons learned and produce the top level design of the envisaged system. The higher level requirements of the system are shown in Fig. 1a and b according to the SysML (Systems Modeling Language, http:// www.omgsysml.org/), requirements diagram standards. Each requirement is represented by a box with a unique id and a short text providing a concise description; some boxes include a “refinedBy” section that contains the associations of these requirements with the use cases produced by TESS (Kenward et al., 2013b). The system design is internet based, accessing large external public databases plus smaller ones held privately by individuals, with a configuration approximating Fig. 2. The TESS project revealed that the environmental modeling and database community is largely fragmented, disparate and uncoordinated, with diverse (input/ output) metrics and without any effective demand in the model creation phase for compatibility among models. Therefore, agreement on environmental indicators and model metrics plus evaluation criteria will be needed for harmonizing the models before entering them in themodel base.While models and toolkits remain accessible as separate modules, it will probably be necessary to distinguish models and toolkits according to their complexity, such that non-experts can be encouraged initially to interface with the less complicated ones. All models in such a system must be accompanied with adequate and easily comprehended documentation, with as much as possible for non expert users. A comprehensive system would become very large and require techniques like managed evolution (Murer et al., 2011). However, it could be started bymergingmodels as a series of toolkits in separate sectors, such as farming and forestry (Piirimäe, 2011). TESS differs substantially from other initiatives such as IPBES (Intergovernmental Platform on Biodiversity and Ecosystem Services, http://ipbes.net/). IPBES is an interface between the scientific community and policy makers that aims to build capacity for and strengthen the use of science in policy making at high level, while TESS favors a bottom-up approach; it aims to mobilize local human resources, by producing a system capable of handling huge amounts of diverse information in a coherent and easy way for the local farmer, gardener, hunter, etc. This is a system to motivate conservation of biodiversity and ecosystem services, because it benefits land-managers financially (Ayoo, 2008) or culturally (e.g. for recreation), in effect providing payments for Ecosystem Services (Ferraro and Kiss, 2002) through private spending of money or time. Moreover, be self sustaining in the long run, the system needs J. Papathanasiou, R. Kenward / Environmental Modelling & Software 55 (2014) 92e106 95 to become attractive enough to users to gain subscriptions. In this context it also needs to motivate its users; researchers, administrators, managers or other local residents need to receive the credit for the data they provide and to get other data and feedback in return (Fig. 2). 5. TESS case studies analysis and results In order to discover how hidden local knowledge could be brought to the surface, TESS partners assessed local capabilities and willingness to adopt new technologies. Could credible data from the local level be fed to the regional and national level, both for making policy across Europe and to help formal Environmental Assessments, like EIA and SEA to producemore robust results? Case studies of local communities tested whether volunteers (based on schools, NGOs, local community groups or individuals motivated by «require Operating En id = “TESS 1” text = “The system shall be architecture must be flexibl frontends may be developed id = “TESS 11” text = “The system must be able to accept donations, subscriptions and payments on account for models and data” «requirement» Subscriptions id = “TESS 17” text = “There must be scope documentation, help and tutor «requirement» Help refinedBy «useCase» Help and tut navigation (uc 18) id = “TESS 13” text = “The user must be able to create a user account so the system remembers the user’s details (name, addr subscription and account details) at login; the system maintain a list of accounts in its central database” «requirement» User accounts refinedBy «useCase» User registration (uc 11) «useCase» User login (uc 9) Data and manage «deriveReqt» req [Package] TESS top level requirements a Fig. 1. a and b. SysML requirements diagram standards for a TESS; each requirement is repre some boxes includes the associations with TESS use cases (Kenward et al., 2013b). use of natural resources) could provide effective local monitoring that meets central policy requirements, and what could motivate them to do so. The project partners were asked to choose a local community in their countries at the lowest level of government (LAU2) and then to organize a team of local residents as volunteer ‘helpers’. The project teams offered training, guidance, equipment and collation of the results, but field work was done by the helpers only. The aims of each case study were to test how best to meet a local decision support need in exchange for local monitoring that could meet central policy requirements. Projects typically required mapping of ecological information, for combination with socio-economic information. Helpers therefore worked in each case on a socioeconomic project and a mapping project (except in the German case, which had only a mapping project). However, in order to assess motivations, case studies also assessed other relevant local factors,