Conference Agenda

110R: Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Friday, 26/Apr/2019:
10:30am - 12:00pm

Session Chair: Jonas Schwaab
Session Chair: Sven Lautenbach
Location: MB-205
Main Building, room 205, second floor, east wing, 90 seats
Session Topics:
What are the visions for the planetary land system?

Session Abstract

Understanding the underlying processes of land-use systems is a highly complex task. Yet, land-use system science goes even beyond this task towards designing visionary futures and developing advice for decision-makers on how to reach these futures. To design visionary futures and transformations requires the use of innovative modelling techniques and close cooperation with stakeholders and decision-makers. Multi-objective optimization approaches are such innovative modelling techniques that bare high potential for developing positive visions about the future. Their most desirable properties are that (1) they can be used to deal with conflicting objectives and reveal trade-offs, (2) they breathe a certain air of normativity that allows to implement visionary ideas and (3) they can help us to make the right decisions in very complex systems which will ideally lead towards the most desirable and sustainable futures. Although the use of multi-objective optimization approaches has increased during recent years, there are still some major challenges ahead before they may become an established tool to support visioning and decision-making. Challenges that need to be addressed are (1) designing optimization algorithms that can efficiently solve complex land-use system problems and depict optimal solutions, (2) show how the analysis of optimal solutions can be used to derive useful recommendations or even rules for decision-makers and (3) how multi-objective optimization can be used in real-world decision-making requiring close cooperation and interaction with stakeholders and decision-makers. A wide range of multi-objective optimization approaches that deal with one of the outlined challenges are invited to be presented and to facilitate a discussion on future research directions of optimization approaches in land-use system sciences.

Full talk
ID: 574 / 110R: 1
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: landscape, land system, optimization, allocation, policy

Pareto frontiers for exploring land use policies

Elizabeth Law1,2, Leandro Macchi1,3, Tobias Kuemmerle1,4

1Conservation Biogeography Lab, Geography Department, Humboldt-University Berlin, Unter den Linden 6, 10099 Berlin, Germany; 2Norwegian Institute for Natural Sciences (NINA), Høgskoleringen 9, 7034 Trondheim, Norway; 3Instituto de Ecología Regional, CONICET Tucumán, Argentina; 4Integrative Research Institute on Transformations of Human Environment Systems (IRI THESYs), Humboldt-University Berlin, Unter den Linden 6, 10099 Berlin, Germany

Understanding how different land uses impact trade-offs between production and the environment, and how these trade-offs play out in space, are critical to transitioning to sustainable land systems. Pareto frontiers offer a powerful policy analysis tool to describe fundamental trade-offs among landscape objectives, as well as determining the efficiency of past, current, and proposed land use allocations, and the impacts of different policy and condition scenarios. Here we discuss the development of landscape-scale Pareto Frontiers describing the major trade-off dimensions between agricultural production, carbon, and biodiversity. We do so for the example of a globally relevant deforestation hotspot, the Dry Chaco in Argentina. We outline the data challenges, simplifications, and assumptions made during the development of the landscape model, and the implementation of the optimization using the R package Prioritizr, solved with Gurobi. Finally, we discuss output options and reflect on the lessons learned during the presentation of the initial results to stakeholders.

Full talk
ID: 376 / 110R: 2
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: Agricultural landscapes, biodiversity, ecosystem services, trade-offs, stakeholder involvement

Using optimization to minimize trade-offs between ecosystem services and biodiversity in agricultural landscapes

Martin Volk, Andrea Kaim, Anna Cord, Michael Strauch

UFZ-Helmholtz Centre for Environmental Research, Germany

Different examples will be presented that use multi-criteria optimization to minimize trade-offs between ecosystem services and biodiversity in various agricultural landscapes across Europe. In a case study in Central Germany, biophysical models were used to simulate agricultural yield, stream flow and water quality, whereas biodiversity, represented by the size of breeding bird habitat, was predicted using statistical (Random Forest) models. These models were applied to stakeholder-defined land use scenarios referring to either land sparing, land sharing or business-as-usual for the year 2030. The scenarios differed in terms of land use and agricultural management as well as in the amount of linear elements in the landscape. Among the scenarios, land sharing was evaluated best for providing bird habitats and water in good quality. However, this came at the cost of a significant decrease in agricultural gross margin. As a next step, the models were coupled with the NSGA-2 optimization algorithm. The algorithm identified a set of Pareto-optimal land use strategies consisting of spatial combinations of the three scenarios. The results show possible improvements in the provisioning of ecosystem services (agricultural yield, water purification) and biodiversity (bird habitat) and can help deriving policy recommendations for future land use allocation. Additionally, challenges for the use of mathematical optimization as a decision support tool will be discussed. This includes, for example, algorithm selection, clustering of objectives, visualization of results and dynamic (i.e. intertemporal) optimization.

Full talk
ID: 572 / 110R: 3
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: Multi-Objective Game Theoretic Models, environmental-economic conflict resolution

The decision-maker matters: An operational Multi-Objective Game Theoretic Model for environmental-economic conflict resolution

Dani Broitman1, Mashor Housh2, Shapira Naama2

1Technion - Israel Institute of Technology, Israel; 2University of Haifa, Israel

Multi-Objective Game Theoretic Models (MOGTM) were developed in order to analyze scenarios of multiple conflicting objectives using the rich toolbox developed in the framework of Game Theory. Typical applications of MOGTM to environmental-economic conflicts, include two players, one advocating the environment preservation, and the other pursuing economic development. Since these models include several objective functions that should be optimized simultaneously, they usually result in a set of optimal responses (called Pareto optimal responses) instead on a single one. MOGTM were used in the last years for the analysis of a plethora of environmental-economic conflicts and have demonstrate their applicability and usefulness. Despite these advantages, there are three issues with MOGTM which prevent their use as a clear-cut tool for decision-making in environmental-economic conflict resolution. The first is their inability to recommend, within a given set of Pareto optimal responses, a single solution which should be implemented. This inability stems from the very definition of multi-objective models, which limits themselves to find the optimal frontier, or at most, restricted subsets within the set of optimal Pareto responses. The second issue is the use of game theoretical abstractions as if these were accurate reflections of the institutional arrangements and relations of power in the real world. In reality, environmental and economic players face much more constraints than possible to introduce in formal models and, above all, they are generally subject to the final decision of an upper instance, not included in the model, which ultimately takes the decision (government, administrative agencies, etc.). The third issue is a direct consequence of the decision-maker’s absence from these models: Its own beliefs, viewpoints and management style cannot be modeled, despite their importance for the conflict outcome. In this paper, we suggest an operationalization of MOGTM which addresses all these concerns. We start from a conceptual inversion of the MOGTM paradigm, focusing first on the decision-maker characteristics and her knowledge of the system she is responsible for. For example, whether she advocates command-and-control or participative policies, what is her room for maneuver to implement these policies, what are the relations of power among the environmental and the economic coalitions in dispute, how the institutional channels available for conflict resolution works, etc. Only after this stage, and based on this background, the decision-maker can choose the appropriate game theoretic tool (which can be a competitive or collaborative game) and its parameters. The solution of the game, likewise in traditional MOGTM, is the last step. The result of this tailor-made game is a single Pareto optimal response which reflects both the decision-maker characteristics, the real-word relations of power between her and the players and among the players themselves. Any change in the perceived or real relations among the players will lead to different game parameters and, hence, a different result. But a modification in the decision-maker characteristics may lead to the implementation of a totally different game, even if the players remain similar. The suggested operationalization of MOGTM results in a single Pareto optimal response, transforming this type of models from a general consultancy tool to a clear-cut tool for decision-making in environmental-economic conflict resolution. Moreover, introducing explicitly the decision-maker preferences in the model, results in more efficient solutions, and allows for a clear explanation about why the chosen solution is better than any other one, subject to the initially defined settings.

Full talk
ID: 853 / 110R: 4
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: Multi-objective optimisation, Settlement network, Socio-economic indicators, Biodiversity

Optimising settlement network topology to maximise socio-economic and biodiversity indicators

Amin Khiali-Miab1, Maarten J. van Strien1, Kay W. Axhausen2, Adrienne Grêt-Regamey1

1Planning of Landscape and Urban Systems, ETH Zurich, Switzerland; 2Institute for Transport Planning and Systems, ETH Zurich, Switzerland

The continuous expansion of settlement areas and the growth of per capita utilization of environmental resources not only increases the pressure on natural habitats and their connectivity, but also affects the human-well-being. Analogous to habitat networks, settlements connected by roads and traffic form complex spatial networks. Polycentricity, which refers to the existence of multiple centres in the structure of a settlement network, is a normative planning goal suggested by many organisations such as the United Nations (UN) and the European Spatial Planning Observation Network (ESPON) and is believed to improve the socio-economic status of a region. In a previous study, we propose that a polycentric settlement networks have a relatively low hierarchy and found that hierarchy indeed was negatively correlated to median personal incomes. Yet it is unclear whether polycentricity also benefits the survival of animal species in habitat networks. The purpose of this study is to use a multi-objective metaheuristic algorithm to find optimal settlement network topologies in which both the survival and mobility of animal species as well as socio-economic indicators are maximised. Our case study area is the densely populated Swiss Plateau. A dynamic model of the settlement network was coupled to a dynamic ecological habitat network model, from which the occurrences of animal species could be predicted. We study the commonalities between the settlement network patterns on the Pareto front and determine whether such optimal settlement network structures can be reached within the framework of polycentric development. We conclude our research with some policy-related suggestions for steering the settlement development process in a direction that satisfies both socio-economic and ecological goals.

Flash talk
ID: 428 / 110R: 5
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: urban renewal, land use trade-off, grey target decision, Xuzhou

Trade-off of land uses in the process of urban renewal: taking xuzhou city as a case study

SHI AN, Shaoliang Zhang, Yunlong Gong, Huping Hou

china university of mining and technology, China, People's Republic of

The urban renewal process is a decision-making process of land redevelopment, and it is also a trade-offs process among land users and stakeholders. Due to the complicated interweaving of renewal costs, social equity, environmentalism, political struggle and other factors, land reuse decision-making in urban renewal is a hot topic for scholars. China's urban renewal is accompanied by rapid economic development and the improvement of urban residents' ecological environment awareness, along with social equity and stability. Therefore, land use trade-offs in China's urban renewal is not only a trade-off between land uses, but also a trade-off between economic, ecological and social benefits. Based on this fact, this paper proposes a trade-off model of urban land uses based on the grey target decision principle, and presents the basic process of trade-off decision-making in the process of urban renewal. Finally, an empirical analysis is carried out on two case studies about demolition and transformation in Xuzhou City, China. The results show that land use decision-making in urban renewal does not have the maximum benefit and optimal utilization structure, but the result of repeated trade-offs between multiple interests.

Flash talk
ID: 898 / 110R: 6
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: biodiversity, human, linear programming, agricultural rents, global

Human-centered systematic global conservation planning

Yuchen Zhang, L. Roman Carrasco

National University of Singapore, Singapore

Systematic conservation planning has helped conservation to be more transparent, repeatable, efficient, and cost-effective. However, they mainly focus on biodiversity and neglect needs of humans. My study aim to advance this field by approximating the problem of decision makers that need to consider biodiversity conservation together with growing human needs. Specifically, there are three objectives: 1) to devise a human-centered systematic conservation-planning model; 2) to apply the model to identify cropland expansion areas that reconcile biodiversity conservation and agricultural production; and 3) to compare the solutions based on five different conservation strategies. I created a human-centred systematic conservation-planning model to plan for the distribution of new cropland by integer linear programming. This model maximized the amount of agricultural rents received by farmers, within constraints of both human crop demand satisfaction and conservation targets achievement. I found suitable locations for cropland allocation were mainly along the tropical belt including the Guiana Shield, the Brazilian Shield, the Western African coast, Eastern African plateaus and valleys, and South East Asia. There were also large areas of new cropland allocated to China, especially in the eastern region. Cotton, maize, oil palm, rice, sugarcane, soybean, and wheat were the most frequently allocated crops. Different conservation strategies presented distinctive economic and environmental outcomes. Strategies that required no allocation into protected areas or key biodiversity areas gave less total rent. National conservation constraint on forest loss could save 12.1 million ha forests, but further decreased rent by $0.52 trillion compared to a global conservation constraint. Although the study is only a small step forward in terms of data resolution and amount of variables considered, the presented framework allows addressing human needs in the conservation planning processes. This could serve as a basis for future conservation planning work in more complex contexts.

Full talk
ID: 323 / 110R: 7
110R Multi-objective optimization approaches to support visioning and decision-making in land-use system science
Keywords: carbon sequestration, land use, scenario analysis, trade-off analysis, optimization

Trade-offs between carbon storage, crop yield production and water supply at the global scale – where to put which land use?

Anita Bayer1, Sven Lautenbach2, Almut Arneth1

1Karlsruhe Institute of Technology, Germany; 2University of Heidelberg, Germany

The way we use our land affects natural ecosystem functioning and ecosystem services (ES). The present pattern of land-use types developed over the last millennia in response to a complex interplay of natural-system constraints and socio-economic pressures, however, the current land use configuration might not provide the optimal in terms of a variety of ES. We evaluate the global configuration of different land uses under the sole premise of optimizing for carbon storage, crop yield production and water supply, three services that are key in the land-use nexus and context of the SDGs. The LPJ-GUESS dynamic vegetation model is ran at 1°x1° grid cell resolution to simulate the ES provision of global land under different allocation of land use (potential natural vegetation and 4 major crop types, each rainfed or irrigated) considering historical and future climate, atmospheric CO2 levels and the distribution of protected areas. A multi-objective genetic algorithm is used to identify Pareto-optimal solutions with respect to the three objectives and constraints in form of protected areas and biophysical growing limits of the different crop types. Our analysis concentrates on the identified optimal land-use allocations providing higher global ES totals than would be achieved under current land use and future climate conditions. Two time horizons are compared, a short-term (20 years into the future) and a long-term planning horizon (end of century). Results indicate a potential for a simultaneous increase of all three services. We highlight opportunities in land management and possible pathways to adapt current land use to increase ES provision towards higher performing allocations and evaluate the effort for transformation. We identify global regions that are crucial for the provision of the three ES and are stable across different optimal solutions and compare them to regions with variant land-use allocation across solutions.