S4: Optimization in Environmental Modeling

Organised by Stefan Reis (Centre for Ecology and Hydrology (CEH) Edinburgh, UK)


Environmental models have become more and more complex over time, both with the increase of computing power and storage making it technically feasible and with a growing understanding of processes and connections between different phenomena. One example of this is the way how the full impact pathway of air pollutants from emission source to receptor is nowadays modeled in a high temporal, spatial and substance resolution. Furthermore, in recent years, the linkage of air quality and climate change via emissions of precursors for instance of tropospheric ozone or substances with a radiative forcing or cooling effect, which have, in addition, adverse health effects was further investigated and is today cast into integrated models. But the connection is not limited to the effects, as means to reduce air pollutant emissions and greenhouse gases often generate significant synergies, in some cases however are subject to clear trade-offs.

In many cases, models are designed for decision support, aiding policy makers and experts to derive solutions for a complex problem, where often the objective is multi-dimensional and involves many criteria for the assessment of an 'optimal' solution. Methods such as Multi-criteria assessment (MCA) or Integrated Assessment (IA) are key building blocks of such models. For most of these problems, linear optimization methods, which were often used in reduced-form models, create problems rather than solutions, as it is utterly difficult to cast complex, non-linear problems into a set of restrictions and constraints. For instance, the combined optimization of air pollution control and greenhouse gas reduction strategies mentioned above has a solution space of over 104000 with non-linear relationships between cause and effect (i.e. 'emission' => 'concentration'). Hence, an imminent need for non-linear optimization methods and extremely efficient and fast algorithms emerges in the field of environmental modeling.


This session aims to be a forum for a set of presentations focusing on methodological development and application of optimization methods in general, and the use of Genetic Algorithms and non-linear optimization methods in particular, in environmental modeling.




Part 1, Monday 13:30 - 15:15

13:30. An Efficient Algorithmic Framework for Environmental Modeling. Prof Urmila Diwekar, Vishwamitra Research Institute

13:45. Nesting genetic algorithms to solve a robust optimal experimental design problem. Mr Dirk De Pauw, BIOMATH, Ghent University

14:00. Multiobjective Optimization Procedure For Control Strategies In Environmental Processes. Mr Xavier Flores Alsina, University of Girona

14:15. Optimization of Grazing Management for Watershed Sediment Control Mr Yanxin Duan, USDA-ARS-SWRC

14:30. Prescriptive Treatment Optimization Using a Genetic Algorithm: A Tool for Forest Management. Mr Frederick Maier, University of Georgia

14:45. Landscape design and agricultural land-use allocation using Pareto-based multi-objective Differential Evolution. Dr J.C.J. Groot, Wageningen University & Research Centre

14:45. Solid Waste Management in Urban areas: a multiobjective approach Ms Michela Robba, University of Genova

Could not attend:

Experiences in using evolutionary and non-evolutionary optimization methods in models calibration. Dr Dimitri Solomatine, UNESCO-IHE Institute for Water Education