In general, models are designed to approximate real-world situations and, in this context, form an often simplified or reduced representation of reality, influenced in their design by the questions they have been developed to answer. The purpose for the development of models can for instance be driven by the need to achieve an understanding of ecological processes, or by specific applications, such as the investigation of cause, effect, and mitigation of climatic change. Understanding-driven and application-driven models, however, have different requirements for optimization methods.
As ecological models are often process oriented, representing species, habitats, ecosystems, system interactions and such like, optimization algorithms are mostly applied for curve fitting, that is, trying to improve the agreement between model predictions and observations. On the other hand, most application-driven environmental models are designed for decision support, helping to understand the cause-effect chain, for example, of environmental policy decisions. Their core aims are to increase the efficiency and effectiveness of solutions, finding (the) best solution(s) out of a vast number of possible solutions and combinations. For the first group of problems, generic mathematical optimization methods are typically applied, while the latter require more complex and advanced methods, for example, linear programming or nonlinear approaches.
In this context, it is quite difficult as well to draw a clear system boundary of methods, algorithms, and applications. As ecological and environmental modeling per se spans a variety of scientific disciplines and science areas, algorithms and optimization methods have often been applied in quite different areas before making their appearance here.
Was this article helpful?