Speaker: Steve Railsback
Abstract:
My colleagues and I have produced eight major versions of “inSTREAM”, an individual-based stream trout model, since 1999. The model is complex, representing how population ecology emerges from habitat conditions that vary over space and time, the behavior and physiology of individual fish, and competition among the fish for food and habitat. Yet this model has been successful, at least in the sense that we still adapt and apply it to a wide range of basic and applied research questions. Lessons learned about multi-scale modeling from this experience include: (1) Keeping models simple but not too simple is a fundamental challenge. A well-specified problem is critical, and “pattern-oriented modeling” provides a successful strategy for designing models that capture just the essential mechanisms of the real system. (2) The second phase of pattern-oriented modeling is a way to develop useful theory for adaptive behaviors of agents, by testing alternative hypotheses. Multiple qualitative patterns can be more useful than extensive data sets. (3) Validation is not as straightforward a process as for traditional models. (4) Multi-scale models that capture the essential mechanisms of the system they represent can be used, even without extensive calibration and validation, to develop important new understanding.
About the speaker:
Steve Railsback is an environmental engineer and ecologist who specializes in the use of individual-based (or ‘agent-based’) models to address ecological management problems. His experience includes an MS in environmental engineering (University of Illinois, 1981), five years as a staff scientist at Oak Ridge National Laboratory, a PhD in biology (University of Bergen, Norway, 2004), over 20 years experience as a consulting ecological modeler, and supervising students in the Environmental Modeling Graduate Program at Humboldt State University.
Steve’s interest in river management led to involvement in some of the earliest applications of individual-based models in applied ecology. This experience resulted in concern about general issues of across-level simulation modeling. With Volker Grimm, Steve authored the first full monograph on individual-based modeling in ecology (2005) and the first textbook on individual- and agent-based modeling (2012). He is an author of more than 40 publications on specific model applications and modeling methods.