Agent Based Modeling Applications for Geosciences

Agent-based modeling techniques have successfully been applied to systems in which complex behaviors or outcomes arise from varied interactions between individuals in the system. Each individual interacts with its environment, as well as with other individuals, by following a set of relatively simple rules. Traditionally this "bottom-up" modeling approach has been applied to problems in the fields of economics and sociology, but more recently has been introduced to various disciplines in the geosciences. This technique can help explain the origin of complex processes from a relatively simple set of rules, incorporate large and detailed datasets when they exist, and simulate the effects of extreme events on system-wide behavior. Some of the challenges associated with this modeling method include: significant computational requirements in order to keep track of thousands to millions of agents, methods and strategies of model validation are lacking, as is a formal methodology for evaluating model uncertainty. Challenges specific to the geosciences, include how to define agents that control water, contaminant fluxes, climate forcing and other physical processes and how to link these "geo-agents" into larger agent-based simulations that include social systems such as demographics economics and regulations. Effective management of limited natural resources (such as water, hydrocarbons, or land) requires an understanding of what factors influence the demand for these resources on a regional and temporal scale. Agent-based models can be used to simulate this demand across a variety of sectors under a range of conditions and determine effective and robust management policies and monitoring strategies. The recent focus on the role of biological processes in the geosciences is another example of an area that could benefit from agent-based applications. A typical approach to modeling the effect of biological processes in geologic media has been to represent these processes in a thermodynamic framework as a set of reactions that roll-up the integrated effect that diverse biological communities exert on a geological system. This approach may work well to predict the effect of certain biological communities in specific environments in which experimental data is available. However, it does not further our knowledge of how the geobiological system actually functions on a micro scale. Agent-based techniques may provide a framework to explore the fundamental interactions required to explain the system-wide behavior. This presentation will present a survey of several promising applications of agent-based modeling approaches to problems in the geosciences and describe specific contributions to some of the inherent challenges facing this approach.