Numerical models are a critical tool in many disciplines for generating and testing hypotheses, examining sensitivity to perturbation, hindcasting or filling in data gaps, and making predictions about future behavior. Numerical experiments offer advantages over field or laboratory experiments in that they provide complete control over critical variables and can expand the spatial and temporal scales over which experiments are run. Although improvements in computing technology have allowed for highly detailed simulations, does the ability to see everything (i.e., the perfect experiment) improve our overall understanding of phenomena? An alternative to highly detailed simulations, reduced complexity models (RCMs) have persisted and are even becoming more pervasive in the environmental sciences. RCMs offer advantages in their ability to couple physical, biological, and chemical dynamics and provide highly intuitive yet quantitative understanding of system behavior and sensitivities. However, these advantages depend on the rigor of the processes simulated and the assumptions involved in model simplification. A diverse array of assumptions and strategies used in formulating RCMs is present in ecology, hydrology, climate science, and other environmental science fields, highlighting a need for the community to come together for synthesis. In this workshop, participants from diverse fields will come together to discuss common strategies for reducing complexity in models, probe assumptions, and evaluate issues of scale and parameterization of RCMs. Strategies for testing the rigor of RCM processes and assumptions using direct numerical simulations, physical models, and/or databases will also be developed.
RCMs are common elements of larger climate change simulations and are also used to predict how large ecosystems such as deltas, wetlands, and desert landscapes will likely change as a result of restoration, urbanization, and/or changes in climate. The workshop will result in tools for the scientific community that will make the use of RCMs more efficient, reliable, and effective for use in management of thee complex environmental systems. The tools include online resources (code, model test cases, data, descriptions of common steps in formulating RCMs) that will help streamline and validate RCM development. Two synthesis papers, targeted for general publications of geophysical and ecological societies will also result. Because of their simplicity and their ability to produce intuitive understanding of how complex environmental systems function, RCMs, or ?toy models,? are an ideal teaching tool. Part of the workshop will focus on developing online teaching resources using RCMs. An additional workshop session focused on graduate student research will reinforce an emphasis on training a diverse assemblage of early-career scientists.
This award funded a workshop in March 2013 that brought together a diverse community of biological and physical scientists to address strategies for modeling/simulating environmental systems. Within these two disciplines, scientists have historically approached the challenge of simulation from different directions. Geophysicists have a tradition of working with models that closely adhere to the physical laws governing systems but have been increasingly formulating much simpler models. In contrast, modeling in ecology was initially dominated by very simple models, with later models becoming more highly detailed. The most important outcome of the workshop was synthesis of a systematic approach to modeling environmental systems that recognizes distinct roles for reduced-complexity models and highly detailed models. Reduced-complexity models (a.k.a. "exploratory models") best enable the researcher to test alternative hypotheses about the dominant processes controlling the evolution of landscapes or their ecological function. Because their simple structure enables many different versions of the model to be run in succession, they make it possible to understand the sensitivity of landscapes or their functions to changes in environmental variables. In contrast, more detailed models are most appropriate for making predictions about specific landscapes or understanding the processes controlling more subtle aspects of a landscape’s form or function. They may also be the most appropriate type of model to use in testing the impact of alternative management scenarios. In the Appropriate-Complexity Method (ACME) for formulating models, workshop participants synthesized a systematic strategy for approaching landscape modeling challenges. The goal of ACME is to develop a model with the "appropriate" level of detail, defined as the model that optimizes tradeoffs in tractability, level of representation of the actual environmental system, interpretability, and generality. Decisions about different aspects of "detail" in the model are made based on observable properties of the landscape and a progression of questions that orient modeling objectives. The modeling tool kits developed in the workshop, links to published papers, and other workshop materials are available to the general public on the workshop website, https://sites.google.com/site/rcmworkshop/home. Another broader impact of the workshop is that it contributed to career development and provided professional networking opportunities for graduate students and early-career scientists. Mechanisms through which this occurred were a student and early-career poster session and inclusion of early-career scientists in the panel of invited speakers, workshop leadership, and the group of authors that compiled the synthesis articles resulting from the workshop.
