- Program Areas
- Installation Energy and Water
- Environmental Restoration
- Munitions Response
- Resource Conservation and Resiliency
- Natural Resources
- Infrastructure Resiliency
- Air Quality
- Weapons Systems and Platforms
Empirical Dynamics: A New Paradigm for Understanding and Managing Species and Ecosystems in a Non-Stationary Nonlinear World
Dr. George Sugihara | Scripps Institution of Oceanography
This project will investigate and further develop empirical dynamic modeling (EDM) as a practical nonparametric approach for environmental science that explicitly acknowledges the reality of natural (non-engineered), nonlinear (state dependent, non-separable, and ephemerally non-stationary), interconnected systems. It is aimed at addressing the need for credible ecological forecasting in both the short-term and 25 years or more from now, and it will emphasize the ongoing development of methods for short data sets and extensions to non-analogue environmental states as project goals.
The EDM approach uses time series data to reveal the actual dynamic relationships operating among variables. The approach is non-parametric and involves few, if any, assumptions (e.g., it does not assume a stationary equilibrium or make any assumptions about functional form). EDM exploits Takens’ Theorem, which states that any one variable in a dynamic system contains information about all the other causal variables. It uses the extractable information arising from dynamic interdependence among variables (ecosystem complexity). The basic concepts are summarized in three brief videos at http://youtu.be/7ucgQE3SO0o?list=PL-SSmlAMhY3YOxFJ6TZ5S0cLTjv8emcR2.
EDM represents a paradigm shift away from current parametric models (the conventional parametric approach involves “hypothesized” model structures and key variables, not empirically determined ones), and it uses true out-of-sample forecast skill as the rigorous measure of model merit. The approach will be developed specifically to address short data sets and extensions to non-analogue states. It will be applied with non-analogue scenario extensions to two test cases of relevance to the Department of Defense (DoD): predicting red tides in Southern California and predicting coral reef dynamics in the Pacific Islands. The former is relevant to Southern California DoD installations and as a test case for management of freshwater reservoirs by the U.S. Army Corps of Engineers, and the latter is a direct investigation on islands with current (Wake, Guam) and historical (Midway, Johnston, Palmyra) DoD presence (i.e., military installations). Baselines will be established for benthic community dynamics in Pacific Island reefs (neighboring islands without a DoD history), and these control baselines will be used to address actions required to evaluate and restore potential environmental impacts on islands with DoD history.
An important national need is emerging for better analytical tools and fundamental theory that realistically addresses natural, non-engineered dynamic systems. This project will serve the need for such tools—tools that better match the non-ideal reality of actual complex nonlinear interconnected systems. It will directly address various strategic global needs for improved environmental forecasting (e.g., in response to ocean acidification or global warming) and will lay the groundwork for next-generation risk assessment methods, based on skillful forecasts that reduce risk in a nonlinear interconnected world. The two test cases are expected to provide the first predictive understanding of red tide occurrence in the California Current Ecosystem and a scenario-predictive understanding of the impacts of military presence on coral reefs in tropical Pacific islands. (Anticipated Project Completion - 2020)