|
The field of ecology is becoming increasingly aware of the importance of accurately accounting for multiple
sources of uncertainty when modeling ecological phenomena and making forecasts. This development is
motivated in part by the desire to provide an accurate picture of the state of knowledge of ecosystems and
to be able to assess better the quality of predictions of local and global change. However, accounting for
various sources of uncertainty is by no means a simple task. Ecological data are almost always observed
incompletely with large and unknown amounts of measurement error or data uncertainty, and often the
expense of data collection prohibits collecting as much data as might be desirable. In addition, most ecological
phenomena of interest can only be studied by combining various sources of data; aligning these data
properly presents interesting statistical challenges. While data plays a large role in most ecological analyses,
incorporating scientific knowledge into the analyses through substantive modeling of ecological processes is
essential. Often such theoretical contributions are based on competing scientific theories and simplifications
of reality. This results in an additional source of uncertainty termed model or process uncertainty. Finally,
substantive models must acknowledge parameter uncertainty. For example, more realistic descriptions of
ecosystems might allow parameters to vary over space and time.
The aim of this workshop is to present a thorough investigation and discussion of these various sources of
uncertainty that typically play a role in ecological analyses and of the statistical techniques that enable
proper inferences and predictions to be made in light of these uncertainties. These concepts will be illustrated
using new data sources and sophisticated modeling tools developed for studying a diverse collection
of ecological phenomena. In addition, there will be a discussion of strategies for reducing some of the sources
of uncertainty, including improved design of monitoring networks. This discussion will promote increased
communication between the theoretical and empirical communities as to prioritizing data-collection efforts.
One of the largest communities to use these methods for important decision-making is state and federal
governments, and they will be involved in the workshop as well. In summary, this workshop will provide
an opportunity for the ecological science community to interact with the statistical and abstract-modeling
communities and will promote novel, interdisciplinary research developments on complex models, inference,
and design in the face of various sources of uncertainty.
The organizers of the workshop are Kate Calder (Ohio State Univ.), Jim Clark
(Duke Univ.), Noel Cressie (Ohio State Univ.), Jay Ver Hoef (National
Marine Mammal Laboratory), and Chris Wikle (Univ. of Missouri).
Details on the workshop can be found on the MBI Workshops website.
|
