Spring 2008 Seminar
Model-Data Fusion in Ecology

February 1

Tom Hobbs, NREL, Colorado State Univeristy
Title: "Training Graduate Students to Assimilate Data with Ecological Models"\

February 8

Tom Stohlgren, NREL, Colorado State Univeristy
Title: "Species Environmental Matching (SEM) Models for Invasive Species "

February 15

Nate McDowell, Earth and Environmental Sciences Division, Los Alamos National LabTitle: "Models of plant physiology and ecosystem processes as hypotheses"

February 22

Don Estep, PRIMES, Dept of Mathematics, Colorado State Univeristy
Title: "Sensitivity analysis of ecological models"

February 29

Becky McKeown, NREL, Colorado State Univeristy
Title: "The effect of numerical error on forward and inverse model solutions"

March 7

Steve Ogle & Dan Cooney, NREL, Colorado State Univeristy
Title: "An approach to resolving CO2 fluxes in the mid-continent region of the US by combining top-down atmospheric-based inversions with bottom-up inventories in a state-space model application"

March 14

Catherine S Jarnevich, USGS
Title: "Forecasting species distributions under changing climates"

March 21

* Spring Break - no lecture *

March 28

Troy Butler, PRIMES, Dept of Mathematics, Colorado State Univeristy
Title: "Determination of unique characteristics of parameter estimation using statistical inversion"

April 4

Dave Schimel, National Center for Atmospheric Research, Boulder
Title: "Ecosystem modeling and data assimilation: nonlinear parameter estimation and the scientific method"

April 11

Dale Lockwood, PRIMES, Biology Dept., Colorado State Univeristy
Determination of population structure based on genetic data.

April 18

Tomi Vukicevic, Atmospheric and Oceanic Sciences, University of Colorado, Boulder
Title :A New Look at Data Assimilation and Inversion Problems for Geosciences

Jennifer Hoeting, Dept of Statistics, Colorado State Univeristy
Title: "Spatial models for ecology"

Abstract
Many problems in ecology involve the collection of spatially-referenced data for which traditional statistical models with independent and identically distributed errors are not appropriate.  The first part of this talk will focus on the perils of ignoring spatial correlation in statistical modeling including the impact of spatial correlation on model selection and the impact of sampling design on parameter estimation for spatially correlated data.  The second part will focus on the advantages of adopting a Bayesian hierarchical modeling approach for spatially correlated data. To demonstrate these ideas, we describe Bayesian methodology to predict a spatially correlated ordinal response. Our methodology for spatially correlated ordinal data was motivated by work for the U.S. Environmental Protection Agency to evaluate stream health. This talk includes joint work with several research groups including M.D. Higgs, K. Irvine, A. Gitelman, R. Davis, A. Merton, and S. Thompson.