Climate Change Impacts on Ecology and the Environment

The workshop will bring together a variety of environmetricians in order to bring to bear modern methods of space-time modeling to assess and model the impact of climate change on forest and human health conditions. There has been only limited work by statisticians and stochasticians on the effect of climate change, and this workshop will enable us to set up a rich and politically important research agenda.

A primary aim of the workshop will be to discuss models and methods for the analysis of climate events and the assessment of the sensitivity of environmental outcomes to changes. More precise methods for assessment of evidence for existing impacts will be considered.

Topics include:

• Agroclimate risk analysis:
  Climate change has led to the need for agricultural risk management. For example, because drought has assumed increasing importance, Agriculture Canada has now allocated funding for managing associated risk. Such management entails the development of appropriate models for precipitation and methods for finding optimal management strategies. In particular, new methods are needed for spatial precipitation analysis that admit covariates such as elevation, wind direction and velocity, along with other climate covariates to improve the accuracy of precipitation. Moreover, recent advances in statistical theory make it possible to estimate aspects of the entire distribution of rainfall at any given location and time point. Such advances include work on quantile regression and the development of functional analysis methods for estimating the distribution of a response variable at any point over the continua over which the data are distributed. Alternatively, advances in hierarchical Bayesian modeling enable continuous covariate fields (like temperature) to be input as latent variables into discrete (or mixed) process models, such as the binary process of the presence or absence of precipitation, using the clipped Gaussian (or multivariate t) approach. This leads to a Bayesian version of disjunctive kriging appropriate for use in climate change impact modeling of precipitation fields.


• Statistical and deterministic models in georisk analysis:
  Neither statistical nor physical models like those in climate modeling suffice for modeling large-scale processes like those involved in climate change. A new modeling paradigm is spawning that seeks to combine the best elements of both. One promising approach uses Bayesian melding to integrate them. Central to this approach is the "truth” represented by a latent random Gaussian process. Deterministic models are assumed to be affine transformations of integrals of the truth over the grid cells that derived from the difference equation methods used to solve these deterministic models; monitoring site measurements, such as those involved in climate modeling, are similarly biased point measurements of the truth. A Bayesian hierarchical framework enables the integration of these two very different outputs. This framework enables uncertainty estimates to be added to deterministic outputs and those outputs to be dynamically tuned over time as data flows in. At the same time, the deterministic models may well provide statistical models with the ``backbones” they lack when confronting large - scale climatic processes. However, a number of research questions arise in connection with the use of deterministic models in a statistical domain. How should environmental processes best be monitored taking account of the need to protect human populations from the deleterious societal impacts of such change? How can deterministic models constructed at different scales of resolution be statistically integrated by the use of wavelets, for example. How can deterministic models be turned into practical forecasting tools, say to predict drought or exceptionally hot weather? Most environmental fields are not Gaussian processes but rather (after suitable transformation) multivariate-t processes. How can melding be adapted for use with such processes? How can the spatial melding approach best be turned into a space-time melding approach?


• Impact of Climate Change in Animal Populations and Human Health:
  In order to assess the impacts of climate change on animal populations, it is necessary to know basic biological parameters about animal populations such as survival rates, population numbers, population movement rates etc. These are often obtained using methods such a capture-capture, point-count, or distance-sampling. The analysis of such studies requires sophisticated analysis tools not usually available, nor accessible to most biological researchers. Discussions will focus on the development of suitable methodologies in capture-recapture, distance-sampling, and point-count methodologies for these large scale problems. In addition, tools for both modeling and analysis of current ecosystems and inferring past conditions from natural recorders will be investigated. Identified needs for improved statistical methodology include relating deterministic, fluid-dynamic modelling to observations containing considerable stochastic components. In the area of human health, methods for quantifying risk for certain diseases (asthma, cardiovascular disease) associated with climate change and specific strategies to assess health impacts will be considered. In a particular and important application, methods for forest fire smoke modeling and linkages with adverse events related to asthma will be discussed.


• Aims and Scope:
  The purpose of the proposed workshop is to engage climate change researchers in the scientific enterprise of developing novel methods for addressing these problems. It is envisioned that the proposed workshop would build upon collaborative initiatives already conceived through the environmetrics collaborative research group. Gaps in methodological developments will be identified, for example, methods for isolating the species and ecosystems most vulnerable to climate change. In addition, some techniques discussed will cross several of the themes, for example, detection of changes when observations are available at several spatial and temporal scales. This workshop will also play the important role of providing opportunity for discussion of timelines and progress toward the goals identified in the collaborative research group application on ``Georisk and climate change" by these organizers and will provide a forum for interim reporting on research objectives of that collaborative research group. It is envisaged that this workshop will bring together researcher in diverse scientific fields. It will also have a strong focus on student participation, bringing together students working in this area in a unique networking opportunity including both environmental and statistical sciences.