CANSSI Annual Meeting Scientific Program: Spatial and Temporal Modeling in Climate Science and Public Health
Date and time: Saturday May 25, 2013, from 4:00 pm to 8:20 pm.
Place: Alumni House, University of Alberta
Co-sponsored by the Pacific Institute for the Mathematical Sciences (PIMS)
Each session will feature a lecture followed by discussion.
All are welcome, but space is limited. Please let Mary Thompson know by May 10, 2013 if you plan to come. Email: email@example.com
4:00 pm to 5:20 pm: Session I
“Influence of climate change on extreme weather events”
Richard L. Smith, University of North Carolina and SAMSI
Abstract: The increasing frequency of extreme weather events raises the question of to what extent such events can be attributed to human causes. Within the climate literature, an approach has been developed based on a quantity known as the fraction of attributable risk, or FAR. The essence of this approach is to estimate the probability of the extreme event of interest from parallel runs of climate models under either anthropogenic or natural conditions; the two probabilities are then combined to produce the FAR. However, a number of existing approaches either make questionable assumptions about estimating extreme event probabilities (e.g. inappropriate assumption of the normal distribution) or ignore the differences between climate models and observational data. Here, we propose an approach based on extreme value theory, incorporated into a hierarchical model to account for differences among climate models. A related technique, based on the same modeling approach, leads to quantitative estimates of how the probability of an extreme event will change under future projected climate change. We illustrate the method with examples related to the European heatwave of 2003, the Russian heatwave of 2010, and the Texas/Oklahoma heatwave and drought of 2011. This is joint work with Michael Wehner (Lawrence Berkeley Lab).
5:30 to 7:00 pm: Barbecue
7:00 pm to 8:20 pm: Session II
“An overview of spatial statistics in public health”
Patrick Brown, Cancer Care Ontario and University of Toronto
Abstract: This talk will first review the current `state of the art’ of spatial statistics as applied to problems in the health sciences, and follow up with a description of several examples of active methodological research in the area. Spatial health data is inherently non-Gaussian, being binary, count, or event-time data, and many problems in the health sciences fall within the framework of the Generalized Linear Geostatistical Model (GLGM). The use of Bayesian inference with the GLGM is demonstrated with applications including tropical disease risk in Africa and cancer risk around a nuclear generating station near Toronto. The conditional independence assumption in the GLGM is incompatible with the research questions typical in analyses of infectious disease incidence data, and MCMC-based inference for spatial infectious disease models is described. The remainder of the talk will describe current research in spatio-temporal modelling, fitting the GLGM to spatially aggregated data, and inhomogeneities in infectious disease models.