Statistics Professor Peter Guttorp fiddles with a remote control, pointing it at a device to make a videoconference connection with the University of British Columbia, Vancouver. The image of a classroom pops up on a large projection screen, but the room is empty aside from a lone custodian tidying up. “They’re not ready yet,” Guttorp says with a laugh.
Guttorp teaches a statistics graduate course to not one, two, or even three universities this quarter — ultimately, he connects with UBC Vancouver, UBC Okanagan and Simon Fraser University to teach seven students via videoconference, in addition to the five UW students who attend in person twice a week.
The course is spatial statistics — meaning statistics with a specific geographic focus — and it’s part of Guttorp’s broader vision to develop a multi-site environmetrics research center and certificate program as part of a collaborative research group funded by a grant from the Pacific Institute for the Mathematical Sciences (PIMS).
Environmetrics is the quantitative analysis of environmental data, and can also involve mathematics and computer science. “It’s dealing with scientific questions in the environment,” Guttorp explained. “It can be climate research, looking at [carbon footprints] in climates or looking for things we expect to see in data because of climate change.”
Guttorp teamed up with Charmaine Dean of SFU, Jim Zidek of UBC Vancouver and Sylvia Esterby of UBC Okanagan to form the environmetrics collaborative research group in 2007.
“These were people that I’d worked with, and we decided we wanted to do this,” Guttorp said. Zidek is a former UW statistics professor who worked with Guttorp on environmental statistics in the early 1980s; Esterby was Guttorp’s predeccessor as president of the International Environmetrics Society (he held the position from 2002-2004); and Dean and Guttorp share similar research interests.
“I used to run a research center [at UW] funded by the EPA called the National Research Center for Statistics and the Environment, and [the environmetrics group] was sort of a natural extension of that,” Guttorp said. “A lot of us have done work on air quality, so we’re looking at things like ozone and particulate matter, both health effects and models for how it spreads.”
Guttorp teaches models for analyzing data related to other environmental questions as well, like how to estimate groundwater permeability to figure out where pollutants in water might end up, and how to determine oil permeability through different types of rock to find out where it flows underground in an oil field. The class has also studied biological monitoring — for example, examining how water creatures are affected by pollution.
One of the most important things Guttorp teaches is how to estimate data where no concrete information is available. “We have monitoring stations here and there, but we want to know what the pollution level was at a place where we didn’t have a monitoring site,” he said. For this, he teaches the Sampson-Guttorp spatial defomation method, an approach he developed with Paul Sampson, a UW statistics research professor, when there were no other models that could estimate data in such a way.
Guttorp is grateful to be able to offer the spatial statistics course to the Canadian universities using videoconference technology, which was funded by the College of Arts and Sciences. Otherwise, the course would not have been offered because too few students were registered at each campus. “It gives other universities access to the expertise we have, and in turn we get access to their expertise, which allows us to teach a broader set of topics than we do otherwise,” he said.
The collaborative research group uses its PIMS grant to take students to conferences and workshops, allowing them to meet with people from all different parts of environmental science — statisticians, mathemeticians, engineers and even political scientists. Guttorp, graduate students and their colleagues from UBC Vancouver and SFU plan to attend one such workshop in Malta this March. The focus will be trends in extreme climate events. “We’ll be looking at very high temperatures, very large rainfall events, things like that,” Guttorp said.
Guttorp and the group hope to develop the environmetrics program so that graduate students at any of the four universities can take courses to earn an environmetrics credential on top of their statistics degrees.
“The basic idea is that one-third of the credits would be environmental sciences and two-thirds statistics or applied mathematics that’s relevant to modeling environmental data,” he said. “The difficulty is that we need to sit down and see what the exact rules are at each university.” Students currently taking spatial statistics do not have the option to pursue the environmetrics credential, but Guttorp hopes the program will be finalized next year.