View video in MPG format

Transcript

[Parks Canada beaver logo]
[title: parkscanada.gc.ca]

[text on screen: A group led by Donald Mclennan from Parks Canada builds a model to predict ecotypes and create an ecotype map of Ivvavik National Park.]

[scientists plot out GPS co-ordinates in valley]
Phil Wilson: 690943.177

[Donald McLennan talks to us as he and his colleagues analyze a meadow and it’s habitat]
Donald McLennan: What Paul is doing, we described this plot in some detail, but the fact is because there’s variability in the structure, in the composition of this site, there’s also going to be difference in reflectance. So with a difference in reflectance we want to capture all of that variability because what we want to map is this site, this spruce, mesic willow-birch site. That’s what we want to be able to map. It’ll have a range of different signatures in the electromagnetic spectrum because of the difference in the structure of the ecosystem. So what we’re getting him to do is walk around and take waypoints in all the kind of variability that’s in here so we can have that to characterise the entire ecosystem.

[text on screen: In order to build a model of the entire park, geographic and ecological data is collected on the ground.]

[text on screen: The ground data is then compared to a satellite image of the same area in order to understand what the satellite is seeing (“reflectance”.) ]

[text on screen: This information is then used to infer ecotypes in areas showing similar satellite characteristics that have not been visited on the ground.]

[Donald McLennan talks to us as he and his colleagues analyze a meadow and it’s habitat]
Donald: and we have two ways of inferring that. One is a satellite data, and the other is a geographic map data. And we’re finding in our models that the geographic map data, where the ecosystems are in the landscape and how they’re characterised in terms of their shape, what the slope is, what their elevation is, what their aspect is, as it might relate to things like the amount of heat they’re getting in their angle to the sun or the amount of snow that they can accumulate in the winter to protect them from the cold desiccating winds, we’re finding that is a stronger predictor in our models than the actual satellite signal. So these two things together are giving us a very powerful model for how to reach out to the rest of the park, those two kinds of information. So each one of these sites, while we’re here, we have the map data for it, we have the map data for every one of those waypoints that Paul collected so we have the map data, we know it’s flat, we have the ecosystem data, we know what kind of a land form it is, and we have the satellite data to give us the electromagnetic signature for this area. And based on that from points we do know, we know everything except the ecotype we’re trying to model in the other parts of the park, so we build a mathematical model that characterises these types and when we find those conditions again we infer that that is this ecotype across the park. That’s how our model works.

[Paul Zorn discusses the habitat]
Paul Zorn on camera: When we generate these models, actually, model simplicity is a value in models. Generally speaking, the more complex these models get the more room for uncertainty and error there will be. So, models are almost always a simplification of nature. So when we make these predications we’re not trying to identify every single relevant factor that predicts these ecotypes. What we’re trying to isolate are those few simple things that really drive the occurrence of these ecotypes. They may not explain with perfect precision and accuracy all the ecotypes and where they are and where they start and where they stop, but what we want to do is make these models accurate enough to make useful predictions, but not so complex enough that they’re hard to understand, that they’re hard to interpret because that will decrease the usefulness of these models. So we want a balance between simplicity and accuracy, generally speaking.

[text on screen: To learn more about Parks Canada visit www.pc.gc.ca]

[title: Created for Parks Canada by HyperfocalFilms www.hyperfocalfilms.com]
[title: Parks Canada/Parcs Canada government logo]
[title: Copyright - Her Majesty the Queen in Right of Canada, represented by Parks Canada, 2010]
[title: Government of Canada logo]
[end]