I am most curious whether it is possible to use Landsat or other freely available imagery (but info on any other method would great as well) to locate and track the movement of woodland caribou herds over time. I only have some experience doing this type of work with forest depletions and vegetation analysis but am curious whether similar techniques exist for locating and tracking animal herds.
there are a few examples of animal counting by remote sensing (whales, gnu, crocodiles, seals...), but they used higher resolution satellite images (<1m) or aerial photographs (see this paper) and there was a clear spectral difference with the background (sometimes in UV or infra-red)). As a rule of thumb, you should have around 10 pixels to detect an "object", so it is obviously not possible with Landsat (15 m).
Maybe you could detect the herd as an object, but I seriously doubt that it would work because the distance between animals of the same herd can be larger than the pixels.
In my experience, satellite imagery, even high resolution, has not been too effective in tracking caribou, and the best option in applying landsat or other imagery is to use predictive habitat models derived from gps tracking, or directly utilize gps tracking. Landsat based landscape classification provides a very good basis for where to go looking for caribou, but wouldn't provide much in the way of actual caribou observations. In some cases, you could probably track major catering areas of larger herds, but unlike barren ground caribou, many woodland caribou form only small, less cohesive groups, often only during the winter.