The maps purposely avoid counting any grade-separated crossings where bicycle facilities were included in a road project - ones that had to be built to get the road across a railway, river or another road. These are common. I tried to avid counting those in Oulu and Davis as well. The reason being that, while these are indeed useful, the application of these facilities to locations where bike paths meet collectors/arterials is fairly unique. 

What struck me about Oulu and Davis was threefold:

a. their similarity in urban/residential structure to Winnipeg

b. the similarity in terms of weather extremes and topography to Winnipeg

c. the usage of bicycle underpasses built specifically for AT that were intended only to avoid crossing collector/arterial streets or railways at grade. 

To me, what this represents is an acknowledgement by the designers/planners/engineers that, to a bicycle rider, even a collector street is akin to a river (or freeway or railway track) if it has to be crossed at grade. There is value in sending them underground instead.  Note that the relative traffic volumes of the roads being avoided is also interesting. Even roadways with the equivalent amount of traffic as say, McLeod or DesMeurons, seem to be candidates for an underpass in both Oulu and Davis. Note that I avoid discussion of bridges. They have their benefits, especially with high groundwater, but bridges have the drawback of being exposed to the wind and usually require steeper approaches (to clear large vehicles or containers, for example). Lastly, with regards to mode share, from what I remember, Davis had an overall 15% share of commuter traffic in 2010. For Oulu, it was a 12% bicycle mode share in winter, and around 32% in summer. Winnipeg, of course, sadly still hovers at around 2.6% overall (at least 5X less than Davis and 12X less than Oulu in the summer). Another reason the correlation stuck with me.