[Talk-us] mapRe: (Second attempt) Potential data source: Adirondack Park Freshwater Wetlands

[Kevin Kenny]

On 03/16/2016 06:12 PM, Andy Townsend wrote:
> What I am a bit surprised about is that in the Adirondacks there's 
> relatively little track data in OSM.  Sure, New York State is big, but 
> it's not _that_ big.  It's roughly twice the size of Scotland and 
> (excluding New York City) about twice the population.  Parts of the 
> Adirondacks are about as far from major centres of population as parts 
> of the Cairngorms in Scotland are, and the Cairngorms seem to have 
> many more hiking trails mapped*.
Some areas of the Adirondacks, like Eastern High Peaks, are starting to 
have decent coverage (there are still trails missing, but there's 
enough, say, to complete a traverse of the Great Range or a trip over 
Cascade and Porter). With the default renderer, you have to zoom in 
pretty tight to see them. 
http://www.openstreetmap.org/#map=15/44.1196/-73.8963&layers=N is an 
example.

There are some other parts of the park that really are trackless 
wilderness; there are no tracks shown because there are none in the 
field! The Oswegatchie River, for instance, is best explored with a 
canoe. There are a fair number of marked campsites, but no trails other 
than portage routes. Even about half of the 46'ers (a list of high peaks 
- similar to the Munros) have no established trails. The would-be 
climber must find his own way among the dense vegetation, sucking mud 
and rock slides.
> One thing to be said in favour of a wetlands import is that these are 
> features that by definition it's difficult to map the entirety of from 
> the ground (it's a problem I'm familar with as it's the same reason 
> I'm only able to map the western part of the Derbyshire Peak District 
> in late summer when it's been dry enough for long enough). However a 
> worry is that because there's so little surveying done here no-one's 
> going to be able to sense-check the data so there's a worry that it'll 
> just "sit there" without any future modification. When I've done 
> stream and river mapping in e.g. South Wales I've always found it 
> useful to compare all of survey, government open data and imagery to 
> see what things should be mapped as, where imagery (or GPS data) is 
> offset and where government open data is inaccurate.  Do you have any 
> way of sense-testing any of the data to be imported?  Maybe it might 
> be useful to create e.g. a umap overlay of some of it that's 
> immediately usable and you can start collecting feedback from hikers 
> about what they'd categorise the features you're suggesting be imported.
Indeed. The APA wetlands inventory also includes all the open-water 
features (lakes, rivers, streams), which also appear in the existing 
"lakes and ponds" import, in the National Hydrographic Dataset, and, of 
course, in imagery. There are therefore at least three or fojr sources 
for cross-checking the more significant features. In the map that I keep 
referring to, I have all three overlaid, which is why you see multiple 
tracings of the shoreline on some of the lakes, as in 
http://kbk.is-a-geek.net/catskills/test3.html?la=43.2196&lo=-74.2107&z=14. 
There are a few areas that look quite the mess - 
http://kbk.is-a-geek.net/catskills/test3.html?la=43.5688&lo=-74.5912&z=15 comes 
to mind, but... with that particular one, I've been to that bog. Even in 
the field, it's hard to tell water from land. An unwary step on 
apparently dry ground might plunge you into knee-deep quicksand. The 
disagreement of the data sources there doesn't worry me in the least.

In fact, I think the best approach might be to confine the initial 
import to the open water features (appropriately conflated with what is 
already in OSM), with the other wetland features being added later. 
Doing the open water in small sections and conflating as we go would 
give us confidence in the alignment.

I can say that in several hundred km of walking on- and off-trail in the 
Adirondacks, I've quite reliably gotten my boots muddy right where the 
data set says to expect such a thing. I also, on examining the data, do 
not see contour lines drawn from NED crossing into open water. It's 
pretty well aligned with reality.

-- 
73 de ke9tv/2, Kevin