[Talk-us] TopOSM

[Kevin Kenny]

On 05/22/2016 02:39 PM, Lars Ahlzen wrote:
> On 05/21/2016 01:54 PM, Clifford Snow wrote:
>> TopOSM looks like a good candidate for a hack session at SOTM-US. Let 
>> me know if you are interested so I can find a room for people to meet 
>> on July 25th.
>
> That's not a bad idea. Sounds like there's some interest in the OSM 
> community, and I'd be happy to take a look at it again. I'd say go ahead!

Alas, I'm not going to SOTM, but put me down as someone who's interested 
in the project, with some 'skin in the game' already.

As I already posted privately to Clifford:

I picked up TopOSM's code for my own purposes and added quite a few 
twists of my own. I use the result as a basemap for several of my own 
projects. You can see what it looks like at 
https://kbk.is-a-geek.net/catskills/test3.html.

The 'catskills' in the name is something of a misnomer - it started as a 
Catskill Mountain project, but got out of hand and now covers most of 
the Eastern Seaboard. The 'test' in the name reflects that the page in 
question was intended largely as a demonstration viewer. I mostly use 
the tiles in apps on my phone.

At the moment, it's several months out of date. I've generally brought 
it fully up to date only a few times a year, because I've never troubled 
to get a pipeline going for automatic rendering with a tool such as 
tilestache. Given the limited area that it supports, I do all right 
serving static tiles. It generally takes overnight to produce them all, 
and warms my home office quite nicely when it's running. Clearly, I am 
NOT tooled up to serve it up on a large scale.

It depends on a good many publicly-available data layers with 
ODBL-incompatible terms. Life is full of tradeoffs. I think that a 
'sanitized' version with only ODBL and US Government data wouldn't be 
too difficult to put together.

I'm more than willing to share the code, but it would be a bit of a 
nightmare to set up. I think that the best approach would be to share it 
with a willing apprentice (if you will) in pieces, reworking as we go to 
make sure that each shared piece runs for more than just me and the 
setup is better documented than it is now. I'm willing to put in the 
effort to make such a project succeed, but would find it immensely 
difficult without a guinea pig to try stuff out and provide ongoing 
feedback.

Significant public layers that I include that I believe TopOSM does not 
include:

(1) NLCD, the National Land Cover Database. This provides the base 
colour, atop which is added hillshading. For my personal map, this is 
not something I'm willing to live without. It's important to me, when 
planning an off-trail trip, to know when to expect coniferous forest 
(near me, this is generally dense balsam-and-spruce that slows travel to 
a crawl) and when to expect deciduous forest (which, by contrast, is 
relatively open). It also pretty much obviates the need to use 
NaturalEarth's 1:10million populated areas shading, which is at an 
inappropriately coarse scale for most of our use of it.

In order to do this, I had to replace styles describing land use/land 
cover/cadastre to use shading on the boundaries rather than area fills. 
(I think the appearance in the end is nicer in any case.) I confess to 
colour-blindness and would welcome a more attractive colour scheme.

(2) A fair number of FCODES from NHD (National Hydrographic Dataset) 
that TopOSM ignores. Ignoring these FCODEs loses, for instance, the 
lower Hudson River on TopOSM.

(3) The USFWS National Wetlands Inventory. Most of the marshlands that 
are shown on my map come from there.

There are also a few state-level data sets describing foot trails and 
public land boundaries in my map that I am NOT comfortable including in 
TopOSM without consulting with the issuing agencies. Some of the 
agencies are not amenable to allowing redistribution of their data, even 
when the law requires them to make the data available to the public. 
Most of those are shown on the map in magenta, and serve me as a "things 
to do" list when I'm thinking up ideas for a short hiking trip.




-- 
73 de ke9tv/2, Kevin