Re: [Talk-us] Questions about bsupnik's NHD OSM files

Wed, 13 Mar 2013 14:41:20 -0700

Thanks very much!  I've spent some time looking over your ogr2osm translations.  They are a huge improvement over the translations I managed to find a year and a half ago.

I think it's just ARCGis 9.3.1 gbd files that USGS has "prestaged" for easy downloading, isn't it?  I can still get shapefiles through the National Map Viewer.  It's just that it's tedious to do. I've worked through all of California, Oregon, and Washington; most of Nevada and Arizona; and parts of Idaho with shapefiles. If the mdb files are prestaged, too, maybe I ought to try to build ogr2osm with mdb support. Though I have been warned...

This is old business for you, so I understand if you have more pressing things to do right now that look over USGS fcodes.  But here are some comments, anyway.
So far I've just looked through your NHDArea translations carefully. It probably would have made more sense to start with Flowlines.  Several of your fcode translations are big improvements to mine, which in a couple of cases, I now see were just wrong. Here are the others:

NHDArea:

fcode 33600 (Canal/Ditch no attributes): I use waterway:riverbank rather than
natural:water.  I understand from the wiki that natural:water is for lakes.

fcode 33601 (Canal/Ditch aquaduct): I use waterway:riverbank.
This fcode seems to be used for some major canals, though not specifically for drinking water.
Example : Folsom South Canal, 38.3336308072 -121.187999122
A description is at
http://www.usbr.gov/projects/Project.jsp?proj_Name=Auburn-Folsom+South+Unit+Project
Another example: 34.1782742136 -114.613353799

fcodes 34305 and 34306 (Dam/Weir earthen and nonearthen, respectively):
The "dam=..." key appears to be novel, but looks good to me.  Would
earthen/nonearthen, the attributes usgs assigns, be more likely to be widely
adopted than earth/artificial?

I don't know if FType 343 includes real weirs as
well as dams.  The closest example I found was a 34306 spillway, cf,
38.2520654073 -120.846461923. The rest of 34306 seems to be concrete dams.

fcode 36200 (Flume) : These are rare in Area files.  In Flowline files they seem
to be more or less canal/ditches.  The one Area example I found looks like an
overpass for runnoff:
32.7471792158 -114.708445399

fcode 36400 (Foreshore) : I've been using natural:wetland rather than
natural:beach. At least the examples I know in  subregion 1804 make for poor sunbathing, eg
38.0659132742 -121.581538522

fcode 40300 (Innundation area no attributes) :  You leave this untagged.
I can't find any instances.  Maybe it should be tagged natural:water
intermittent:yes like other FType 403 fcodes?

fcodes 40307,40308,4039 (Innundation Area) : The use of the key 'intermittent' with
value other than 'yes' appears to be very unusual in taginfo, but it seems like an
excellent way to distinguish these three fcodes, at least so long as people don't
start using things like intermittent:no, which would make intermittent=* meaningless.

fcode 44500 (Sea/Ocean) : Is it because this is redundant with OSM that you don't
tag it natural:ocean ?

fcodes 46000, 46003, 46007, and 48400: USGS uses the last three
inconsistently, so maybe simply waterway:riverbank would be good for these four
relations, augmented by intermittent:yes for the last three. Some comments
regarding the last three of these fcodes:

fcode 46003 (Stream/River intermittent) 
I would guess that this is a stream that may dry up sometimes,
eg, 37.7107086081 -120.626779323 and
East Fork Carson River, N38° 57.305 W119° 46.945
But there is also
34.4807920798  -114.271963533
which looks more like what I'd expect for 46007 or 48400.

You are tagging 46003 with natural:water intermittent:yes, but suggest tagging
waterway:river or waterway:stream instead.  My understanding from
http://wiki.openstreetmap.org/wiki/Map_Features is that natural:water is for lakes
and that waterway:river and waterway:stream are are intended for
ways rather than relations. (Taginfo shows that a good deal of tagging doesn't
conform to the these wiki definitions, though.)

fcode 46007 (Stream/River ephemeral)
This seems to be used for usually dry stream beds, but used so rarely I can't be
sure,
37.612028475 -120.016223924

fcode 48400 (Wash) - Here are three very different examples with this fcode.
The first looks like a classic wash to me, ill-defined and normally dry,
N40° 23.424 W118° 46.276
Another is indistinguishable for me from 46007:
46.7309427941 -121.872430188
Another is indistinguishable by me from 46003:
36.8003856762 -120.164653791

Thanks again, Mack


On 03/11/2013 11:01 PM, Paul Norman wrote:

With all the address stuff I’ve completely dropped NHD off my plate.

 

I had been working on a NHD to OSM conversion and had gotten so far as to get a Potlatch2 instance with it as a background layer, see http://lists.openstreetmap.org/pipermail/talk-us/2012-November/009515.html and http://took.paulnorman.ca/potlatch2/potlatch2.html.

 

The NHD data model changed shortly after bsupnik wrote his conversions and some fcodes changed. This makes using his conversions difficult.

 

I would welcome a review of the tagging which I am proposing, which can be found at https://github.com/pnorman/ogr2osm-translations/blob/us_nhd/us_nhd.py. I can also generate a .osm file for any sub-basin (4 digit code), or extract part of a subbasin.

 

ogr2osm isn’t too hard to run, but compiling ogr to have .mdb support to open up the files from the USGS is difficult, so if you want me to convert an area let me know the bbox and sub-basin.

 

From: Mack Stanley [mailto:[email protected]]
Sent: Monday, March 11, 2013 9:34 PM
To: [email protected]
Subject: [Talk-us] Questions about bsupnik's NHD OSM files

 

I have some questions regarding bsupnik's osm xml files made from the NHD
flowline, waterbody, line, and point shapefiles.   

I was very happy, belatedly, to find these files graciously created and hosted by
bsupnik. Fall of 2011, I worked up some ogr2osm translations to do the same thing
on shapefiles laboriously downloaded using the balky USGS National Map Viewer,
unaware of bsupnik's work.

I've read a lot about the tagging and integration issues that make bulk uploading
NHD data to OSM a bad idea.  My goal was just to use it to make a hydrography
layer for my own use, while conforming to consensus tagging conventions,
potentially to make it useful to others.

I'd love to use bsupnik's files, instead.  My main problem is that GNIS names of
streams and waterbodies aren't included.  I've seen this mentioned in this list's
archives,  but couldn't find why this is, or how to fix it.  

I looked into adding name tags from GNIS feature files, but this looks more or
less impossible. Names in feature files are attached to points.  The name point
for a tributary is often at its confluence with the stream it feeds, making the
name assignments difficult to determine mechanically.

The rest of my questions are about tags. I've looked in detail at four files
(15010014, 16040201, 17090006, and 18040011).

Three great things about bsupnik's files are that they are small, they are easy to
read by humans, and the ultimate objects (points in point files, ways in flowline
files, and relations in area and waterbody files) are tagged with their fcodes.
This last feature means that other tags are frosting on the cake---it's trivial to
rewrite with different tags. Bsupnik's wise preference seems to be to simply
leave objects without regular osm tags in cases where the correct tagging is
uncertain.

Nevertheless ...

In Flowline files, why is

fcode 33400 (connector) is tagged waterway:stream rather than waterway:canal ?

fcode 46000 (Stream/River no attributes) is not tagged, rather than
waterway:stream ? (Other Stream/River fcodes are tagged waterway:stream.)

fcode 46007 (Stream/River ephemeral) is tagged just waterway:stream, rather than
waterway:stream intermittent:yes ?

In Waterbodies files, why is

fcode 39001 (Lake/pond intermittent) tagged natural:water
occurrence:intermittent, rather than natural:water intermittent:yes ?

All of the other intermittent tags that I have run across in the bsupnik nhd files
follow the more common usage (judging by taginfo) intermittent:yes rather than occurrence:intermittent.

fcode 46600 (Swamp/Marsh no attribute) is tagged natural:wetland, but
fcode 46601 (Swamp/Marsh intermittent) is not tagged, and
fcode 46602 (Swamp/Marsh perennial) is not tagged.

Why not tag all three natural:wetland with intermittent:yes added to 46601 ?

Thanks for your time and interest,

Mack


_______________________________________________
Talk-us mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/talk-us

Reply via email to