[Talk-ca] [Imports] Proposed import: Surrey, BC waterways

[Paul Norman]

I've rendered an area locally at
http://maps.paulnorman.ca/surreywater.html?bbox=-122.88,49.08,-122.82,49.12

This is OSM data + drains + streams. Culverts are not included. There is
some overlap between NHN and Surrey data which would not be present in the
import.

An example of the worst case for waterways is can be found in Bridgeview at 
http://maps.paulnorman.ca/surreywater.html?bbox=-122.8773,49.2068,-122.8680,
49.2108

Hopefully this will help with determining if we want the roadside ditch
information.

> From: Paul Norman [mailto:penorman at mac.com]
> Subject: Re: [Talk-ca] [Imports] Proposed import: Surrey, BC waterways
> 
> > From: Richard Weait [mailto:richard at weait.com]
> >
> > On Tue, Mar 29, 2011 at 2:31 AM, Paul Norman <penorman at mac.com> wrote:
> > > In short, the waterways in Surrey BC are largely from NHN data which
> > > in some cases is accurate, but tends to be 20-30 years old. I'm
> > > proposing importing data from the city, licensed under PDDL. I'm at
> > > least 2-3 weeks away from being ready to do the import as I have
> > > some other priorities currently but I wanted to get feedback.
> > >
> > > I see this import as fixing two problems. The first is the old NHN
> > > data. The second is the high number of unmapped ditches in Surrey.
> > > Because of the nature of the soil, it is very common for houses to
> > > have drainage ditches in front of them. These are largely unmapped,
> > > but the shapefiles appear to have all ditches.
> > >
> > > Legal: The data is licensed under PDDL, a compatible license.
> > > Accuracy: The positioning is very good.
> > > Completeness: Not all culverts are mapped. Random checks revealed no
> > > missing ditches or streams.
> > > Existing data: It was all mapped by me or imported. I am proposing
> > > removing it.
> > >
> > > There are more details at
> > > http://wiki.openstreetmap.org/wiki/Canada:British_Columbia:Vancouver
> > > /I mports /Surrey/Waterways about the exact tagging. I intend to
> > > refine this, but this may require a couple of site surveys. Before
> > > importing I will post a sample .osm file for review.
> >
> > I think that the Surrey, BC data under PDDL is wonderful.  And I know
> > that you have previously looked at their address data and road data
> > and found it very helpful with addr:, maxspeed and hgv information.
> 
> This data is more controversial, which is why I'm asking for feedback
> early in the process.
> 
> >
> > I wonder if these common roadside ditches will share the same
> > drawbacks as sidewalks in some other areas.  In fact, if sidewalks
> > exist in this area as well, the ditches may compound the problems of
> > sidewalks.  In summary portions of the OSM community are divided about
> > sidewalks.  Some feel that mapping sidewalks isn't needed where they
> > are common.  If all roads have sidewalks, why map them separately?
> > Others feel that sidewalks should be indicated as tags on roads, like
> > sidewalk=both for both sides.  Others feel that mapping each sidewalk
> > with a separate highway=footway way adds important detail to the map
> > for pedestrians.  Still others find that editing an area with a
> > substantial number of separate sidewalk ways makes editing more
> > difficult;  smaller download areas are permitted, editing the shape of
> > a road means editing the shape of two sidewalks as well, adding other
> > objects like building and POI have additional ways to "snap to"
> > unintentionally.
> 
> Good points. If we decide against importing the ditch data, I can rework
> the files to only include streams and named ditches.
> 
> I think the ditches are worth mapping. The ditches in question are
> typically 2-5m wide and can be seen on imagery.
> 
> They have culverts to allow foot and vehicle traffic to pass over them.
> I view them in many ways as analogous to fences on private property
> which are often mapped. Like fences, they are physical features which
> are significant if crossing, but placed in a way that doesn't impact
> most foot traffic.
> 
> The shapefiles can be broken down into 4 relevant groups of ways.
> 
> The first of these is streams. These typically meander, are in forested
> areas, are natural as opposed to constructed, and often have names. In
> my view, these are definitely worth importing.
> 
> The second of these is ditches in rural and agricultural areas. These
> are relatively long with unbroken stretches. They sometimes have breaks
> where they go into culverts to go under roads or footways. These are
> also worth importing, but hard to distinguish from the third category.
> 
> The third is ditches in suburban areas. These are used for drainage.
> They are 2-5m wide, and frequently broken by culverts. They go around
> blocks often and don't cross under roads. They eventually drain into the
> storm drain system. This is what the debate should be about. I feel
> they're worth mapping as they are significant physical features.
> 
> The ditches are not found everywhere, I found them mainly in areas
> constructed in what was damp soil in the 1970-1990s.
> 
> The fourth is culverts. They come from a separate shapefile and join the
> breaks in the above classes.
> 
> > A road with two  sidewalks and two ditches sounds like a further
> > complication.  There would then be five ways for each road?  Perhaps
> > mapping each ditch, where they are common is over the line towards
> > photo-realism or "micro-mapping" in my book.
> 
> The ditches are typically in areas without sidewalks.
> 
> >
> > Would these ditches then need additional tunnel / culvert splits at
> > each intersection?
> 
> The data has them, but they typically curve around the block and don't
> go under roadways.
> 
> I've extracted an area and posted it at
> http://maps.paulnorman.ca/surrey/samplewaterways.osm
> 
> Some of the tagging still needs work, but it gives an idea of the
> various areas encountered. To the north are ditches, split by culverts
> at driveways.
> In the south are agricultural ditches. Running down the middle is a
> stream.