Impromptu meeting tonight, Burgerville on Powell Blvd, 7 ish to 8:15 ish

[Katie Urey]

Hi All,

Brylie Oxley and I will meet tonight at the Powell Blvd Burgerville, 7ish to 8:15 PM.  Last time we met we discussed outlining user-stories to start QGIS addin development.  I expect that tonight we will either return to the user stories or look at tutorials or documentation on writing QGIS addins.

The Jammers are also setup for weekly PSU, Cramer Hall, lab 469 meeting, 6-8PM.  Non-students are welcome in the lab during this 2 hour slot.

best regards,

--
Katie Urey
Portland, Oregon
503-228-3083 (land line with saxophone greeting)
503-780-9034 (cell, though not always "on")
ksu...@gmail.com

[Scott Parker]

That's Thursdays right?

 

Scott

--
You received this message because you are subscribed to the Google Groups "Oregon Walks GIS Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to OregonWalksGISJa...@googlegroups.com.
To post to this group, send email to OregonWalk...@googlegroups.com.
Visit this group at http://groups.google.com/group/OregonWalksGISJammers.
For more options, visit https://groups.google.com/groups/opt_out.

[Katie Urey]

yes, to clarify, our regular meeting is Thursdays, 6-8PM, Cramer Hall lab 469.   NOTE-PSU student status is not required,   The lab is secured by a locked door, During this time, please remember to knock and our contact us ahead of time so that we listen for people knocking.

My cell-503-780-9034

[Scott Parker]

Let's see if we can make a walkway network from the merged data that Michael sent.  Travis can pick the neighborhood.

[Michael Arnold]

The area of the merged network was constrained by the 'sample' Portland Centerline data that Scott sent me way back.  If this doesn't include an area you are interested, I can easily generate merged data for a different area - just need the relevant centerline data.  But note that this version only matches about half the OSM segments.  I'll have a better network soon. :)

[Scott Parker]

Good point.  I'm experimenting with Powellhurst-Gilbert.  We have clipped datasets for street centerline, grocery stores, and dwellings for this neighborhood.

[Scott Parker]

The attached illustrates some of the issues in automated walkway generation from OSM.

 

Note the different categories of ways

 

 

[Travis Driessen]

Let's do it for Arnold Creek!

On Mon, Feb 24, 2014 at 9:32 AM, Scott Parker <jscott...@comcast.net> wrote:

--

Travis Driessen

Graduate Certificate in Geographic Information Systems 2014

Portland State University

travis....@pdx.edu

[Scott Parker]

It will need to be somewhere in East Portland to use the data that Michael has created already.  I would not recommend trying to use OSM data for your GIS II project.  It will be a dozen or so hours to get a walkway network made from the OSM data.  What we can do Thursday evening is to examine some of the problematic situations and discuss possible schemes for processing them into walkways.  I doubt if we will have time to fix them.

[Travis Driessen]

Alright, let's talk about it Thursday.

[Scott Parker]

This was a long somewhat technical explanation of various features that was lost by my e-mail bug.  I thought I had solved it but apparently not.   I'm rewriting it as best I can. 

 

Walkway routing is problematic where the walkway network is incomplete.  It is necessary to connect the incomplete walkway network to the street network.  Thus the 1/2 crosswalks connecting the diagonal walkway to 82nd and to Holgate.  This requires nodes a few yards from the center of the intersection and that requires confusing, short sections of 82nd and of Holgate.

 

 

Note the different categories of ways in joined_clipped in the TOC (left of screen).  Every way has a category in the HIGHWAY attribute and some footways have a category in the FOOTWAY attribute.

[Scott Parker]

This was a long somewhat technical explanation of various features that was lost by my e-mail bug.  I thought I had solved it but apparently not.   I'm rewriting it as best I can. 

 

Walkway routing is problematic where the walkway network is incomplete.  It is necessary to connect the incomplete walkway network to the street network.  Thus the 1/2 crosswalks connecting the diagonal walkway to 82nd and to Holgate.  This requires nodes a few yards from the center of the intersection and that requires confusing, short sections of 82nd and of Holgate.  If there was a consistent way of doing this we could find a way of dealing with it in the walkway generator.  My observation of OSM data is that there are many schemes for incorporating walkways.  This is to be expected with data entered by the public.  Observation and data entry using standards requires organization and training.  We can't expect this to be done on any significant scale without paying people to write the standards, organize for coverage, train people, and do the data entry.  In fact much of the walkway information in the Portland area was entered by paid staff at TriMet.

 

To use OSM data we'll need to resolve cases where existing manually entered walkways conflict with generated walkways.

 

Some of the segments categorized as "service" are good pedestrian ways.  In some cases they have real separated walkways.  Others are just painted lines on a large paved parking area.  This isn't too much of a problem but we either remove them manually or accept a somewhat silly set of walkways.

 

Note the different categories of ways in joined_clipped in the TOC (left of screen).  Every way has a category in the HIGHWAY attribute and some footways have a category in the FOOTWAY attribute.

 

The HIGHWAY and FOOTWAY attributes illustrate a common problem in database design.  The categories are not mutually exclusive and there is more than one real world attribute implied in the categories.  As I interpret it there are four real attributes involved: function, permitted users, physical facility, and land use.  Motorway (better named limited_access), primary thru tertiary, footway_crossing, footway_sidewalk, footway unspecified, service, and track belong as categories in the function attribute.  Pedestrian, motor vehicle, except trucks, bicycle, wheelchair would be typical categories for the permitted users attribute.  Only pedestrian is actually available and it is mixed with categories from other real attributes.  One is forced to choose whether function is the important feature and categorize a way as footway-sidewalk or if user is the important feature and categorize it as pedestrian.  The physical facility attribute would be the place for information about steps and curb ramps.  Steps and path are available, but again, they are mixed in with other types of categorization.  Curiously, there is only one land use category available, and that is residential.  It doesn't make sense to me without other categories such as commercial and industrial.  I think it would be better to remove this from the network attributes.  Land use is better as a thematic overlay.

 

Cycleway is a unique problem.  Some judgment is needed to categorize a facility as part of a street or as a separate way.  In the same way that OSM has added "_link" to some street types it may be good to add "_with_bike_lane" to some street types.  This gets messy if you want "secondary_ramp_with_bike_lane", etc. so it may be better to add another attribute column.  This allows WAY = secondary with BIKELANE = paint_line and also WAY = bikeway with BIKELANE = NA.

 

The real problem with all this is that it isn't at all clear how to enter information about walkways.  Some of this is due to the nature of data entry by citizens who are not trained on well documented standards.  Some of it is due to the design of the OSM segment attributes and categories.  We can find most places with steps but it will be difficult to turn this into a sidewalk or crosswalk that is weighted by the existance of steps.  Just because a way is categorized as "path" or "pedestrian" or "footway" doesn't mean that it does not have steps.  We won't know if a walkway is a manually entered sidewalk that duplicates a generated walkway of an off street path that needs to be added to the walkway network.  In the end we'll need to examine each way that is not definitely a street and resolve it manually.

 

Scott

[Michael Arnold]

Wish I could tele-attend your meeting tomorrow, but unfortunately am busy elsewhere.   Hopefully I can get involved in this discussion soon.

[Katie Urey]

Hi All,

Add to tomorrow's agenda- test the capability of google hangouts in this PSU lab.   We are not able to install software on the computers in this lab, and I doubt that this will change.

Scott and or Travis, please bring you personal laptops so that we can test those devices's ability to get through the PSU network to the real world.   I will also bring my mobile hotspot.

Katie

[Scott Parker]

We have an opportunity to work with a student from the Portland Community College (PCC) GIS program.  The student would get credit for working as an intern at Oregon Walks.

[Katie Urey]

Scott- much to consider here.

I think the basic question is how do OSM tags (key,value pairs) associate with walkway elements. 

The freeform tagging of OSM may require someone to become more expert in composite keys or in the postgres hstore data type.

While trying to understand each of Scott's comments, I've learned a bit about OSM and about the data in the Portland area.

-There are many many ways tagged highway:residential.  My hunch is that if this is from a bulk upload, we might discuss tagging some ways in the trunk,primary, secondary,tertiary,unclassified hierarchy and dropping the land-use-descriptive tag, "residential."

OSM validity checkers find node and connectivity errors in the Portland area.  We should clean this up in OSM.

- this is only a partial reply to Scott's comments, but I send it anyway to keep the valuable discussion alive.

best,

Katie

--

You received this message because you are subscribed to the Google Groups "Oregon Walks GIS Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to OregonWalksGISJa...@googlegroups.com.
To post to this group, send email to OregonWalk...@googlegroups.com.
Visit this group at http://groups.google.com/group/OregonWalksGISJammers.
For more options, visit https://groups.google.com/groups/opt_out.

[Cone, Paul]

Should loop in TriMet on any proposed OSM changes to the data in the local area.  I hear they have hired an FTE to do OSM maintenance.

---

From: OregonWalk...@googlegroups.com [mailto:OregonWalk...@googlegroups.com] On Behalf Of Katie Urey

Sent: Tuesday, March 04, 2014 5:29 PM
To: OregonWalk...@googlegroups.com
Subject: Re: Oregon Walks GIS Jam:: using OSM for walkway generation