TIGER 2011

[maps4gps]

Census TIGER 2011 files were released late yesterday afternoon.

Any thoughts on how often a mapset should be revised/updated?

[Boyd]

Good question. All of my maps have their origins with TIGER 2007. My problem now is that I have heavily edited this data and made it my own to the point where it would be difficult to start over again with new TIGER. I have added roads, reclassified roads based on comparison with NJDOT's maps, deleted roads and replaced them with new ones traced from aerial imagery.

At some point I need to figure where to go from here. I suppose a good start would be downloading some of the new stuff and comparing it my own. One big issue in southern NJ is a large quantity of bogus "paper streets" in communities that were planned during the 19th century. But all the sources I know (TIGER, Navteq, Teleatlas, Google) show these bogus roads as though they were real. Today those roads could never be built there because the land is now protected.

[-Oz-]

Since I don't manually edit very often I just update whenever the new roads come out.  However it takes quite a bit of effort since I feel obligated to update GNIS data; trails, etc each time.

[maps4gps]

TIGER 2007 , which you, OSM and Ibycus use, is not as 'good' as more recent versions.  Slightly over 1/3 of the counties had not undergone the multiyear spatial realignment process.  All counties were completed for the 2008 version.  The 2008 version was the first where Census released the full extent of TIGER.  This included 'minor' features like private drives from county roads to farm/ranch houses and non-linear features being defined with twice or more coordinate pairs.

The total 2011 zipped file sizes are little changed from those of 2010, some are even slightly smaller.  In a quick check of some files, I did not note any change in the all to numerous miscodings of the data.

For the previous versions, I have also tried to correct/improve on some these miscodings.  The hundreds of hours I spend doing this each year is just too much time when the other data sources also need so much error correction and 'improvement'.

Using the 2010 files or the 2011 as downloaded looks more managable.  Until the recent bumping of one of the four most dowmloaded mapsets, three were done 3 or more years ago (two as simple data 'dumps').  Neither old age nor data content, etc. seam to be criteria for 'editor's choice' selection.

Maybe I will just concentrate on the improvements for the home State that I have had on the 'to-do' list for the past four years.  I have never received a comment, positive or negative (even when asked for on the download page) on any of the 'extra' data or processing I have put into many of my mapsets.     

[Boyd]

I'm definitely going to look at the new files. One problem I found with the 2007 roads is that there were duplicates in some areas (stacked on top of each other). Since I've massaged the data so much myself, it's always possible I screwed something up, but I don't think so. It resulted in much larger file sizes where this happened. The way I first noticed was because some road segments had a different appearance in Globalmapper when selected with the digitizer, due to duplicate below.

Is that a known issue with 2007 and has it been cleaned up?

[maps4gps]

I do not recall hearing or observing a duplicate issue.  I checked the user notes on the Census website for the 2007 version and do not see anything.  You may find it interesting to also check. 

[Boyd]

Just now downloaded my own county and am taking a very quick look. I see the same major problems as the 2007 version. Even though there are MTFCC codes for many different types of roads, they are not being used. Everything is classified as either S1200 (secondary road) or S1400 (local neighborhood road). So that puts the Garden State Parkway and NJ Turnpike in the same class as a 2 lane 50mph country road. And it puts county routes in the same category as neighborhood streets and dirt roads. Furthermore, the county route numbers are generally not given.

These are the same issues that I've worked so hard to correct in my own version. I'm going to study this more closely, but I think my own roads are probably a lot better. It is a lot of work to hand edit the roads. If I start making maps of larger areas, I can see how that won't be practical. But NJ is only something like 8000 square miles, and I'm getting to know it pretty well. 

[Boyd]

I do not recall hearing or observing a duplicate issue.

Just ran my globalmapper script to check for duplicates and didn't find any in 3 different counties. Then I checked one of my old 2007 downloads and didn't find any there either. Hmm. Maybe I duplicated the roads myself somehow along the way. 

[maps4gps]

The coding is very lacking in usefullness.  I have seen about every type of 'road' classified as some other type of road and even as a non-road feature.  Lots of short segments of city streets (even a few voter district segments a few miles from the house) being coded S1100 (Interstate highway) - makes the overview map very cluttered.  USGS mentioned on the National Map website they will use vendor supplied transportation in place of TIGER starting in 2012.

As the mission and funding of Census is for head (and house) counts; the street name and address is takes precidence over highway number, etc.  There used to be a database file (one per country) which gave names and/or route numbers for all the features which had one or more names/numbers.  It was not ordered, so the record for a county road might come before the record(s) for a State and/or Federal highway for the same segment of road.
I have not been able to code a name to display both a shield with number and a street name.  Census works on a county by county basis - nearly all the country level roads might be coded S1200 in one county and S1400 in the adjacent county. 

[-Oz-]

Until the recent bumping of one of the four most dowmloaded mapsets, three were done 3 or more years ago (two as simple data 'dumps').  Neither old age nor data content, etc. seam to be criteria for 'editor's choice' selection.

You know editor's choice maps are just recommended by people who mention them to me.  If you want a change just mention it.  I can't possibly have all the maps installed.  I try to remove and add as I notice major problems (for example ibycus being soooo old but still so popular0.

[Seldom]

(even a few voter district segments a few miles from the house) being coded S1100 (Interstate highway)

Some of those were intentional, like Abbey Road in SanJuan County, UT.  On the 2010 Tiger it shows as an Interstate, and did for several releases before that.  Alas, some more serious mapmaker has re-classified it to "residential road" in 2011.

[maps4gps]

Boyd,
   
     In my local area the spatial realignment was done in the 2008 version and there were noticable improvements.  There have not been noticable changes since.  I would expect some of the Denver metro area to be changing as some of the counties rank in the 5 fastest growth counties in the US.    The 2010 version was likely as up to date as possible for roads as it would have been used in the 2010 census of population and housing and for political reapportionment based on those counts.

    I considered only updating some of the State topos each year; however, working with quads I would process so many adjacent States almost all would have been processed anyway.  My thinking now is to update transportation one year, hydro another, and GNIS, etc. a third; with contours every year where needed (which would likely be everywhere as within a year almost every State has some some new/updated NED data). 

     Thanks for your input.