Geographical boundaries for the continental US tz zones
I am trying to figure out the geographical extent of the 19 continental US timezones defined by tz. Because my goal is ultimately to create a freely available digital map, I would like to get definitions in terms existing, freely available digital maps, and the ones at nationalatlas.gov seem appropriate. Specifically, I plan to use countyp020 (county boundaries 2001), statesp020 (states) and timeznp020 (time zones) [see below for more details on those maps]. I tried to translate the comments in the "northamerica" file in terms of operations on the elements of those three maps, but I have a few questions/comments. 1. I would think that America/Boise, America/Denver and America/Phoenix together correspond exactly to the Mountain time zone of timeznp020. However, the later includes the southern quarter of Idaho County, Idaho, while northamerica suggests that this county is entirely in America/Pacific. Who is right? 2. The description of America/Denver includes "far eastern Oregon", and the description of America/Boise includes "eastern Oregon". Comparing with timeznp020, I suspect that America/Denver should not include "far eastern Oregon". (and if timeznp020 is right, "eastern Oregon" could be better described as "most of Malheur County, Oregon"). 3. What does America/Indiana/Indianapolis cover? Is it all the counties of Indiana not listed in any of the other zones? 4. The overall description of Indiana includes: # - Daviess, Dubois, Knox, Martin, Perry, and Pulaski counties # have been like America/Indiana/Vincennes. # # - Crawford, Pike, Starke, and Switzerland counties have their own time zone # histories as noted below. However, it seems that Pulaski now has it own zone (America/Indiana/Winamac), so it should probably be moved from the first to the second bullet in the quote above. 5. The description of America/Indiana/Marengo is that it covers the eastern part of Crawford County. How is that county divided, and in which zone does the western part fall? 6. What part of Kentucky goes in America/Kentucky/Louisville? At this point, I am not trying to handle border towns such as West Wendover, mostly because I have no idea of where to find a boundary for such towns. Any suggestion on what to do for those cases is welcome. For those of you who are curious about the content of the maps cited above: you can download them from <http://nationalatlas.gov/atlasftp.html> and visualize them using standard GIS tools (I use the freely available OpenJUMP and uDig). Alternatively, I have created a KML version of timeznp020 (using org2ogr) and placed it at <http://efele.net/timeznp020.zip> (4MB, 15MB uncompressed); you can visualize it in Google Earth (File->Open); I assume that the built-in 1st level and 2nd level admin regions (under "borders") are close enough to statesp020 and countyp020. Below is my current recipe to build my map in terms of operations in the source maps. Comments welcome. Thanks, Eric. --- Phoenix timeznp020/(Mountain & no DST) Boise statesp020/(Idaho+Oregon) intersected with timeznp020/Mountain North_Dakota/Center countyp020/(Oliver County, North Dakota) North_Dakota/New_Salem countyp020/(Morton County, North Dakota) Indiana/Marengo countyp020/(Crawford County, Indiana) // should only be the eastern part of Crawford County? where is the // limit? where does the western part go? Indiana/Vincennes countyp020/(Daviess, Dubois, Knox, Martin, and Perry Counties, Indiana) Indiana/Petersburg countyp020/(Pike County, Indiana) Indiana/Knox countyp020/(Starke County, Indiana) Indiana/Winamac countyp020/(Pulaski County, Indiana) Indiana/Vevay countyp020/(Switzerland County, Indiana) Kentucky/Louisville countyp020/(Clark, Floyd, and Harrison counties, Indiana) // what part of Kentucky is in this zone? Kentucky/Monticello countyp020/(Wayne County, Kentucky) Indiana/Indianapolis ? // statesp020/Indiana minus what is already covered above minus // Dearborn and Ohio County (they are in America/New_York) and // minus Gibson, Jasper, Lake, LaPorte, Newton, Porter, Posey, // Spencer, Vanderburgh, // and Warrick counties (they are in America/Chicago) Menominee countyp020/(Dickinson, Gogebic, Iron, and Menominee Counties, Michigan) Detroit statesp020/Michigan minus what is already covered above Los_Angeles timeznp020/Pacific minus what is already covered above Denver timeznp020/Mountain minus what is already covered above Chicago timeznp020/Central minus what is already covered above New_York timeznp020/Eastern minus what is already covered above ----
Date: Wed, 30 May 2007 22:55:25 -0700 From: Eric Muller <emuller@adobe.com> Message-ID: <465E634D.2080301@adobe.com> | I am trying to figure out the geographical extent of the 19 continental | US timezones defined by tz. Sorry, I can't help with that, but your question is a fairly common kind of one - we keep having people attempt to draw maps from the timezone data. I don't believe that it is really particularly suited for that, nor that it should be. The aims of this project are to identify all of the distinct timezones that exist, as best we can - it isn't part of our aim to specify just who is supposed to use each of those timezones, that's what the local legislature or executive administration (or whatever) does - or attempts to do, usually out near what you'd like to be a nice clear boundary between zones there's actually lots of ambiguity. This project doesn't care - as long as each individual user can decide "my local time is always the same as that in New York city" (or wherever) then they can select data that is going to be correct for them. It is irrelevant whether their neighbours use the same zone, which side of some boundary line they're on, or anything else like that. This all means that we don't much care where the boundaries are and there's no reason that we should, even if there really are nice fixed boundaries. What is more important here is when someone discovers that some region uses a mapping from UTC onto wallclock time that isn't the same as any that we have, or when one of the existing ones changes its rules. The actual set of people that any of this applies to isn't so important (as long as it is more than 0, or it isn't worth distributing). None of this means that what you're doing isn't useful, and if and when you find errors in the zones, or the comments about the zones I'm sure they'll be corrected. It also doesn't mean that there aren't people on this list who can answer some or all of your questions - nor that you shouldn't have asked them. I'm just trying to discourage any (more) bending of the timezone data towards the needs of map makers - that isn't the objective. kre
Eric Muller <emuller@adobe.com> writes:
1. I would think that America/Boise, America/Denver and America/Phoenix together correspond exactly to the Mountain time zone of timeznp020. However, the later includes the southern quarter of Idaho County, Idaho, while northamerica suggests that this county is entirely in America/Pacific. Who is right?
Most likely the National Atlas, if nobody says differently. You can consult the U.S. Department of Transportation (DOT) to make sure. Might have to call them or consult printed records...
2. The description of America/Denver includes "far eastern Oregon", and the description of America/Boise includes "eastern Oregon". Comparing with timeznp020, I suspect that America/Denver should not include "far eastern Oregon". (and if timeznp020 is right, "eastern Oregon" could be better described as "most of Malheur County, Oregon").
Sounds good.
3. What does America/Indiana/Indianapolis cover? Is it all the counties of Indiana not listed in any of the other zones?
Yes.
However, it seems that Pulaski now has it own zone (America/Indiana/Winamac), so it should probably be moved from the first to the second bullet in the quote above.
Yes, thanks.
5. The description of America/Indiana/Marengo is that it covers the eastern part of Crawford County. How is that county divided, and in which zone does the western part fall?
6. What part of Kentucky goes in America/Kentucky/Louisville?
For both questions, I haven't a clue, but you may be able to tell by looking at old DOT records.
My page at http://www.statoids.com/tus.html has a good deal of detail about time zone changes over the years in the United States. I think you'll be able to find the answers to some of your questions there. I've answered #2 because I found the information right away. Most of my description of time zone boundaries comes from published maps, usually state road maps from Rand McNally, the AAA, Gousha, etc. They're pretty reliable, but they can't be considered primary sources. In the list of changes following the table on my webpage, the changes with exact dates often come from primary sources (DOT docket items) or widely disseminated news reports. Changes since about 1988 can usually be confirmed by examining the tz mailing list archives. -- Gwillim Law ----- Original Message ----- From: "Eric Muller" <emuller@adobe.com>
2. The description of America/Denver includes "far eastern Oregon", and the description of America/Boise includes "eastern Oregon". Comparing with timeznp020, I suspect that America/Denver should not include "far eastern Oregon". (and if timeznp020 is right, "eastern Oregon" could be better described as "most of Malheur County, Oregon").
Eastern Oregon is definitely equivalent to "most of Malheur County", although if the road maps are correct, the exact boundaries underwent a small change around 1970. The change affected only sparsely populated areas. As the tz comments say, southern Idaho and eastern Oregon delayed going on DST for four weeks in 1974. Eastern Oregon belongs in America/Boise, because it remained in synch with Boise rather than Denver during that period.
Eric, I can not help you with the time zone boundaries within the Unites States. However, I wanted to let you know that I created a KML time zone map for Canada. It may give you some ideas of what you can do with KML. The file is in KMZ (compressed KML) format; it is 682kB in size and should open in in Google Earth. Use the radio buttons to select between "winter", "summer", and "outlines" views. In the "outlines" view, hover your mouse above the center of an outline to highlight it and make a name appear. Click in the center of a highlighted outline to view more information. http://www3.sympatico.ca/c.walton/canada.kmz Also, you should have a look at Jesper's WTE program. Jesper has put a great deal of effort into making his maps as accurate as possible. http://www.worldtimeexplorer.com/ By the way, the links you provided for the US time zone areas <http://efele.net/timeznp020.zip> seem to have lots of points with many "significant digits" giving the illusion of accuracy. Unfortunately your data shows many sections of US/Canada border to be about 200 meters north of where it is supposed to be. Is this a sign that the US is trying to encroach on Canada? Should the Canadian military be put on alert? I should also mention that the Google Earth yellow borders should not be used as references; they are notoriously inaccurate; I usually turn them off. Jesper, If you are reading this and you get a chance to open my latest canada.kmz file, you will notice that there are lots of new islands that were not shown last time I sent you a copy of the file. -chris
-----Original Message----- From: Eric Muller [mailto:emuller@adobe.com] Sent: May 31, 2007 1:55 AM To: tz@lecserver.nci.nih.gov Subject: Geographical boundaries for the continental US tz zones I am trying to figure out the geographical extent of the 19 continental US timezones defined by tz. Because my goal is ultimately to create a freely >available digital map, I would like to get definitions in terms existing, freely available digital maps, and the ones at nationalatlas.gov seem >appropriate. Specifically, I plan to use countyp020 (county boundaries 2001), statesp020 (states) and timeznp020 (time zones) [see below for more details on those maps]. I tried to translate the comments in the "northamerica" file in terms of operations on the elements of those three maps, but I have a few questions/comments.
Chris Walton wrote:
Eric, I can not help you with the time zone boundaries within the Unites States.
However, I wanted to let you know that I created a KML time zone map for Canada.
Wonderful. Thanks.
By the way, the links you provided for the US time zone areas <http://efele.net/timeznp020.zip> seem to have lots of points with many "significant digits" giving the illusion of accuracy.
That only reflects that the software I used has "printf ("%f")" somewhere. What matters more is the stated accuracy of the map (<http://nationalatlas.gov/metadata/timeznp020.faq.html#what.6>): Horizontal positions are specified in geographic coordinates, that is, latitude and longitude. Latitudes are given to the nearest 0.000278. Longitudes are given to the nearest 0.000278. Latitude and longitude values are specified in Decimal degrees.
Unfortunately your data shows many sections of US/Canada border to be about 200 meters north of where it is supposed to be.
That's in reaction to your map going about 200 meters south of where it is supposed to be in some areas 8-) Eric.
From: Eric Muller Date: Wed, 30 May 2007 22:55:25 -0700 Subject: Geographical boundaries for the continental US tz zones
I am trying to figure out the geographical extent of the 19 continental US timezones defined by tz. Because my goal is ultimately to create a freely available digital map, I would like to get definitions in terms existing, freely available digital maps, and the ones at nationalatlas.gov seem appropriate. Specifically, I plan to use countyp020 (county boundaries 2001), statesp020 (states) and timeznp020 (time zones) [see below for more details on those maps]. I tried to translate the comments in the "northamerica" file in terms of operations on the elements of those three maps, but I have a few questions/comments.
The boundaries of the US timezones are precisely defined by the DOT. One of the problems is that the boundary descriptions are laid down in cadastral coordinates. These can in some cases only with great difficulty be translated into latitude/longitude. More information about the boundaries of USA, Canada and other countries on my Multizones web page. http://home-4.tiscali.nl/~t876506/Multizones.html
Many thanks for the answers. I am slowly making my way toward my goal. Will send an update when I have more concrete results. Eric.
I have completed a first version of a map of the TZ timezones in the US. It is available as a shapefile at <http://www.efele.net/maps>. Comments welcome, Eric.
Eric Muller <emuller@adobe.com> writes:
I have completed a first version of a map of the TZ timezones in the US. It is available as a shapefile at <http://www.efele.net/maps>.
Thanks. Does the following HTML summarize it accurately? <li><a href="http://www.efele.net/maps/">A map of the TZ timezones in the US</a> contains a <a href="http://en.wikipedia.org/wiki/Shapefile">shapefile</a> of the <code>tz</code> regions in the US.</li>
participants (6)
-
Chris Walton -
Eric Muller -
Gwillim Law -
Oscar van Vlijmen -
Paul Eggert -
Robert Elz