The tz_world shapefile (14 Mb) captures the boundaries of the TZ timezones across the world.
The geometries are primarily derived from the fip10s data, augmented with data presented in the pages for the maps of the US, Canada and China.
This is a snaphsot of the zones (many of the smaller zones are not visible at this scale):
The vast majority of the tz timezones are by construction matching a country, so little needs to be said about them: we equate them with the corresponding region in our fips10s data. We also discussed the situation of the US, Canada and China in other pages. That leaves only a few countries with multiple time zones, and again, most of them at straightforward, with timezone matching administrative divisions of the countries: the correspondance is documented in the script we use to build the shapefile (see below). The rest of this section discusses the remaining cases, where the definition of the extent of a zone is open to interpretation.
This country is covered by two tz timezones, Tashkent and Samarkand. There is not enough information in the tz data to figure out how Uzbekistan is divided between them. We use a separation along first-level administrative divisons that matches WTE.
This country is covered by four tz timezones. There is not enough information in the tz data to figure out how Ukraine is divided between them. We use a separation along first-level administrative divisons that matches WTE.
The extent of the Eucla zone is not clear from the tz data. The only clue is that the timezone ends just east of Caiguna. We arbitrarily make that zone the intersection of a rectangle and Western Australia: Caiguna is a 125.490E, so we use somewhat arbitrarily 125.5E as the west boundary; the north boundary is arbitrarily -31.3S; the south boundary is the ocean; the east boundary is the Western Australia/South Australia border.
The Lindeman zone is said by tz to include the three islands of Hayman, Lindeman, Hamilton, all in Queensland. There are other islands in this area, but we don't include them, for lack of better information.
The Broken_Hill timezone is defined by law as the county of Yancowinna, New South Wales. The geometry was obtained by georeferencing and tracing of this image of the counties of New South Wales
It is difficult to obtain reliable and verifiable information on time zones in Mongolia: see this summary. We use the following setting:
According to this report,
Regardless of what Cuba does, the U.S. Naval Base Guantanamo Bay will always be in the Eastern Time zone and will always observe daylight saving time as this allows the naval base to remain in synch with their headquarters in Florida. This information was confirmed with the assistance of the Public Affairs Officer at the U.S. Naval Base Guantanamo Bay.
Accordingly, we assign it the America/New_York timezone.
The Kwajalein zone is certainly no less than the 11 islands leased by the US on the Kwajalein atoll. It’s also very unlikely to be more than the Kwajalein atoll itself. We use the whole atoll as the extent of the zone.
Most of Spain is easy. The open question concerns the Plazas de soberania. Clearly, Ceuta is in Africa/Ceuta, but the timezone of Melilla, Chafarinas, Peñón de Alhucemas, Peñón de Vélez de la Gomera and Isla de Alborán is unclear; we use Europe/Madrid for those.
Except for the Sakha (Yakutiya) republic of Russia which covers three timezones, all the other subdivisions of Russia are in exactly one zone, and cause no particular problem.
In Sakha, the tz documentations puts the Srednekolymskiy and Verkhnekolymskiy rayons in Asia/Yakutsk, but since they are at the eastern end of Sakha, we place them in the Asia/Magadan. tz does not mention the Neryungri nor he Yakutsk gorsovets; based on their location, we place them in Asia/Yakutsk. Finally, we place the artic islands north of Anabarskiy in Asia/Yakutsk and those north of Ust-Yanskiy in Asia/Vladivostok. The result after those corrections to the tz description of the zones matches this map.
We obtain the timezone boundaries within Sakha from the map created by Alexander Perepechko and Dmitry Sharkov and offered by WAGDA.
The zones of those areas are not handled in our map. Suggestion on the best approach (very localized zones vs. complete partition of the territories) are welcome.
In our shapefile, we have a single attribute tzid. In addition to the names of tz timezones, we use these two values: uninhabited when tz has identified an area as such and has decided to not assign a timezone to it; and none for Greenland and Antartica.
For Canada, we use the same technique as in building the Canada map; in fact, we simply follow the same process but include the rest of the world as well.
The same process is used for the US, China, Russia and Brazil, i.e. we have masks we uses to bring in the timezone boundaries that are not already present as subdivision boundaries in fips10s.
The US and China masks have been obtained by simply extending the outside boundaries from our US and China maps.
The Russia mask has been obtained from the Perepchko and Sharkov map mentioned above.
The Brasil mask has been obtained by using 1) the great circle from Tabatinga to Porto Acre to divide east from west Amazonas, and 2) the Xigu and Javary rivers from the VMAP0 inwatera (inland water area) layer to divide east from west Para.
The collection of masks is available here (500 Kb), along with the script that build the resulting map.
History:
