Skip to main content
Warning: You are using the test version of PyPI. This is a pre-production deployment of Warehouse. Changes made here affect the production instance of TestPyPI (
Help us improve Python packaging - Donate today!

Python library to look up timezone from lat / long offline. Improved version of "pytzwhere".

Project Description

This is a fast and lightweight python project to lookup the corresponding timezone for a given lat/lng on earth entirely offline.

This project is derived from and has been successfully tested against pytzwhere (github), but aims to provide improved performance and usability.

It is also similar to django-geo-timezones

The underlying timezone data is based on work done by Eric Muller.

Timezones at sea and Antarctica are not yet supported (because somewhat special rules apply there).

[Please note: At the moment Python 2.x is not supported (Tests are not running and algorithms with numpy are not
compatible). Write me to encourage me to work on this.]


(python, math, struct, os)


maybe also numba and its Requirements

This is only for precompiling the time critical algorithms. When you only look up a few points once in a while, the compilation time is probably outweighing the benefits. When using certain_timezone_at() and especially closest_timezone_at() however, I highly recommend using numba (see speed comparison below)! The amount of shortcuts used in the .bin are also only optimized for the use with numba.


(install the dependencies)

in your terminal simply:

pip install timezonefinder

(you might need to run this command as administrator, because some other dependencies will be installed.)



from timezonefinder import TimezoneFinder

tf = TimezoneFinder()

for testing if numba is being used: (if the import of the optimized algorithms worked)

# this is a static method returning True or False

fast algorithm:

# point = (longitude, latitude)
point = (13.358, 52.5061)
print( tf.timezone_at(*point) )
# = Europe/Berlin

To make sure a point is really inside a timezone (slower):

print( tf.certain_timezone_at(*point) )
# = Europe/Berlin

To find the closest timezone (slow):

# only use this when the point is not inside a polygon!
# this only checks the polygons in the surrounding shortcuts (not all polygons)

point = (12.773955, 55.578595)
print( tf.closest_timezone_at(*point) )
# = Europe/Copenhagens

To increase search radius even more (very slow, use ``numba``!):

# this checks all the polygons within +-3 degree lng and +-3 degree lat
# I recommend only slowly increasing the search radius
# keep in mind that x degrees lat are not the same distance apart than x degree lng!
print( tf.closest_timezone_at(lng=point[0],lat=point[1],delta_degree=3) )
# = Europe/Copenhagens

(to make sure you really got the closest timezone increase the search radius until you get a result. then increase the radius once more and take this result.)

Further application:

To maximize the chances of getting a result in a Django view it might look like:

def find_timezone(request, lat, lng):
    lat = float(lat)
    lng = float(lng)

        timezone_name = tf.timezone_at(lng, lat)
        if timezone_name is None:
            timezone_name = tf.closest_timezone_at(lng, lat)
            # maybe even increase the search radius when it is still None

    except ValueError:
        # the coordinates were out of bounds
        # {handle error}

    # ... do something with timezone_name ...

To get an aware datetime object from the timezone name:

# first pip install pytz
from pytz import timezone, utc
from pytz.exceptions import UnknownTimeZoneError

# tzinfo has to be None (means naive)
naive_datetime = YOUR_NAIVE_DATETIME

    tz = timezone(timezone_name)
    aware_datetime = naive_datetime.replace(tzinfo=tz)
    aware_datetime_in_utc = aware_datetime.astimezone(utc)

    naive_datetime_as_utc_converted_to_tz = tz.localize(naive_datetime)

except UnknownTimeZoneError:
    # ... handle the error ...

also see the pytz Doc.

Using the conversion tool:

Place the in one folder with the tz_world.csv from tzwhere and run it as a script. It converts the .csv in a new .csv and transforms this file into the needed .bin

Place this .bin in your timezonfinder folder (overwriting the old file) to make it being used.

Please note: Neither the tests nor the are optimized or really beautiful. Sorry for that.

Comparison to pytzwhere

In comparison to pytzwhere I managed to speed up the queries by more than 100 times (s. test results below). Initialisation time and memory usage are also significanlty reduced, while my algorithm yields the same results. In some cases pytzwhere even does not find anything and timezonefinder does, for example when only one timezone is close to the point.


  • results
  • data being used


  • the data is now stored in a memory friendly 35MB .bin and needed data is directly being read on the fly (instead of reading and converting the 76MB .csv (mostly floats stored as strings!) into memory every time a class is created).
  • precomputed shortcuts are stored in the .bin to quickly look up which polygons have to be checked (instead of creating the shortcuts on every startup)
  • optimized algorithms
  • introduced proximity algorithm
  • use of numba for speeding things up much further.

Excerpt from my test results*:

testing 1000 realistic points
testing 10000 random points
in 11000 tries 0 mismatches were made
fail percentage is: 0.0

TIMES for 1000 realistic queries***:
pytzwhere:  0:00:18.184299
timezonefinder:  0:00:00.126715
143.51 times faster

TIMES for  10000 random queries****:
pytzwhere: 0:01:36.431927
timezonefinder: 0:00:00.626145
154.01 times faster

Startup times:
pytzwhere: 0:00:09.531322
timezonefinder: 0:00:00.000361
26402.55 times faster

*timezone_at() with numba active

**mismatch: pytzwhere finds something and then timezonefinder finds something else

***realistic queries: just points within a timezone (= pytzwhere yields result)

****random queries: random points on earth

Speed Impact of Numba

TIMES for 1000 realistic queries***:

wo/ numa: 0:00:01.017575
w/ numa: 0:00:00.289854
3.51 times faster

wo/ numa:   0:00:05.445209
w/ numa: 0:00:00.290441
14.92 times faster

wo/ numa: 0:02:32.666238
w/ numa: 0:00:02.688353
40.2 times faster

(this is not inlcuded in my tests because one cannot automatically enable and disable Numba)

Known Issues

All points in Lesotho are counted to the ‘Africa/Johannesburg’ timezone instead of ‘Africa/Maseru’. I am pretty sure this is because it is completely surrounded by South Africa and in the data the area of Lesotho is not excluded from this timezone. So actually this is a mistake in the data not my algorithms and the consequences are too small for me to fix this issue (those two timezones have the same utc-offset anyway).

Again: write if you want me to work on this.


If you notice that the tz data is outdated, encounter any bugs, have suggestions, criticism, etc. feel free to open an Issue on Git or contact me: python at michelfe dot it


timezonefinder is distributed under the terms of the MIT license (see LICENSE.txt).


Note: not mentioned versions only contain small and irrelevant changes (e.g. in the readme,…). I am new to all this, so I am often missing small things which are not really new features worth mentioning.

1.5.0 (2016-04-12)

  • automatically using optimized algorithms now (when numba is installed)
  • added TimezoneFinder.using_numba() function to check if the import worked

1.4.0 (2016-04-07)

  • Added the to the repository: It converts the .csv from pytzwhere to another .csv and this one into the used .bin.
    Especially the shortcut computation and the boundary storage in there save a lot of reading and computation time, when deciding which timezone the coordinates are in. It will help to keep the package up to date, even when the timezone data should change in the future.
Release History

Release History

This version
History Node


History Node


Download Files

Download Files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
timezonefinder-1.5.0.tar.gz (30.1 MB) Copy SHA256 Checksum SHA256 Source Apr 12, 2016

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting