City by Month

Methodology

This site is built and run by AI agents (Anthropic's Claude) for a human owner. Programs assemble each page from a fact record. No fact comes from a language model.

Last updated : first sourced version.

Where the facts come from

The 11 sources: what each is, who publishes it, its licence and the date it was last read
WhatPublisherLicenceChecked
Wikidata Query ServiceWikimedia FoundationCC0 1.0
GeoNames dump (cities15000, countryInfo)GeoNamesCC BY 4.0
OurAirports airports.csvOurAirports (David Megginson)Public domain
WMO Climatological Standard Normals for 1991-2020 (NCEI Accession 0253808)WMO Member Nations (each national weather service's own submission); archived by NOAA National Centers for Environmental InformationNo licence stated. Published by NOAA NCEI for WMO; citation requested,
U.S. Climate Normals 2020: U.S. Monthly Climate Normals (1991-2020)NOAA National Centers for Environmental InformationNo licence stated; citation requested
Hong Kong Observatory daily climate series on DATA.GOV.HKHong Kong Observatory, Government of the Hong Kong Special Administrative Region; published on DATA.GOV.HKTerms and Conditions of Use of DATA.GOV.HK, version 1.2
National Environment Agency monthly climate series on data.gov.sgNational Environment Agency, Singapore; published on data.gov.sgSingapore Open Data Licence version 1.0
NOAA Solar Calculator equations, computed locallyNOAA Global Monitoring Laboratory (method); City by Month (computation)Method, not data
IANA Time Zone Database, read through the PyPI tzdata packageIANAPublic domain
holidays Python library, run locallyVacanza teamMIT
Official sourcesThe statistics office, government body or law named beside each figureFacts and figures, each linked to its source page,

How a page is made

  1. A program reads each source and writes one fact record for each city.
  2. Some values are entered by hand from the source named beside them: population, area, languages, the airport list, the choice of weather station and the holiday checks.
  3. Each value in the record carries its source, the source address and the date it was read.
  4. A second program fills fixed sentences and tables with the record's values.
  5. Where the record has no value, the page says "no data" and gives the record's note.
  6. Temperatures are in °C and rain is in mm. °F and inches are converted and marked.
  7. Ranks, conversions and daylight are computed. Each shows its rule.

The rule for each fact

Country
The country's short English name. Three use a short form of the UN's name: Türkiye, Netherlands, and Hong Kong SAR, China.
Population and area
Entered by hand from the statistics office named beside each figure. Each figure says which place it covers and its date. We print no density.
Coordinates
One point for the city centre, from Wikidata; from GeoNames where Wikidata gives more than one.
Time zone
The IANA time zone name, from GeoNames.
Languages
From the law or government page named beside them, with the status that source gives: official, national or de facto.
Currency
Code and name from the ISO 4217 list.
Airports
Listed by hand: passenger airports with scheduled flights within 70 km of the centre point, or further where named. Name, type and position come from OurAirports. The distance is ours: the shortest line over a sphere of radius 6,371.0088 km, from the centre point to the airport's point, rounded to the nearest km.
Climate
1991-2020 normals as each national weather service filed them with the World Meteorological Organization, read from the copy at NOAA. For New York and Los Angeles: NOAA's own U.S. normals, converted from °F and inches. For Hong Kong and Singapore: our own 1991-2020 mean of the government's open data series for the station, not the weather service's normals; a measure is printed only if all 360 months of the series are complete. One station per city: the service's main station inside the built-up city, 30 km from the centre point at most. A measure with a missing month is no data. A city with no such station, or whose weather service's reuse terms are not settled, has no climate data and no month pages.
Station note
A station more than 5 km from the centre point, or more than 50 m higher or lower, is named with its distance and direction. The distance is measured as for airports, to 0.1 km.
Daylight
NOAA Solar Calculator equations (Meeus, Astronomical Algorithms); zenith 90.833 deg; two passes; sea-level horizon; rounded to the minute. Computed for the 15th of each month, 2026, from the city's coordinates. Local time from the tz database, version 2026d. Day length is shown in hours and minutes: minutes ÷ 60.
Public holidays
From the holidays library; the page gives its version. A year's list is printed only after we matched it, date by date, to the government's own list, which the page links. Names follow the government's English list where it differs. Where the library's list is wrong, we copy the government's list row by row and check it a second time against that list; the page then names that list as its source. Otherwise: no data.

How we compute

°F and inches
Figures in brackets are converted: °F = °C × 9 ÷ 5 + 32, to 1 decimal; inches = mm ÷ 25.4, to 2 decimals.
Warmest, coldest and wettest month
Warmest, coldest, wettest, most and least daylight: the highest or lowest of the 12 monthly values. Months with the same value are all named and marked equal.
This month against the year
Rank = 1 + the number of months with a larger value, or with a smaller value from the low end. The nearer end is shown. Months with the same value share a rank and are marked equal.
Public holidays: library and region
Library: holidays 0.105. Region: the city's state or province, where the library has one.
Month pages
A city has month pages only when its record has a daily high and a daily low for all 12 months.

Sources

Cite this page: City by Month. "Methodology: sources, rules and how AI is used". https://citybymonth.com/methodology/. Last updated 2 October 2026.