Our Weather Data and Methodology: Sources and Processing
Every number on this site is computed, not copied. This page identifies the source, period, aggregation and limits behind each figure so you can judge whether it fits your use.
Climate averages (the historical layer)
One fixed period makes every city comparable
Each city page shows monthly climate averages built from ten full years (2015–2024) of daily weather history:
- Source: the Copernicus ERA5 reanalysis dataset, accessed through the Open-Meteo Historical Weather API.
- Resolution: daily highs, lows, precipitation, snowfall, sunshine and humidity for each city's exact coordinates.
- Aggregation: roughly 3,650 daily records per city are averaged into monthly averages.
- What "record high" means here: the highest and lowest daily values in that ten-year window, at that grid point - not an all-time station record. ERA5 is a gridded reanalysis, so its values are smoothed across a cell and can sit a few degrees below a nearby weather station's official extreme. We label these as decade extremes everywhere they appear, because calling them records would overstate them.
- Climate types follow a simplified Köppen classification computed from those averages.
Because the window is fixed (2015–2024), the averages are stable and comparable across every city we cover - they don't drift with each site update. That comparability is what makes the "how this city compares" block on each climate page possible: when we say a city is drier than 94% of our dataset, every city in that comparison was measured the same way, over the same years.
Live conditions and forecasts
- Source: WeatherAPI.com (current conditions, 48-hour hourly, 14-day daily, air quality, alerts, astronomy).
- Refresh: each city's bundle stays cached for up to 8 hours, then the next visit requests a refresh.
- Freshness labels: every live page shows when its data was last updated; if the upstream feed is briefly unavailable, we show the last good observation and say so.
Rankings
The hottest and coldest lists compare the most recent cached observation for every covered city - a live snapshot, not a climate ranking. No modeled or interpolated values are mixed in.
Editorial content
City and country descriptions are computed from each location's own climate figures by our own software - not written by hand per city, and not produced by a language model. The hottest month named in a city's text is that city's actual hottest month in the data, and every comparison is a real ranking against the rest of the dataset.
Which facts a page leads with depends on what is unusual about that place: a city whose rain falls in three months is described through its wet season, one with a wide annual swing through its seasons, one that barely changes through what changes instead. Two cities read alike only when their climates genuinely are alike.
Guides are handled differently from measured city pages. They are maintained as structured editorial records against the primary sources listed at the bottom of each guide (NWS, USDA, FDA, EPA and peers), with a visible content-check date. Automation may assist drafting and quality control, but it is not evidence and is never presented as a credential. Our editorial policy explains the release checks and specialist-review limits.
Blog publication controls
Long-form blog submissions are kept separate from the city dataset and are not published merely because they arrive. A submission must stay within temperature, weather or climate, provide a substantial structured explanation, and not substantially duplicate an article already on the site. Imported HTML is sanitised, and external links in imported articles are qualified so the blog cannot be used as a followed-link placement channel.
Corrections
Spotted something off? Tell us - we check reports against the source data and update the page when a correction is verified.