Methodology

Data

Daily weather comes from the Open-Meteo Historical Weather API, which is based on ERA5 reanalysis. It is modelled data on a grid of roughly 10 to 25 km, not a single weather station, so very local effects such as a coastal breeze or a hilltop are smoothed out.

Comfort score

Each month and week starts at 100 points. We subtract 4 points for every degree the average feels-like temperature is outside 18 to 26 °C (up to 60), 1.5 points per rainy day (up to 25), 0.5 points per muggy day (up to 15) and 0.4 points per windy day (up to 10). A score of 70 or more is what we call comfortable.

Definitions

A rainy day has at least 1 mm of precipitation. A muggy day has a dew point of 18 °C or more, which we work out from temperature and humidity. A windy day has a top wind speed of 30 km/h or more. Daylight is calculated from the latitude, not measured. Reanalysis data tends to count more light-rain days than a rain gauge would, so rainfall totals are more reliable than rainy-day counts.

How the sentences are written

The text is generated by fixed rules, not by an AI model. We compare yearly averages, find peak months, count the months each city leads, and detect seasons where the lead changes. Small differences are reported as nearly identical.

Limits

Three years is a short record. Use these comparisons to get a feel for each place, and check a longer climate record before planning anything weather-critical.