Disclaimer, methodology and license

Italia Rovente's code is public on GitHub under the MIT license: you may inspect, use, modify and redistribute it freely, provided the copyright notice is retained. The software is provided without warranties.

The weather data is not covered by the MIT license: it belongs to Open-Meteo, based on the ERA5 reanalysis by ECMWF / Copernicus Climate Change Service, and is subject to their respective license and attribution terms.

The site only collects anonymous, aggregated counts (visits and votes), with no personal data, accounts or advertising tracking.

Not just people: search engines and AI assistants come to read this data every day too — we allowed them on purpose (see /llms.txt). That OpenAI, Anthropic and Google come here to fetch it says the essential thing: the numbers are treated as a source to consult, not an opinion. An anonymous per-category count — only the request's User-Agent, no personal data:

• ChatGPT / GPTBot: 2,294 requests
• Claude: 2,732 requests
• Perplexity: 57 requests
• Google: 21,314 requests
• Bing / Copilot: 255 requests
• Meta: 1,222 requests
• Other crawlers: 12,784 requests

Trend

One snapshot a day: human visits vs. total bot requests.

Last 7 days

How much has Italy warmed since 1940?

Comparing the 1991–2020 climate normal with the 1961–1990 one, the cities monitored by this site show an average warming of +1.0°C. That's the same method — the difference between two 30-year averages — used identically across every page of the site.

Why is your figure lower than the ones I read elsewhere (Copernicus, ISPRA)?

Because it measures something specific, and different. The figure we show is the gap between two 30-year climate normals: 1991–2020 minus 1961–1990. It is not the warming since the pre-industrial era (1850–1900), which our series does not cover: ERA5 starts in 1940, and the 1961–1990 window was already warmer than pre-industrial — so against that reference the total is larger. On our own data, the linear trend over the whole series since 1940 is higher than the gap between the two normals, precisely because it also includes the warming before 1961 and after 2020. A figure published by Copernicus for Europe or ISPRA for Italy can therefore be much higher without contradicting ours: the baseline, the period and often the territory are different. Comparing two numbers means first checking they measure the same thing; we use the two-normals comparison because it is the World Meteorological Organization standard and the most robust to year-to-year noise, not because it yields the smaller number.

What data does Italia Rovente use?

The ERA5 reanalysis by ECMWF/Copernicus (C3S), distributed by Open-Meteo: air temperature at 2 meters above ground since 1940, updated automatically every day.

How many cities does the site cover?

107 Italian cities, from the main ones (with precise max/min data) to provincial capitals. Full list at italiarovente.app/en/citta.

Why do the 12 main cities show two slightly different warming numbers?

On the 12 main cities we show a second comparison, computed with the same method (1991–2020 normal minus 1961–1990 normal) but on an independent source: E-OBS, built by the ECA&D/Copernicus project solely from direct European weather-station observations, not from a model like ERA5. The two sources use different methods, resolutions, and station networks, so a difference of a few tenths of a degree between the two numbers is normal, not an error — what matters is that neither contradicts the other on the direction of warming. E-OBS data: EU-FP6 project UERRA and the Copernicus Climate Change Service, provided by the ECA&D network (ecad.eu); source: Cornes et al. 2018, An Ensemble Version of the E-OBS Temperature and Precipitation Datasets. Non-commercial use, per the E-OBS license.

Is this an official weather service?

No. It is informational and educational: it does not replace official weather services or ISPRA data, and should not be used for critical decisions (safety, agriculture, navigation).

How is a city's warming calculated?

As the difference between the 1991–2020 climate normal and the 1961–1990 one (the WMO reference period): a comparison between two 30-year averages, not between a recent year and the past, because it's the most robust and least noisy measure.

Can I cite the data from this site?

Yes, with attribution to Italia Rovente and the data source (Open-Meteo / ERA5 / ECMWF-Copernicus). The site's code is open source under the MIT license.

Why doesn't the site cover ice ages or much older climate?

Because we use direct instrumental data (the ERA5 reanalysis), reliable only from 1940 onward. Older periods require paleoclimatology, which reconstructs climate through indirect proxies — ice cores, tree rings, sediments — with uncertainty that grows the further back in time you go. A different method, same principle: science is built on verifiable facts and data. Everything else — alarmism or minimization — is narrative, and we don't show any of it here.

Which Italian cities have the most extreme summer heat days ("African heat")?

The up-to-date ranking is at italiarovente.app/en/classifiche, "The most scorching days" section: days with a high ≥30°, averaged over the last 5 full years for each monitored city, refreshed automatically — we don't quote a fixed number here because it would change and quickly go stale.