Daily temperature vs.

Pick a city and year. Each column is one day's high or low minus NOAA's 1991–2020 normal for that date at the same weather station.

Historical Trends
Show list of cities

Data: NOAA daily observations and 1991–2020 daily normals, fetched live from the ACIS web service (data.rcc-acis.org). Each city uses its official airport (or Central Park for New York) climate station. The normals are the same 1991–2020 baseline for every year shown. Data for the last few days is preliminary.

I would prefer not to use the normal because of the ongoing secular warming trend in the data, but the app is using NOAA data and that is the word they use. Here is more .

There are two kinds of average in the app.

1. The “normal” for each day (the baseline every column is measured against)

The app doesn’t calculate these itself. It downloads NOAA’s official 1991–2020 daily normals for each weather station, published by the National Centers for Environmental Information. Roughly, NOAA makes them like this:

Averaging each date directly, say all 30 March 19ths, would give a jagged line, because one freak warm day moves that date’s average by a degree (1°F, about 0.6°C). The curve instead rises and falls smoothly through the year. For example, Denver’s normal high goes 44°F, 44°F, 44°F (6.7°C)… in early January and creeps up a degree Fahrenheit (about 0.6°C) every few days.

A few things follow from this:

2. The averages in the summary boxes (“Highs vs. normal +5.2°F” or +2.9°C)

The app calculates these: it adds up each day’s difference (actual minus normal) and divides by the number of days with data. Days with a missing reading are left out, so the box also shows the day count.