NCEI United States Daily Record Count Archive

Updated AUG 2022/FEB 2024/JAN 2025/JAN 2026/SEP 2026

The purpose of this post is merely to catalogue counts of daily record high maximums, high minimums, low minimums and low maximums coming into the National Center for Environmental Information’s site and all related charts and graphs produced in my Excel files for those data sets for the United States. I am in the process of constantly updating this data verifying the 2009 Meehl et. all surface Records published in geophysical Science that I initiated from the year 2000. Each individual count could be a tied surface record or one broken by several degrees Fahrenheit.

Here is the link to the NCEI site that I glean data from:

https://www.ncdc.noaa.gov/cdo-web/datatools/records

More from NCEI:

“The daily records summarized here are compiled from a subset of stations in the Global Historical Climatological Network. A station is defined as the complete daily weather records at a particular location, having a unique identifier in the GHCN-Daily dataset.

For a station to be considered for any parameter, it must have a minimum of 30 years of data with more than 182 days complete each year. This is effectively a “30-year record of service” requirement, but allows for inclusion of some stations which routinely shut down during certain seasons. Small station moves, such as a move from one property to an adjacent property, may occur within a station history. However, larger moves, such as a station moving from downtown to the city airport, generally result in the commissioning of a new station identifier. This tool treats each of these histories as a different station. In this way, it does not “thread” the separate histories into one record for a city.

This tool provides simplistic counts of records to provide insight into recent climate behavior, but is not a definitive way to identify trends in the number of records set over time. This is particularly true outside the United States, where the number of records may be strongly influenced by station density from country to country and from year to year. These data are raw and have not been assessed for the effects of changing station instrumentation and time of observation.”

An updated 2016 study from Dr. Jerry Meehl indicates that the ratio from year to year will average around 15 to 1 by 2100:

https://www.usatoday.com/story/weather/2016/11/21/us-record-high-temperatures-overwhelm-record-lows/94234824/

Per one of the authors of both the 2009 and 2016 studies, Claudia Tebaldi said “This climate is on a trajectory that goes somewhere we’ve never been. And records are a very easy measure of that.”

All of the data listed below is part of this one chart. Prior to 1890 there were very few reports of records, so I have opted not to catalog data prior to 1880. The ratio of daily record high maximums to low minimums for the 2020s (so far) is higher than any decade since the 1880s:

Here are the current daily record counts per decade, which have gone into the prior chart:

We now have another five years of completed data as of DEC 2025. Here is what we have at five-year increments:

The above chart can also be broken down to yearly increments:

*I have come up with new programmed files such that daily record counts for Alaska are subtracted from NCEI values for the United States as a whole. This will enable us to better compare apples to apples values for the lower 48 states and NCEI average temperature rankings for the lower 48 states since 1895. These new record scoreboards and charts should be better, but they aren’t perfect since Hawaii and other U.S. tallies from all territories are still included in total counts. However, since the bulk of reporting stations outside of the lower 48 do come from Alaska, this is a good first fix.*

Summary of the U.S. daily and monthly record findings

The central result is a strong, nonlinear relationship between temperature anomaly and the balance of warm versus cold records. As monthly temperatures move from strongly negative anomalies toward strongly positive anomalies, the record distribution shifts decisively from cold records to warm records.

In the daily DHMX/DLMN analysis shown here—daily highest maximums versus daily lowest minimums—the transition is especially clear. Cold months strongly favor DLMN records. Near the climatological average, the two record types approach parity, and once anomalies become positive, DHMX records increasingly dominate.

My table illustrates the progression particularly well:

Approx. anomaly	DHMX/DLMN ratio
−2.9°F	0.386
−1.9°F	0.396
−0.9°F	0.611
−0.5°F	0.755
−0.1°F	0.962
+0.1°F	1.166
+0.7°F	1.583
+1.3°F	1.929
+2.1°F	2.955
+2.5°F	4.461
+3.3°F	4.979
+4.1°F	9.160
+5.1°F	13.211
+6.5°F	33.402

So the important result isn’t merely that warmer months produce more warm records. The imbalance accelerates toward the anomaly extremes. Around normal temperatures, warm and cold records remain relatively competitive; several degrees above normal, cold records become increasingly scarce while warm records proliferate.

The reverse occurs on the cold side. For example, at approximately −4.3°F, the ratio is only 0.328, meaning DLMN records outnumber DHMX records by roughly 3:1. At −5.9°F the ratio falls to 0.084, or nearly 12 cold records for every warm record.

What the four principal datasets told us

Our recent work wasn’t limited to DHMX/DLMN. We compared the complementary daily and monthly record pairs:

DHMX vs. DLMN — daily record high maximums versus record low minimums.
DHMN vs. DLMX — daily record high minimums versus record low maximums.
MHMX vs. MLMN — monthly record high maximums versus monthly record low minimums.
MHMN vs. MLMX — monthly record high minimums versus monthly record low maximums.

All four produced the same broad physical signal: negative temperature anomalies favor cold records, positive anomalies favor warm records, and the warm/cold ratio becomes increasingly extreme as the magnitude of the anomaly increases.

That agreement is scientifically important. We’re not seeing a relationship peculiar to afternoon maximum temperatures or to one definition of a record. It appears in maximums and minimums and in daily and monthly record behavior.

The monthly datasets are particularly useful as a complementary test because monthly records require warmth or cold to persist sufficiently to affect a month’s extreme statistics. Daily records can result from a short-lived weather event; monthly records impose a different threshold. Yet the same anomaly dependence emerged.

The larger finding

I think the most interesting conclusion from this work is that temperature anomaly acts almost like a control variable for record occurrence.

Near climatological normal, the record system remains comparatively balanced. Move sufficiently to the cold side and cold records rapidly dominate. Move sufficiently to the warm side and warm records rapidly dominate. The farther the climate state moves from normal, the more asymmetric the record distribution becomes.

And because the climate has shifted toward positive anomalies over time, the United States now spends substantially more time in the portion of this relationship where the mathematical odds strongly favor warm records.

That distinction matters: the underlying relationship between anomaly and record production is weather/climate sensitivity; the increasing frequency with which the country occupies the warm side of that relationship is the climate-change signal.

The work therefore provides a useful bridge between two ways of describing climate change: mean temperature anomalies and extreme-temperature records. They aren’t independent phenomena. Our analysis shows quantitatively how movement in the mean climate state translates into an increasingly lopsided distribution of extremes.

The 2020s:

For this data set all monthly ratios of > 10 to 1 DHMX to DLMN or > 10 to 1 DLMN to DHMX are in bold type. Rankings are for the lower 48 states with the warmest ranking since 1895 of average temperatures being 132 and 1 being the coldest as of 2026. Blue colors represent cold months and red warm. Those months with counts close to a 1 to 1 ratio of highs to lows are colored black. All-time record hottest or coldest months and years are boldly colored in purple.

NCDC rankings have been color coded (under tabs in each file) such that values of 55 to 75 are black representing neutral months or years (+ or – 10 from the average ranking of 66). Values below 54 are color coded blue and values above 75 are color coded red.

Time stamps for when I last updated counts are located in the upper left-hand corner of each chart. Drop me a note if you see an error or if you have suggestions for improvements.

Average temperature anomalies with departures from 1901-2000 have been added to charts on this update.

The 2010s:

The 2000s:

The 1990s:

The 1980s:

The 1970s:

The 1960s:

The 1950s:

The 1940s:

The 1930s:

The 1920s:

The 1910s:

The 1900s:

The 1890s:

The 1880s:

All of the data listed below is part of this one chart. So far, the ratio of daily record high minimums to low maximums for the 2020s is higher than any decade since the 1880s:

Here are the current daily record counts per decade, which have gone into the above chart:

Nights have been warming faster than days across the U.S. looking at the following trends:

Current five-year and annual ratios:

The 2020s:

For the following data sets of record high minimums and low maximums I have highlighted months with over a ten to one or under one to ten ratios in bold type. The rankings are for the lower 48 states with the warmest ranking since 1895 of average temperatures being 132 and 1 being the coldest as of 2026. Blue colors represent cold months and red warm. Those months with counts close to a 1 to 1 ratio of highs to lows are colored black. Near average rankings between 54 to 74 are also colored black. Time stamps for when I last updated counts are located in the upper left-hand corner of each chart.

All-time record hottest or coldest months and years are boldly colored in purple.

Anomalies from 1904-2000 temperature averages have been added on this update.

Drop me a note if you see an error or if you have suggestions for improvements.