Rising Maximum Temperatures

Guest Opinion by Kip Hansen — 5 January 2024 — 2500 words/15 minutes

Roger Pielke Jr. recently posted a piece at The Honest Broker titled:  “U.S. Climate Extremes: 2023 Year in Review   – A Very Normal Year” – which was subsequently reposted at WUWT.

In that post, he uses this graphic:

(I have increased the size of the titles for clarity – kh)

It is easy to see that the trends of both the Maximum January temperatures and the Maximum July temperatures have been rising — more so for January temperatures than July — though this is somewhat obscured by the different scales of the two graphs.   [Caveat:  The temperature record on which this graph is based is not scientifically reliable before about 1940.] Also, one has to be careful to note what exactly they are really measuring. 

This is not the usual average temperature.  Not monthly average temperature. 

It is Contiguous U.S. Maximum Temperature for these two months, January and July – based on the assumption that these are the coldest and hottest months.    At least we can say they represent a cold month and a hot month.

So how do we calculate such a record?  Let’s just assume NOAA has done what is usually does – it took some kind of an average of the maximum temperatures reported each day by its weather stations in the Contiguous United States – those temperatures reported usually as “Tmax” in the daily station records. 

Let’s leave aside all my usual arguments about the inanity of such averages and just accept the idea that they are trying to represent.  (None of this is Roger Pielke Jr.’s  fault – he is just reporting what they say in the NOAA produced graphics.)

But rather, consider exactly what they are reporting – Maximum Daily Temperatures (averaged somehow).  But how is this measured?

How was this measure in the early 20th century?  They used something like this, the Six’s Min-Max Thermometer. 

However, the Wiki explains:

“MMTS (meteorology)

A Maximum Minimum Temperature System or MMTS is a temperature recording system that keeps track of the maximum and minimum temperatures that have occurred over some given time period.

The earliest, and still perhaps most familiar, form is the Maximum minimum thermometer invented by James Six in 1782.

Today a typical MMTS is a thermistor. This may be read locally or can transmit its results electronically.”

Weather.gov offers this information:

[If you are familiar with weather stations you can skip this section.- kh]

Temperature Sensors – Liquid

Thermometers used in a CRS [Cotton Region Shelter] are Liquid In Glass (LIG) and are either alcohol or mercury.   Alcohol thermometers are employed in the colder climates where winter temperatures drop below -40 degrees, the freezing point of mercury.   Minimum temperature thermometers have a small bar embedded in the liquid that is pulled down the tube as the temperature falls.   As the temperature warms again and the liquid moves back up the tube, the bar remains at the minimum temperature.   This allows the observer to read the lowest temperature.   Maximum thermometers have a small break near the base of the well of liquid at the bottom of the thermometer.   As the temperature falls from the maximum, this break in the liquid keeps the liquid in place at its high point.   The maximum and minimum thermometers are mounted on a rack.   After noting the highest and lowest temperatures, the observer then tilts the rack.   This resets the thermometers by rejoining the liquid in the “maximum” thermometer and sending the bar back to the top of the liquid in the “minimum” thermometer.   The thermometers are now reset, allowing observation of the highest and lowest temperatures for the next day.

Temperature Sensor – Electronic

[The newer electronic MMTS can look like the one picture here.)

Another and newer type of thermometer is the Maximum Minimum Temperature System (MMTS).   An MMTS is an electronic thermometer, not too different from the type one might buy at a local electronics store.   The MMTS is a thermistor.   This thermistor is housed in a shelter similar in appearance to a bee hive.   This design is similar in functionality to the CRS.   Currently, the MMTS requires a cable to connect the sensor with a display.   Future plans are for wireless displays.   This would eliminate many of the problems associated with cabled systems.

In the 1980s, thermistor MMTS units began to be introduced into the NOAA and NWS systems.  [ source ]

In a larger application, such as the NY State Mesonet, a typical station looks like this:

This is a Mesonet station high in the Catskill Mountains, I took this photo a couple of weeks ago.  Circled is the 6-foot temperature sensor.  It is specifically a RM Young 41342  [spec sheet] in a RM Young 43502 – Aspirated Radiation Shield.  The standard version has an accuracy (at 23°C ) of ±0.3°C (about 0.5°F), with a 10 second response time. 

Why do we need to know the response time when measuring 6-foot (2 m) air temperature?   Well, when you were a child (most of us, anyway) the doctor and your mother took your temperature with “liquid-in-glass” oral thermometer (for me, a mercury-in-glass then later alcohol-in-glass) which you were required to hold “under your tongue”  for how long?  “for 3 minutes.”  That’s how long it took to get a “liquid-in-glass” [LIG] thermometer to reliably change and record temperature.   Our original Six’s Min-Max Thermometer-style thermometers, used for many years and still in use in some places today, had a similar response time to changes in temperature measured in minutes – not seconds.

This becomes important when looking at the Maximum Temperature record for any weather station.  For a properly sited weather station, which would look something like the Mesonet station pictured above in many ways, there is little chance of spurious very-short-term temperature changes being recorded by an electronic MMTS.  There are no parking lots, no air conditioners, no jet exhaust, no delivery trucks, no buildings reflecting heat, no odd little shifts of blowing a stream of uncharacteristic hotter air over the sensor for a minute, etc.

Many NOAA weather stations consist of a MMTS alone, on a pole. (see Anthony Watts’ Surface Station Project reports.)

[Interesting note:  The NYS Mesonet station TANN, pictured here with a UFO (unidentified finger object) in the upper left, looked to me, when I visited it on a cold snowy morning, to be a very well sited weather station.  But on its site data page it gives a siting rating for various measurements according to WMO SITING CLASSIFICATIONS FOR SURFACE OBSERVING STATIONS ON LAND [required reading for anyone concerned about station siting and the temperature/weather record] in which “WMO guidelines give different variables a classification number, with 1 being the best [on a scale of 1-5]. Higher-numbered classifications indicate that the site’s surrounding environment may cause uncertainty in the data.”  Of the three categories rated for this station, it received a “4” for Temperature/Humidity and a “5”s for Surface Wind and Precipitation.]

Back to response time:  Why would this make a difference?

I didn’t know but I had suspicions….so, naturally, I asked Anthony Watts, probably the man most knowledgeable about how temperatures are measured inside of Stevenson Screens, in the somewhat similar Cotton Region Shelters (CRS), and in modern electronic weather stations, this question:

“Pielke Jr has published these graphs of January and July Maximum temperatures.  (his substack and at WUWT.)

What are the chances that some of the rise is due to the use of electronic weather stations which report INSTANTANEOUS highs and lows?”

With Anthony’s permission, I quote his answer:

“Absolutely, I’m convinced that short, local events, such as a wind shift bringing heat from pavement can contribute to a high that is spurious. The MMTS system as well as the ASOS system logs the Tmax – it does not log the duration.

The response time of a mercury or alcohol max/min thermometer basically makes it a low pass filter, and such spurious events don’t get recorded.

The solution is to install a “mass hat” on an electronic thermometer sensor to get its response time down to that of a mercury or alcohol max/min thermometer.” — Anthony Watts (personal communication)

[“Mass hat” – this would be something like a sleeve that slips over the thermistor, which is the long skinny probe seen in this image, with sufficient mass that has to change temperature before the thermistor is affected – thus slowing the response time of the thermistor to more closely match that of liquid-in-glass max/min thermometers.]

How’s that, you say?  It is a result of exactly what they are measuring and recording: the Maximum temperature reading.  Here are the hourly Maximums from an imaginary weather station:

We can see the usual diurnal shape, warming to midday, cooling overnight.  But, hey, what’s that sticking up in the middle? At 1000 hrs?  That, my friends, is a spurious instantaneous temperature reading.  You see, this below is the imaginary Anywhere, U.S.A. station:

You can see the MMTS there on the left-hand side, on the lawn, and the five air conditioning units 6 to 8 feet away.  Maybe, around 10 o’clock, the buildings air conditions all started up, timer controlled, and kicked out lots of heat just as an errant puff of wind came around the corner from the building on the right, blowing all that extra heat over the MMTS for a minute.  The MMTS dutifully records a new Maximum Temperature.  That little spike would be reported as the Maximum for the day, averaged into the Monthly Maximum.  As more and more MMTS are added to the network, the more spurious instantaneous Maximums can be recorded, slowly driving the Contiguous U.S. Maximum Temperature of January or July up a bit each year, as the number of MMTS units increase the number of spurious readings.

These types of spurious Tmax readings can be caused by all sorts of things.  See Anthony’s two Surface Station reports [2009 here and 2022 here].  At airports, a badly sited MMTS can be influenced by passing or turning jet airplanes on the runway, taking off or landing. For parking lot sited stations, a UPS or Amazon truck parked right next to the MMTS can reflect extra heat onto the MMTS for a minute or two.  An odd little puff of wind picks up the hottest air six inches above the black asphalt and wafts it up over the MMTS.  The point is that it doesn’t have to last long – 10 second response time!  New Tmax!

Let me provide a real-time, real-life example from a weather station I have visited many times:  C – Turkey Point Hudson River NERRS, NY (NOS 8518962).   We can find the real time standard meteorological data for last 45 days from this page.  Temperatures are recorded at six-minute intervals (which are averaged instantaneous measurements).  [Note:  ASOS stations, on the other hand,  use five-minute intervals.] Looking closely at the data for examples, we find this 30 minute period on December 18^th 2023 from 1436 to 1500 – five six minute records:

   

A temperature jump of 4.2°C  (or 7.6°F)  in six minutes?   For over 20 minutes, the recorded temperature remains higher, and then is aback at 10-11°C. [See note just below]  There are lots of instances of these types of oddities in the record of this station.  In this case, we have a 10-11°C (about 50°F) day suddenly transformed into a 15°C (60°F) day  — for 15 minutes.  That 15°C is a Tmax for the day – almost 4°C higher than the rest of the day.  The average temp (including the spurious reading)  for the hour (all six-minute records) in which this oddity occurs is 12.7°C.

Note :  “Once each minute the ACU [which is the central processing unit for the ASOS] calculates the 5-minute average ambient temperature and dew point temperature from the 1-minute average observations (provided at least 4 valid 1-minute averages are available). These 5-minute averages are rounded to the nearest degree Fahrenheit, converted to the nearest 0.1 degree Celsius, and reported once each minute as the 5-minute average ambient and dew point temperatures. All mid-point temperature values are rounded up (e.g., +3.5°F rounds up to +4.0°F; -3.5°F rounds up to – 3.0°F; while -3.6 °F rounds to -4.0 °F).”  [source:  ASOS Users Guide, 1998]  The station illustrated is a NERRS station and uses 6 minute intervals, but the algorithm is similar – kh.

Here is how that works to bias the temperature record – both the Tmax record and the Tavg record:

(Click image see full size in a new tab or window)

The five latest 1-minute values are averaged, giving a new 5-minute-average every minute.  In the NERRS network,  1-minute values are averaged every five minutes to created the Recorded 5-minute Temperature record.  [Note: different agencies use slightly differing algorithms and timings – NERRS uses six minute averages, while ASOS uses five minutes.]  A single 1-minute spurious temperature causes five spuriously high 5-minute averages in the “averaged every minute” system used in ASOS (the orange trace in the graphs above) .  In the NERRS network, 1 spurious reading creates at least two spuriously high Recorded 5-minute values (red trace and stars). 

The graph at the top of this essay – Contiguous U.S. Maximum Temperature – is created by this process:  “Once each day (at 23:59 LST), the highest and lowest ambient temperatures for the current month, along with the date(s) of occurrence, are computed and stored in memory until the end of the following month. On the first day of the following month, ASOS outputs the Monthly Maximum Temperature and date(s) of occurrence, plus the Monthly Minimum Temperature and date(s) of occurrence.”  It logically follows that spuriously high instantaneous readings can easily make that monthly Maximum Temperature listing and thus create the graph from NOAA highlighted by Pielke Jr.

Yes, it can be confusing, but:  The NERRS network does not record each 1 minute temperature, only an average every six minutes.  ASOS and MMTS record a new 5-minute average every minute, which are also not a record of the 1-miute temperature measurements themselves.
These are examples of spurious instantaneous MMTS/ASOS temperature readings and their effects – and lead to the bottom line:

Bottom Line:

There is a reasonable hypothesis that could or should be investigated:

With the widespread introduction of MMTS and ASOS weather stations over time since 1980, which record instantaneous temperatures every minute with a 10 second response time, spurious instantaneous high temperatures can be recorded as Tmax driving up both the daily temperature average (Tavg) and the daily, weekly monthly and annual Tmax records.

# # # # #

Author’s Comment:

A good question based on curiosity about an observation of something (anomalous or not) is the basis of all good science and research.

This topic could be important – as all the temperature records (local, US Contiguous, Regional, Global)  are based on the record of Tavg – the daily “average” temperature of a weather station.  That “average” is not the average of all the temperature measurements for a 24 hour period, but rather the “average” of the Tmin and Tmax  of that 24 hour period.  Thus, Daily/Weekly/Monthly Average Temperatures are highly influenced by Tmax.  [for details, see this document from the National Centers for Environmental Information]

It is not just the Tmax record that can be nudged higher by anomalous instantaneous affects all of the subsequent  temperature metrics.

Nearly 20 years ago, K. G. Hubbard et al. produce a paper titled “Air Temperature Comparison between the MMTS and the USCRN Temperature Systems” which found MMTS systems biased Tmax high and Tmin low. It was based on a single year’s worth of data but claims that MMTS data was or maybe still is being “corrected”.

Just to be clear:  This Opinion piece represents my personal investigation and  opinion on the topic – I have quoted Anthony Watts’ response to my emailed question.  Everything else, every word, is my responsibility and does not necessarily represent his viewpoint.

Thanks for reading.

# # # # #