Temperature Measurement Fact of the Day

Climate scientists know this of course, but there is something I learned about surface temperature measurement that really surprised me when I first got into this climate thing.  Since this is a blog mainly aimed at educating the layman, I thought some of you might find this surprising as well.

Modern temperatures sensors, like the MMTS that is used at many official USHCN climate stations, can theoretically read temperatures every hour or minute or even continuously.  I originally presumed that these modern devices arrived at a daily temperature reading by continuously integrating the temperature over a 24-hour day, or at worst averaging 24 hourly readings.

WRONG!  While in fact many of the instruments could do this, in reality they do not.  The official daily temperature in the USHCN and most other databases is based on the average of that day’s high and low temperatures.  "Hey, that’s crazy!" You say.  "What if the temperature hovered at 50 degrees for 23 hours, and then a cold front comes in the last hour and drops the temperature 10 degrees.  Won’t that show the average for the day around 45 when in fact the real average is 49.8 or so?"  Yes.  All true.  The method is course and it sucks. 

Surface temperature measurements are often corrected if the time of day that a "day" begins and ends changes.  Apparently, a shift form a midnight to say a 3PM day break can make a several tenths of degree difference in the daily averages.  This made no sense to me.  How could this possibly be true?  Why should an arbitrary begin or end of a day make a difference, assuming that one is looking at a sufficiently long number of days.  That is how I found out that the sensors were not integrating over the day but just averaging highs and lows.  The latter methodology CAN be biased by the time selected for a day to begin and end (though I had to play around with a spreadsheet for a while to prove it to myself).  Stupid. Stupid. Stupid.

It is just another reason why the surface temperature measurement system is crap, and we should be depending on satellites instead.  Can anyone come up with one single answer as to why climate scientists eschew satellite measurements for surface temperatures EXCEPT that the satellites don’t give the dramatic answer they want to hear?  Does anyone for one second imagine that any climate scientist would spend 5 seconds defending the surface temperature measurement system over satellites if satellites gave higher temperature readings?

Postscript:  Roger Pielke has an interesting take on how this high-low average method introduces an upwards bias in surface temperatures.

  • Simon

    “What if the temperature hovered at 50 degrees for 23 hours, and then a cold front comes in the last hour and drops the temperature 10 degrees. Won’t that show the average for the day around 45 when in fact the real average is 49.8 or so?”

    Yep, I bet that happens ALL the time, probably every day. No wonder the results are skewed! Sheesh!