Weather Sensors Cause Climate Errors

May 11, 2007

Weather observations have been taken around the world for centuries. Up until the early 1980s, a majority of the temperature observations were taken with a Liquid in Glass (LIG) mercury thermometer. Special LIG thermometers, known as minimum and maximum thermometers, were used to record the daily high and low temperature. These thermometers worked very simply and were quite accurate. The mercury rose (or fell) and marked the high (or low) temperature for the day. The temperature using this method was the absolute maximum or minimum; no averaging or sampling was used.

In the 1980s, technology allowed for sensors to become automated and computerized. Instead of using the absolute maximum or minimum, the new digital thermometers utilize algorithms to calculate the high and low temperature. The algorithm uses a 60 second sampling rate and calculates a running 5 minute average.

So how does this affect temperatures? A joint study conducted by the University of Nebraska and Nation Climate Data Center demonstrated that the difference between the automated sensors and LIG thermometers was between .15 and .5 degrees C.

I have yet to read any global warming research that accounts for this variation. This reason is simple – far too many environmental scientists, geologists, ecologists, and others with no expertise in meteorology are conducting global warming research. Without doubt, there is plenty of research funding to go around for global warming. Unfortunately it too often winds up in the hands of amateurs with no background in meteorology.


15 Responses to “Weather Sensors Cause Climate Errors”

  1. Patrick Farrington said

    Is the difference a warm bias, or on the cool side? Or can it be either way?

  2. climatepolice said

    For minimum temperatures, digital thermometers have a warm bias. For maximum temperatures it is less clear but most other studies suggest it also is a warm bias.

  3. Vincent Gray said

    The idea that the “mean daily average” is the sum of the maximum and the minium divided by two violates the basicv principles of mathematical statistics. It is always biased, by an unknown amount. The method described, in which it is measured by an “algorithm” may well introduce an extra bias, but since we have no mathematically acceptable data on the error involved in choosing the max/min avergae instead of a statistically acceptable average, we cannot teel what the bias is, except that it is probably higher than the 0.7ºC alleged temperature rise since 1850.

    The US workers have trid to assess the bias from using the max/min average, including the fact that it does not apply to a standard 24 hours in their “homogenisation exercise got tyhe GHCN series. However, they have to admit that such an exercise is impossible in the vast majority of the latitude/longitude boxes used for the “surface anomaly record” because very few countries have sufficient stations for a statistically valid comparison.

    I do not understand the above comment. How can a single reading, such as a minimum have a “bias”?
    The bias is in the wrong method for determining the average temp[erature, and there is no way we can find out its magnitude.

  4. John McL said

    The “official” definition of the “mean temperature” is the arithmetic average of the minimum and maximum temperatures. These figures were measured according to your description above but how accurate are they?

    Last Saturday at Coldstream, an observation station just outside Melbourne (Australia) the temperature varied between 5.2 and 9.7 deg C from midnight until 11am, rose to 22 at 2pm, was still above 20 until 4:30pm, fell to 10 at 6:30pm and remained below 10 until midnight. In other words a cloudy and cold morning was followed by a warm afternoon before the cool of evening.

    “Mean temperature”(as mid point of min and max) was 13.7 degrees. Arithmetic average of 48 recordings made at 30 minute intervals was 10.11 degrees.

    Wait a moment, the designated minimum on any one day is from 9:00am on the previous day to 9:00am on the current day, and the designated maximum is from 9:00am on the current day to 9:00am on the next (Yes, the min and max don’t apply to the same period of time!) Calculated by this method the mid-point of minimum and maximum temperatures was 13.4 degrees.

    That’s at least 3.3 degrees between the arithmetic average for the 24 hours and the convoluted method used to determine the official “mean”.

    If that doesn’t shock you, then one more factor might. A short period of bright sunshine on an otherwise cloudy day – or a short period of warm winds – will have a disproportionate effect on the official “mean temperature”. Outside the tropical regions (20S to 20N) low dense cloud has reduced since 1999, mid level cloud has increased sligtly and thin upper level cloud also increased slightly. This in combination with the corrupt calculations may suggest far more warming than has really occured.

  5. Jeff Norman said

    It was the change in methodology that introduced the alleged positive bias into the data set and may or may not have effected the estimated regional/hemispheric/global average.

    Therefore there is reason to believe these estimates are skewed on the positive side and one can no longer express changes in temperature as 0.7±x°C (say) but as 0.7 +0.5x -2x°C.

  6. Walter E. Wallis said

    What we have here is another example of slide rule accuracy misuse. Averages cannot be more accurate than the individual readings. To claim a one degree C rise in average temperature a century is fatuous.

  7. Bob said

    Does this mean that historical temperature records might be higher than those we accept as valid?

    It would seem that the bias could be estimated by comparing measurements by the two methods at several selected stations over an extended time period. If we had to go back and stick in mercury thermometors at a number of stations and take the measurements for a couple of years, then do so.

  8. Daniel Mattas said

    And what about overall uncertainty of measurements? It is well known that this is about 0.2 to 1 degree (Celsius) for non-calibrated mercury thermometers. But what is it for electronic sensors? What is their long-term stability? How often are the sensors checked for accuracy/calibrated? Etc.

  9. Cameron Kuhns said

    I think that digital thermometers have a warm bias. At home we have two thermometers, One digital the other a bi-metal spring and I have noticed that the digital thermometer is 3 to 4 degrees warmer than the bi-metal spring one.

  10. A minimum is an absolute minimum. Averaging over a 5 minute period (which is what the digital automatic thermometers do) nearly always raises the minimum, thereby creating the bias. In reality, the true minimum is no longer being reported.

  11. Jim Clarke said

    “…far too many environmental scientists, geologists, ecologists, and others with no expertise in meteorology are conducting global warming research.”

    True, but it goes deeper than that! There is also a gap among atmospheric scientists as well. The AGW crisis is being driven primarily by theorists and academics while the synoptic (forecasting) meteorologists remain largely sceptical. The academics like to believe that this gap is the product of the forecasters ignorance. On the contrary, it is likely the result of the forecasters wisdom!

  12. Stan B said

    So now we have a change in instrumentation, a change in paint on the boxes the instruments are kept in, and the UHIE to contend with.

    Not to mention models that must be continually tweaked to keep some semblance of reality when feeding all this faulty, biased data into.

    And they want me to believe the weather forecast for 100 years from now?

  13. David said

    Kind of on the same subject, I would love for someone to do an analysis on the accuracy of the recorded temperatures that we always hear so much about. Who recorded the data? Where was it recorded? For instance, at, they indicate that the record high on May 14th for Albuquerque, NM was 92 F in 1938 and the record low was 32 F in 1907. I’d be very interested in knowing exactly who, how, and when those measurements were taken, and if there’s any chance that they could be wrong.

  14. Steven Zell said

    If someone tries to calculate the average daily temperature at a given site by measuring the daily minimum and maximum, adding, and dividing by two, there will probably be a seasonal bias relative to measuring temperatures hourly, and dividing the sum by 24.

    During clear weather, temperatures are most dependent on solar heating during the day, and radiational cooling at night. This wouldn’t hold during cloudy or stormy weather, when temperatures are more affected by advection of warm or cold air from other areas.

    During spring and summer, the days are longer than the nights, morning temperatures rise sharply after sunrise, and tend to remain near the daily maximum most of the daylight hours, while night-time temperatures are near the daily minimum for relatively few hours, meaning that a 24-hour average will be closer to the maximum than the minimum.

    During fall and winter, when nights are longer than days, temperatures fall sharply after sunset, and usually flatten out near the dewpoint (condensing dew releases latent heat), spending many hours near the daily minimum, while warmer daylight temperatures are limited to a few hours around solar noon, meaning that a 24-hour average will be closer to the minimum than the maximum.

    Relative to a more realistic average obtained by summing hourly readings and dividing by 24, the old (minimum + maximum) / 2 method would give warmer winter averages and colder summer averages, and would tend to reduce seasonal fluctuations. Then we wonder why there’s more “extreme” weather recently–we’re just measuring it better!

  15. Vincent Gray said

    Her I come again.

    Of course, the Americans (Karl, Peterson, Vose et al) have tried to apply a “correction” to the statistically unacceptable max/min average by comparing it with an average fro hourly reradings from a number of sites. They found a large bias, with a large inaccuracy. They try to “correct” the US (GHCN) readings for “homogeneity’. What comes out is a temperature record for the USA which shows no signs of being influenced by greenhouse gas emissions.

    A similar exercise was carried out in China.I have the paper whih shows the results, in

    Zhao et al. 2005 Acta Meteorologica Sinica Vol 19 pages 389-400. The “corrected” temperature record for China turns out, also, to show no signs of influence from greenhouse gases.

    It is of interest that the second author of this paper was Yasui Ding, who used to be co-Chairman of th WGI Committee of the IPCC, but apparently has been replaced.

    The comparison between max/min averages and horly readings siimply cannot be done on a worldwide scale, The vast majority of the readings used for the “global average” are therefore biased to an inknown degree, but ceratinly by several degrees.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: