Profile photo of henryp



April 23, 2012 in climate change

Global cooling has started.

For those of you who think that earth is still warming: you are wrong. From a sample of 45 weather stations taken randomly from all over the world, I find that a turning point was reached around 1995. Note that the sample of weather stations is well balanced by latitude and 70/30 sea – inland. The plot for the acceleration / deceleration of warming/cooling (in degrees K/ t² ) appears to be natural looking, symmetrical, like the curve of a thrown object.


The (black) figures you are looking at in the tables below (allow some time to load up), represent the average change in degrees Celsius (or Kelvin) per annum, from the average temperatures measured during the period indicated. These are the slopes of the least square fit equations or “ linear trendlines” for the periods indicated, as calculated, i.e. the value before the x.

The average temperature data from the stations were obtained from I tried to avoid stations with many missing data. Nevertheless, it is very difficult finding weather stations that have no missing data at all. If a month’s data was found missing or if I found that the average for a month was based on less than 15 days of that month’s data, I looked at the average temperatures of that month of the preceding- and following year, averaged these, and in this way estimated the temperatures of that particular month’s missing data.


We note from my 3 tables below that Maxima, Means and Minima have all turned negative (from warming to cooling) between 12 and 22 years ago. The change in signal is best observed in that of the Maxima where we can see a gradual decline of the maximum temperatures from +0.036 degrees C per annum (over the last 38 years) to -0.016 (when taken over the last 12 years).

If we plot the global measurements for the change in Maxima, Means and Minima against the relevant time periods, it can be shown that the best fit for each of the curves is given by a polynominal of the 2nd order (parabolic fit).

Namely, for maxima it is

y= -0.00006 X2+ 0.00480X -0.06393

r²= 0.997


I have added a few more stations, (including Washington DC) which ramped up my r²= 0.998

The speed of warming/cooling for maxima now is 0.036 from 1974 (38 yrs), 0.029 from 1980 (32 yrs), 0.014 from 1990 (22 years) and -0.016 from 2000 (12 years).

For means, it is

y= -0.0001 X2 +0.0064X – 0.0778
r²= 0.959

For minima, it is

y= -0.00008 X2 + 0.00408X – 0.04178
r²= 0.985

Using the maxima plot, we note that at 0 (zero) when there was a turning point, i.e. no warming or cooling, we find x=17 years. From this sample of weather stations I can therefore estimate with reasonable accuracy that earth received its maximum energy input from the sun via the atmosphere during 2012-17=1995.

(if we are tempted to look at the root of same binominal on the other side, i.e. when global warming started, we find 68, suggesting that the global warming cycle started officially somewhere in 2012-68=1944. UPDATE: I realised this  result is speculative, as I do not have any real measurements from 1944-1973 but we are using an approximation from a probable plot. However, I did realize since some time ago that the plot I was looking at is really  like an a-c wave. I have subsequently been able to determine that the best sine wave for this plot would be one with a wavelength of 88 years. That would mean that the begnining of warming started somewhere around 1995-44=1951. That means we are now on a cooling curve unil ca. 1995+44=2039.

It can also be shown that the nature of the graph for means is one that lags a bit on the graph for maxima: earth has a store where it keeps its energy and a lot of that energy only comes out a bit later. Although the plot for means with rsquare 0.959 is still impressive, showing there is a definite relationship, I would not use it to determine the roots to give me the actual time when earth reached its maximum energy output (i.e. when it was the “warmest”). However, I would generally agree with the available datasets like RSS, UAH, Hadcrut3 and Hadsst2 that that must have been a few years after 1995.


Like I asked myself, you are going to ask why no one but me has figured that global cooling has already started. Your first point could be that my sample is too small. My argument against this would be that I already got similar results (for the longest period) with a sample less than half the size. See:

I also note that people like Roy Spencer and others got similar results than I report in my tables here. For example, RS reports a warming of 0.14 degree K per decade for the past 33 years. This is almost exactly the same result as reported in my table for Means here: 0.013 degree K/annum over the past 32 years.

The reason why I think scientists have not picked up on this is because they did not follow the same statistical principles that I applied when looking at all the data.

They might also have missed it because they kept looking at absolute data. Remember that we are looking at extremely small change. The problem when you measure temperature is: ERROR. How do you know your “thermometer” (recorder) reads right? You have to calibrate it. At regular intervals. In the method I used, I looked at the differences between the measurements which as such is a lot less dependent on calibration.

People would do well to take note of my results, as the drop in global temperature as shown in my table for Means (=average daily temperatures) of 12 x 0.019 = 0.2 degrees C over the past 12 years has been steeper than what I would have expected from the observed drop in maxima. The observed cooling trend could still accelerate further. It also appears from the results in my tables that earth’s energy store is big, but it looks depleted due to the maxima having been dropping for quite some time now. Earth’s temperature is also dropping now. Periods of cooling in the past have usually been associated with crop failures and subsequent periods of poverty and starvation. Hoping that these results will simply go away is like an ostriche putting its head in the sand. By checking the results at your local weather station you will be able to pick up on the particular trend in your own area and make the right decision on whether to sow or not to sow….Buying some extra warm cloths seems like a good idea to me.

(Please take some time to give my blog a star rating. Thanks)




  1. I found your work above, and especially the result of a cooling trend of .1 C per annum intriguing, given that work, (or an article) by P Gosselin based on US countryside temperature gauges (ignoring any in urban areas) gave the same very scary result of 0.1C decrease per annum.
    Clearly, countryside temperatures represent the true temperature.

    That’s a full blown glaciation in just 100 years. That would make the UK, Canada, the Northern part of the USA, and many Northern Hemisphere countries uninhabitable within a generation.

  2. Hi – I wonder if you are looking at this correctly?

    Your basic data as shown is not temperatures but change in temperatures per year.
    That is equivalent to velocity, which indicates that the rate of increase in temperature itself is slowing down and has turned negative in the last 12 years.

    That is to be expected with cyclic data.
    After the peak has been reached, then the next move is down.

    Now what concerns you is the fact that the rate of slowing is increasing.
    That is equivalent to the rate of downward acceleration.
    That is also quite natural if you think about cycles which can be thought of as sine waves (in an idealised model).

    The rate of change on a sine wave increases ever more slowly as you reach the top, pauses for an instant and then starts to slowly decline again. As it goes, the gownward speed accelerates until a maximum is reached half way down, when the rate of decline slowly dercelerates again until it once more stops and moves up again.

    Your linear projection would suggest that the temperature will eventually fall to -273.15 degrees centigrade and go on for ever lower from there.
    That is not likely.

    You need to fit a sine wave to you data and you need some hundreds or rather thousands of years of data, as the climate is chaotic and not periodic.

    There are some signs that we are in for about 30 years of cooling and some rather unclear signs that it may be much worse than that.
    But I feel that your approach exagerates the likely impact of it all.

    That’s my quick take.

  3. Henry@J Martinwell, looking just at the results of Anchorage (2 weather stations) I would definitely start thinking about packing my bags, if I lived there….

  4. Are you sure that it is cooling 0.1 degree per annum?

  5. GuyL

    I have a nitpicking question: could you clarify the definition of mean temperature:
    With most of the modern equipment temperature is recorded continuously, set at reading at the maximum rate that the equipment allows for, e.g. at least once per second or more.
    The average of all these measurements during a 24 hour day is the mean for that day. The mean for a particular month in a particular year is indeed the average of all the daily means for that month of that year.

    Either way, it does not really matter that much to me if the methodology and frequency of measurement of each weather station is not exactly the same as long as the method stays constant within the station. I am only interested in looking at the differences between measurements taken at that station.

Add Comment Register

Leave a reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>