Guest essay by Mark Fife
In today’s post on the Global Historical Climatology Data I am going to concentrate on daily minimum temperatures for long term stations in North America and Europe. As I mentioned last time the coverage is heavily weighted to the US.
In my last post I talked about the high amount of variance between stations. I conjectured most of those variances were due to localized site changes such as development. I believe that is a safe conjecture.
However, looking at daily minimums yields a different picture. The by site variation is there but it is not as pronounced. The standard deviation of the average of annual station average is only .46. That is a very reasonable value in comparison to my previous data set. The annual range between highest and lowest deviation from station average is consistent on average.
The following chart is the difference by year between the highest and lowest temperatures records for all stations in the study. While there is some variation over time the key point is the lack of any clear trend on average. There is a fluctuation in the magnitude of in year variation but that appears to be due to weather events with in the US in the form of hot and cold waves. Because the data is heavily weighted to the US it is sensitive to such events in the US.
This is the average daily minimum temperature record for all stations as mentioned above. It is a reasonable approximation of individual station records.
The following graph may grab your interest if you are familiar with statistics, especially that brand of statistics used in Quality Engineering. If you are interested in the technique you can Google search for Statistical Process Control. This is a well-established methodology which has been in use since the 1950’s.
What you see here is my twist on the method. I have transposed the data shown in the preceding chart by converting it to standard deviations with the overall average normalized to zero. This is nothing more than a graphical test for equality of the means. Confidence intervals are thus easily defined, such as ± 1.96 standard deviations form a 95% confidence interval. The second key indicator of a shift in the mean is the number of consecutive points above or below the zero line.
There is no question about the clear signal of a pattern here. There are also clear evidence of extreme events occurring in 1904, 1917, 1921, 1931, and 1998.
Thus far I see no reason to doubt the veracity or accuracy of these extreme events. They appear to be accurate. They are, however, out of the ordinary. The other interesting observation is how the year to year variability decreased going into the 1940’s and then again in the 1960’s. That variability increases coming into the 1970’s. That is reflected in the chart of annual ranges above.
The conclusions I draw are as follows:
- There is evidence of a regular pattern about 60 or so years in length.
- There is no statistically significant difference between the 1900’s and the 1960’s. Using my normalized data, the 1960’s is warmer by 0.07 standard deviations. This is insignificant.
- There is no significant difference between the 1930’s and either the 1990’s or the 2000’s. The 1930’s are warmer than either by .08 standard deviations. This is insignificant.
- If this pattern holds true I would expect to see a low point going into the 2020’s. This does appear to be happening, but I would be very careful drawing conclusions from short term data. However, similarities do exist between 1931 to 1942 and 1998 to 2007.
Finally, the last question is why would the daily low temperatures show such a different result? I will hazard a few guesses:
- Daily lows must be unaffected for the most part by site changes which cause higher day time temperatures.
- Structures and surfaces added to a site cause increased temperature due to differences in absorbed energy and in heat capacity or specific heat. Lower heat capacity or specific heat means surfaces and objects achieve a higher temperature for the same energy absorbed than surfaces and objects with higher heat capacities or specific heats. That generally means they cool off more quickly as well. Therefore, the extra heat is not retained.
- The effect just described above is the opposite effect where specific heat is relatively higher. The best example of that effect is water. Water in either liquid or gas form has a much higher specific heat than a normal atmospheric gas mixture, concrete, brick, shingles, and so forth. A body of water not only stays cooler during the day than what is on the land, it also cools off much slower.
The lowest temperature of a typical day in most locations normally occurs within an hour of sunrise. To systematically affect the daily minimum temperature objects, structures, and surfaces that would retain or produce heat must be added to the site. That is possible, certainly adding a pond or lake next to a climate station could have such an affect.
This result and the obviously different outcome from my prior study supports the supposition most instances of higher than typical temperature increases are due to site changes as described above.
I would further conclude the daily minimum temperatures provide a far more accurate picture of what is happening with respect to the anthropogenic global warming theory.
The lack of any evidence of a change in heat retained overnight, if correct, would debunk the concept added CO2 is causing the surface of the Earth to warm up due to downward IR. The logic behind this assertion is simple. If CO2 truly did act as a greenhouse or a blanket to retard cooling that effect would be demonstrable in progressively higher overnight temperatures. There is no evidence that has occurred.
You could conjecture as to whether temperatures have increase during those overnight hours which precede the daily low point. This data does not address that conjecture.