There are two serious problems in measuring global temperatures.

1. The first is that an average is usually meant to be based on a ‘normal distribution.’  But any temperature measure of the earth’s surface is not a normal distribution. Therefore the measures of mean, skew and variance are off the mark.

The Earth’s total surface is roughly 200 million square miles.  Of this 14 million square miles is the ice of Antarctica and the Arctic, which is always at 32 degrees Fahrenheit. That means 7% of the Earth is the same minimum temperature. Not a normal distribution.  Ice does not go below freezing temperature.

The land mass not covered in ice is 40 million square miles. That is 20% with widely fluctuating temperature.  The remaining 73% is ocean which cannot freeze, setting a lower temperature boundary of 32 degrees.  Definitely not a normal distribution for the surface of the planet.

I’ve swum at the equator so the upper temperature of the ocean is below 80 degrees.  The ocean is always within a 50 degree range.  Again, not a normal distribution. The average ocean temperature for the globe is 61 degrees Fahrenheit. Which is close to the midpoint of 80 plus 32 which is 56 degrees. 

My calculation of the global surface temperature based on satellites and balloons is 57 degrees, which has been steady since 1950.

2. The second problem in measuring the surface temperature of the earth is also that it is not a ‘normal’ distribution.  The bottom temperature that can be measured is very sticky around 32 degrees Fahrenheit.  On the 20% that is land mass the temperature can rise to any number or drop to any number, but the remaining 80% of the Earth is bounded on the bottom by the 32 degrees.  That is a very bottom truncated distribution.

This means that the global average can rise faster than it can fall. Making the surface temperature more sensitive to the volcano rises than to any factors that will lower it.

Summary: because the global temperature is not normal, because it is truncated at the bottom 32 degrees Fahrenheit, global temperature can rise faster than it can fall, it will have erratic measurements over time. Global surface temperature will rise rapidly for volcano events such as Pinatubo in 1991 but fall slowly thereafter.  The potential for overstating a global temperature rise is significant but largely a statistical anomaly.