For a description of just how the anomaly stuff is done and some comments about why, see my comment here:
The graphs in the posting are a bit “rough” in that I’m still playing with the scales and colors. I started out “just red” then decided to color code by season but used two different colors for a season and, well, I can either spend a day re-doing them all or just “go with it” for now and describe the Interesting Thing I’ve discovered.
Basically I found that for some Continents (regions) the Winter Data is vastly more volatile than the Summer data. Most (all?) of the “warming” comes out of a reduction of the volatility to the downside of those cool season data. Hot seasons are just not very volatile. Yet they DO have an up tilt at the very end with the advent of electronic thermometers and lots more asphalt around airports.
My conclusion (up here at the top ;-) is that this leans strongly toward “hot black asphalt in the sun” causing the summer “lift” and “loss of high cold volatile places” causing the cool season “lift”. Remove those stations at altitude that tend to have the strongest cold excursions, you get a flatter winter curve with the cold anomalies pruned. Have an airport add 10,000 foot more runway and 20 acres of paved parking for Jet Travel, you get hotter summer readings.
It is my opinion that accounts for most of whatever warming is in this the unadjusted data. (That then gets enhanced with “all one way warmer” adjustments in the adjusted version).
Remember that this is the average of all thermometer anomalies over all years for all regions:
Note that the range is from -3 C to +2 C and the actual data ranges from about -2.5 C to +1.5C or about 1 C of total volatility.
Here’s the graph for Europe:
Very similar. (Other continents are not so similar, you can see them here:
As there are 7 Continents and 12 months, a set of all graphs would be 84 of them. I’ve made them for Asia, North America, Europe, Africa and Australia / Pacific Islands. Missing are Antarctica and South America. I’ll make them later. For now, what side tracked me into this posting was what happened with Europe and Asia (and to some extent North America – though I’ve not redone them to see how much).
The “problem” that popped up was that winter is just way more volatile than summer. Here’s Europe in summer:
Notice that I have expanded the range to be from -8 C to plus 6 C. That was needed to match winter below. Still, the data overall range from about -3 to +3 (one ‘flyer’ nearer 4 C) or about 6 C. The top end doesn’t get all that much warmer compared to 1850 to date, but the low excursions end recently.
What would happen if post W.W.II airports added more concrete and asphalt and more of the thermometers were placed at airports and inside the “Concrete Jungle” of cities? More solar heated pavement carryover into night warmth, less vegetation transpiration making cool and damp.
From -7 to +5 or 13 C of range. The “top end” doesn’t really get hotter. In fact, other than one “flyer” it is on a cooling trend. What DOES happen is that the lower/cold volatility gets pruned greatly.
What would happen with the loss of High Cold Places? Loss of cold directed volatility. What would happen with more stations over pavement (and with snow removal done and with tons of Jet-A kerosene being burned / hour and with constant swarm of “ground transportation” in the cities and at the airports)? Loss of low going anomaly cold.
IMHO, what these two graphs together show is that “altitude matters” in the set of thermometers and that “asphalt matters” especially in summer. Spring and Fall are intermediate between these results.
Very similar things are seen in other regions, and I’ve made a set of graphs by season for all regions that I’m putting in the next posting.
The assertion will be made that “the Reference Station Method using Anomalies” can fix that loss of “high cold places” 100% perfectly. My assertion is that they must fix a 6 C loss and could easily have a 1/2 C residual in that process (or a 1/12 of the anomaly error), or 8%. That’s all it would take. Where are their error bands?
For comparison in Europe, Fall and Spring:
This is the same -3 +2 range as the All Europe graph, so you can see what caused me to go WHA? when it first popped up. Just a huge scatter compared to the aggregate.
Here is the same September data on a -8 +6 scale for comparison to the other expanded range graphs:
Fall is basically dead flat. It isn’t about Fall, it is about Summer a little bit and Winter a lot.
Spring is basically dead flat. A little bit of ‘loss of cold’ at the very far right, but not much. Spring sunshine on the tarmac anyone? Again, it looks to me like it is all about the loss of winter cold excursions from high cold rural places, and a small addition of solar heated tarmac in the summer.
What it is NOT, is a generalize increase in the whole range of temperatures over all seasons from a “well mixed gas” that is ALWAYS present and “working”.
IMHO this is a Giant Dig Here!!!!! Deserving of a lot more looking a lot closer.
I’m going to continue to do my general “look it all over fast” thing first though. If anyone wants to take this and run with it, a footnote in your paper would be appreciated. (Yeah, I’m a “cheap date” ;-)