The Canonical Set of dT/dt Graphs
This is an “aggregator” posting that ties together the entire set of graphs for the world done from the dT/dt “monthly anomaly” program. The purpose of this posting is to provide a single “entry point” into the global set of graphs (so that keeping one link is sufficient to reach them all). While I’ve created a ‘dT/dt’ category (on the right hand side of the blog page) to get to all of them, it is less structured as it is a reverse chronological listing. It will be easier to pick a desired geography from this page. General dT/dt articles will be under that dT/dt category. This posting is just the Canonical Set of Graphs.
I will be listing the regions in GHCN region order. That will place Africa first and Antarctica / Ships at Sea last.
Africa – Region 1
It would seem that we have a rather hard “Pivot” right in the middle of the data. Not exactly what one would expect from a steadily accumulating warming gas; but very much like what one expect from a human change of process and equipment …
Africa has many countries and I’ve divided that continent into several postings. That set of postings can be reached through this link:
Asia – Region 2
Not much happening and then, a really hard core “Hockey Stick Pivot” at the end.
The full set of Asian country graphs is here:
South America – Region 3
seen as a simple graph, it looks like a deep drop and rapid warming:
When we look at it by major thermometer count groups, we can see that the early segments are all dropping or flat, it is the splice at the major change of instruments that leads to the overall “warming” from the bottom. That, and the 1990 process change that induces a ‘warming trend’ after that (accompanied by dramatic reduction in volatility of the monthly anomalies, especially to the downside) We also see that the 1975 “jump” is an important factor. What was it? I don’t know. Perhaps a PDO flip? Or the start of he Jet Age of Airports? Did we change which thermometers were used without a significant count change? Or some other data handling difference? Certainly a “dig here!”. We see a similar effect in Africa. (But not in other regions…)
The full set of South American country graphs is here:
North America – Region 4
In some ways the flattest of the lot. Perhaps because so many eyes were watching the instruments ( such as surfacestations.org ) or perhaps just too many instruments to effectively swap around? Yet we still get a “hockey stick” blade grafted on the end.
The full set of North American country graphs is here:
Pacific Ocean / Basin – Region 5
A long steady slowly dropping sequence that gets a “Pivot” in the end as thermometers used and processes are changed.
When inspected “by segments”:
The Pacific Basin:
Europe – Region 6
We rise out of the Little Ice Age into a long flat history. Just to be “Sticked” in the end…
Europe has many countries and thus, many graphs. You can also see that it has a very long data series. Click on the graph for the full sized version so you can see just how long it is. The full set of postings is reached though this posting:
Antarctica – Region 7
I have only made one dT/dt graph so far for Antarctica. It is this one:
While the bottoms rise, the tops don’t. We are massively crushing the volatility out of the data from Antarctica, but it’s not getting any warmer. Then we get a bit of a “Pivot” to be “Sticked”…
Ships at Sea – Region 8
This is an odd “region” that has just a few fixed points at sea where passing ships will sometimes collect the temperature data and report it. The only graph I’ve made so far is this one:
Dropping. Regularly. Then a bit of “Pivot” on the process changes, and after getting back to zero we drop it. Hard.
Note on Process
In general, the dT/dt method is very straight forward. The first step is the creation of an “anomaly file” where each thermometer is compared to itself, month over month, and the difference between the two monthly entries is placed into a file. As a second step, a report is run that averages those anomalies for each year to give the dT/yr number. That is the amount by which that year is different from the prior year, on average, for that thermometer. A running total of these values is made (the dT field) and that represents cumulative change of temperature ( “Delta T”) to date.
This set of graphs differs from that general statement of the process on a modestly small, but important, point. The “Duplicate Number” or “Modification History Flag” handling. When there is a significant change in the processing applied to a given location, it gets a new “Duplicate Number”. These can often overlap for years or decades.
There are three ways that I’ve considered on “how to handle it”.
1) Average all the anomalies for each “Duplicate Number” together for a given thermometer and otherwise ignore it.
2) Treat each “Duplicate Number” as a unique thermometer record.
3) A hybrid approach. Treat each “Duplicate Number” as unique for the creation of the basic anomaly series. But average them together when there is more than one. Assign that Average Anomaly to the lowest “Duplicate Number” in the averaged set. Then, during the report, reset the annual anomaly values when a change of the “Duplicate Number” comes to the front.
(So, for example, in many locations there are 2 or 3 Duplicate Numbers in the years 1987 – 1990. These get averaged together and assigned to the “Duplicate Number” of 0 or whatever is the lowest number duplicate number in the set. Then, in 1991, when duplicate numbers 2, 1, and sometimes 0 end; the only duplicate number left is typically “3”. This then is visible to the report writer and we ‘take a reset’ on the annual anomalies.) That is what makes the most visible “bullseye” in the graphs as it avoids the ‘blending’ of duplicate number 3 with the earlier duplicate numbers that otherwise hides this change of process.
I chose to use the #3 Hybrid Approach to make this set of graphs as it gives the most clear indication of where there ‘are issues’ in the data. The temperature history gets slightly skewed as there is a dropout of data in that year as the annual anomaly is reset. (The Splice is highlighted.) For the most accurate indication of a probable temperature history, it would be best to use 1 or 2 ( I lean toward #2 but have not settled on one yet). The most commonly used temperature series tend to use #1 (GISS calls this the “as combined” set on their web site, though they average the temperatures rather than the anomalies) but in my opinion that hides too much about what is changing in the process. That is, it ‘hides too well’ that the rise is more from changed thermometers and change processing than from actual warming.
So at some future time I’ll be making another series of dT/dt method reports and graphs, with the intent being to find a more accurate representation of the actual change at each location. For now, this set is “pretty good” and has the “feature” that it highlights the problems spots in the temperature history.