(You can click on the graphs to get a larger version to see the details)
Belize – on the Caribbean Coast of Central America
Belize is an interesting little country. About the size of Vermont, but without all the mountains. It is on the Caribbean coast of Central America. The Caribbean being a place of extraordinary stability in some ways. The shallow water there warms under fairly consistent sun, and the shallowness prevents significant waves. Most of the Caribbean has wonderful beaches with very little in the way of waves and tides. (Unless a hurricane is passing through ;-)
I’ve been on a beach in Jamaica when the air and water were both about 87 F all the time. Just Heaven.
So look at that chart. Darned stable temperature. At the far left edge are a couple of years of data from the 1800s. Then, for unknown reasons, there is a long gap to the 1930’s and from there the series is full for each year. About 26.x C all the time.
About the only thing you can really say about that chart is that the “lower peaks” that make excursions down to the 25 C line get ‘clipped’ in The Great Dying of Thermometers. But I doubt if any of us would look at that chart and claim Belize was burning up with a rising temperature curve.
Costa Rica is a bit different from Belize, but not too much for comparisons. A bit further south. Coasts on both sides of the Central American Peninsula with both Pacific and Caribbean beaches. Stable temperatures and another little bit of Heaven on Earth for vacationers. Yes, a bit larger at about the size of Vermont, New Hampshire, and Connecticut combined. Yes, some bigger mountains down the middle. But still not much very far from water and not a lot of change.
Here “time begins” in 1941 with one relatively cool thermometer (one presumes in the mountains somewhere, but worth checking). In the late ’50s and early ’60s some more thermometers are added (probably near the beach, given the 25 C average temperatures). From that point onward, we see a darned stable 25 C average temperature. There is a bit of a step up in 1981 as a thermometer is deleted, but it steps back down again in 2001 with another deletion.
Again I think most folks would look at that and say “No warming here”. During each segment of stable thermometer counts, the ‘trend’ is basically flat.
Putting Belize and Costa Rica Together
Now let’s put these two on one graph together:
OK, I’ve added some trend lines.
Now first off, here is why I look AT THE DATA and look at tables more than graphs with trend lines. That trend line in Costa Rica looks like a real warming trend, but it hides the fact that the thermometer count jumped up right when the temps made a step function higher. It is the stable part to the right of that thermometer count change that matters. And you can see that both Belize and Costa Rica have more in common than not and that ‘temperatures change’ more in line with instrument changes than anything else.
OK, at that point the “anomaly advocates” like to toss rocks about how this is a silly approach and if you don’t use anomalies why of course you find that instrument change matters. (Often with various epithets included, since they seem to lack manners.)
Well, I look at those charts and see things that matter. I see stable segments when the instruments are stable (not what one would expect from a warming globe being steadily and relentlessly warmed via CO2). I see a ‘clipping of the down peaks’ in Belize coincident with a BIG thermometer count change. And I see a Caribbean that runs right around 25 C all the time from the 1800’s to now.
I think these things matter.
Lets look at it as Anomalies
Now lets look at the same place using the anomalies for these two countries. Realize this is based on ALL the data, unadorned. There are no interpolations, adjustments, in-fills, UHI corrections, whatever. Just the temperature data for each thermometer compared only to itself. This is done “by month” (that will not matter nearly so much for places bordering on the Caribbean as it IS so stable year round) and missing data gaps are bridged by waiting until a new value shows up in a future year so dropouts and gaps don’t have much impact.
The functions graphed here are:
dT/yr the average of the 12 monthly changes of temperature for the thermometers in the country when compared to the last valid value in that month for that thermometer. A very pure form of anomaly.
dT the running total of those anomalies. What is supposed to be the change of temperature to date.
Count is the number of thermometers in each country in a given year
And the trend lines for those dT values.
Here we see a strong trend line of rising anomalies for Belize while Costa Rica is shown as “warming” but much more slowly. Quite different from the actual temperatures in those countries. Anomaly processing has suppressed the “step function” in the early part of the Coast Rica series (what it’s supposed to do). And now Belize has a strong warming trend line.
But notice that the Belize values for dT tend to make “step functions” when instrument change happens. You can see the same thing for the Costa Rica graph as well, though after that first thermometer change, the thermometers don’t change as much as the ones in Belize.
So, IMHO, using “anomalies” is a process that has some sensitivity to instrument change. I suspect it is related to the ‘edge effects’ when a series of changes is left ‘unterminated’ (that is, you take a ‘rise’ in a year, then drop the thermometer and next year you do not get the offsetting drop, so perpetually preserve that ‘rise’). They clearly do NOT present the same information you get from direct inspection of the actual temperature histories for a location.
Think about it.
OK, for the inevitable folks about to launch into a tirade about how anomalies are great and using temperatures is dumb, and I “don’t get it” et. al.: Please don’t bother. I DO “get it” and I DO know why you would use anomalies. It lets you compare widely divergent instruments with the offsets removed and it lets you average a bunch of highly divergent things with less individual instrument bias whacking your average. (It suppresses that first ‘step up’ in Costa Rica when the lone cool thermometer gets a few friends). What I’m looking at is how anomalies CAN go bad and CAN mislead. Please realize that BOTH tools have value. Real temperatures keep you anchored in an absolute reference frame. That matters. Anomalies are subject to runaway accumulation of errors as series are left unterminated and as instruments change.
Now consider that ALL the global temperature series are based on much more complicated and much more “mucked with” anomalies based on temperatures that have had all sorts of adjustments, in-fill, “correction” etc. done to them. Do you really trust that to be valid to the 1/100 C place? Even 1/10 C?
For comparison, here is the GISS view of the world for 2009 ( near the last year of the graphs above):
Belize and Costa Rica are under that +1 C and near the +2 C area of Central America.
Now when I look at the temperature graphs up top, I just don’t see it. I see peaks that are consistently the same, and bottoms that have been clipped. (And with a bit of volatility compression as a result).
And THAT is why I think it is important to look AT the temperature DATA. Sure, use anomalies. But don’t ever forget to anchor yourself back in the real temperatures.