The officially published global temperature records all converge on a total temperature rise since the late 19th century of about 0.9 (0.8–1.0) degrees Celsius:
But to what extent can we be confident that this is how the ‘global average surface temperature’ (GAST) anomaly actually evolved over this time frame?
The truth is: We can’t. At all.
This is fundamentally a matter of data coverage, but – just as importantly – it is also a matter of methodology. How do you make up for a paucity of data? How do you properly compile, weight and interpolate data into a reliable “global average” when that data – the actual observational information that you have collected and thus have at your disposal – provides nothing like a full global coverage? And how do you make this “global average” of yours consistent over time when your data coverage (both in total and in spatial distribution) vastly changes over that same time frame? What basic assumptions will you have to rely on? Because, make no mistake, an interpretive undertaking such as this will crucially have to rest on a foundation of some rather sweeping presuppositions.
The reality is, we don’t have anywhere near global temperature data coverage prior to the satellite era. It is pretty decent, maybe even good, from the late 60s-early 70s onward, but practically globally complete and uniform* from 1979 only. That is, with the satellite records. Even today, only the satellites provide real global data coverage. And this point is important to bear in mind. Surface-based records do no such thing. Despite certain persistent claims to the contrary …
*(Yes, there are effectively a few small white areas still, at the poles and in the highest regions of the world, like Tibet and the Andean Altiplano. But these only include somewhere between 1% (UAH tlt) and 4% (RSS tlt) of the Earth.)
The coverage in the Southern Hemisphere was spotty, to the say the least, prior to the 70s, all but nonexistent south of the 45th parallel. Same with the Arctic. We simply don’t know what was going on in these regions. We have to guess. Here’s Phil Jones of the UEA Climate Research Unit (from the Climategate emails):
The issue Ray alludes to is that in addition to the issue of many more drifters providing measurements over the last 5-10 years, the measurements are coming in from places where we didn’t have much ship data in the past. For much of the SH between 40 and 60S the normals are mostly made up as there is very little ship data there.
You can gather what he’s talking about from this 2010 summary by Bob Tisdale:
ICOADS: THE BASIS FOR LONG-TERM SST DATA
Long-term Sea Surface Temperature (SST) datasets are based on the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) data, which is divided into 2 degree grids. ICOADS link:
Sea Surface Temperature (SST) before the advent of buoys and satellites was measured by ships. The majority of ships traversing the oceans kept to shipping lanes. This limited the spatial coverage of SST measurements, especially in the South Pacific, which continues to have limited measurements in recent decades. Figure 1 is a collection of COADS SST data coverage maps available from the National Center for Atmospheric Research (NCAR). And here are the links to the full-size individual maps:
[Links not working.]
NCAR describes the maps as “Percentage of non-missing data in each time period is plotted. The minimum number of observations needed per month per grid box was 1.” Non-missing data is an interesting but descriptive explanation. It definitely gets the point across that missing data is the norm in many areas. Each of the maps illustrates the monthly SST coverage for a 20-year period, starting in 1861 and ending in 1997. Zero to 10% coverage is in white, while complete coverage (90 to 100%) is shown in gold. Even the last period from 1981 to 1997 shows major gaps in the data for the South Pacific.
[His Figure 1.]
The SST data suppliers take the long-term ICOADS data, make corrections for the transitions from one measuring method to another (non-insulated buckets, insulated buckets, ship intakes, buoys, satellites) and infill missing data using methods that vary from one dataset to another.
I’ll make Tisdale’s Fig.1 my Figure 2:
The compilers of the official long-term anomaly records of ‘global average surface temperature’ (GAST) have necessarily taken huge liberties with their interpretive approach to the available data prior to about 1970. (And as we shall see in a later post, they have actually taken liberties with the post-1970 GAST data as well. Though in a very different way.)
The global estimates as presented simply do not appear to be compatible with reality. Because of fundamental assumptions being made (and broken, within the very same record) about the nature (and level) of temperature covariance over time between certain regions of the world. And this is something that is in fact surprisingly easy to show.
The long-term GAST timeseries perhaps aren’t so much deliberately and actively manipulated [but don’t take my word for it] as simply being based on flawed physical reasoning governing the ‘global averaging’ of actual (observational) and infilled (‘made up’) data. This will become evident from the analysis below.
Here are three specific – and very illuminating – operations that you can perform to discover and highlight the fundamental problem inherent in the long-term GAST timeseries as officially presented to the world:
- Global LAND (T2m anom.) vs. global OCEAN (SSTa),
- Northern vs. Southern Hemisphere, and
- Global LAND+OCEAN vs. North Atlantic SSTa (AMO with natural trend included).
1. GLOBAL LAND (T2m anom.) vs. GLOBAL OCEAN (SSTa)
What do we see?
The two curves definitely covary.
BUT, from the years just prior to ‘the modern era of global warming’, starting in 1976-77, and up till today, the global temperature anomaly over LAND appears to have increased by about twice the amount of the global temperature anomaly over the OCEAN. However, if we follow the two curves backwards in time from that very same period, we find that there is absolutely no overall difference in warming and cooling rates between the two domains. The amplitudes are basically the same.
In what universe does this make any physical sense at all?
2. NORTHERN vs. SOUTHERN HEMISPHERE
Animation 1. Red curve: NH T2m+SST anomaly (HadCRUt4); green curve: SH T2m+SST anomaly (HadCRUt4). Blue vertical line marks 1968.
What do we see? Once again, we see definite covariance. But the way these two curves ultimately diverge is quite different from the way the land/ocean ones diverged.
From the short decade just preceding the start of ‘the modern era of global warming’ (1968-76) and till today, the total rise in NH temp anomalies seem to be double that of the total rise in SH temp anomalies. However, this is not the case at all if we follow the two curves backwards in time from where the 1968 blue vertical line intersects them both. Something huge seems to happen right at that point. During the 78 years leading up to it (1890-1967), the red and green curves track each other to an impressive degree, with more or less equal amplitudes, and there is absolutely no overall divergence between them. But then 1968 arrives, and, out of the blue, the SH temp anomalies are seen to shoot UP where the NH ones dive DOWN, so that the relative alignment of the two curves is permanently shifted by quite exactly 0.2 degrees! And from then on, the NH anomalies all of a sudden rise much faster than the SH anomalies …
In what universe does all this make any physical sense at all?
3. GLOBAL LAND+OCEAN vs. NORTH ATLANTIC SSTa (AMO with natural trend included)
Animation 2. Red+blue+purple curve: gl T2m+SST anomaly (HadCRUt4+HadCRUt3), adjusted down by 0.064 from Jan’98; green curve: the AMO (ESRL), trended (so as to represent actual North Atlantic SST anomalies over time) and scaled slightly down. The two black vertical lines basically delimit the 60s, across which the conspicuous ‘shift’ discussed below occurs.
What do we see? There’s that obvious covariance again. But some strange things are going on …
I’ve talked about this particular relationship before; here and here. It is rather stunning to behold the overall correlation between the GAST anomalies and the North Atlantic SST anomalies (essentially, trended AMO) in Anim.2. The general fit appears to stretch all the way back to the early beginnings of the records, and it thus very much seems to reflect a genuine physical connection of some sort. (Note well from Fig.2 above how the North Atlantic, stretching between Europe and North America, is one of the regions of the world with the best and most consistent data coverage from earliest on, fairly good all the way back to the beginning of the 20th century, but especially so from the 1920s forward.)
What really screams at you in Anim.2, however, is that major disjunction occurring across the 60s. What could possibly be the natural explanation of such a major shift? Almost perfect match from 1890 to 1960, and again from 1970 till today. So what happened during the 60s that was so exceptional!? (Well, Michael Mann and Keith Briffa’s “divergence problem” arose at just that time. That’s when the trees of the world apparently decided to follow the AMO downhill rather than global temps flat and up …)
So what is there to be said about the particular assumptions necessarily made by the compilers of the official temperature anomaly datasets in order for them to come up with the three temperature correlations (the three plots) seen above?
As you can well gather from the plots, in each case it makes sense to compare and separate between their pre and post ~1970 sections, and it also makes sense to assume that the particular relation seen between the two curves in the POST 1970 sections are closer to reality than in the PRE 1970 sections, simply because that’s when the data coverage is best by far:
- Global LAND (T2m anom.) vs. global OCEAN (SSTa):
Assumption: LAND warms (and cools) considerably faster than the OCEAN. Which seems plausible. So why not at all pre 1970?
- Northern vs. Southern Hemisphere:
Assumption: The NH warms (and cools) considerably faster than the SH. Which seems plausible. So why not at all pre 1970?
- Global LAND+OCEAN vs. North Atlantic SSTa (AMO with natural trend included):
Assumption: GAST anomalies more or less mirror trended AMO (North Atlantic SST anomalies, slightly downscaled: x0.8). Which seems plausible. So, with the same tight match all the way from the 19th century to 1960 AND from 1970 till today, why that conspicuous 0.25 K block shift across the 60s?
The official estimated average temperature data PRE 1970 simply makes no physical sense.