I’ve moved to WordPress. This post can now be found at Part 2 of The Large SST Discontinuity Also Appears in Cloud Cover and Marine Air Temperature Data##################
And A Look At Two More COADS Datasets - SST & Wind Speed
In The Large 1945 SST Discontinuity Also Appears in Cloud Cover and Marine Air Temperature Data, I illustrated that the 1945/46 shift SST anomalies also appears in Cloud Cover and Marine Air Temperature data. In this post, I provide comparative graphs of HADISST anomaly data from January 1930 to December 1959 and the Marine Air Temperature and Cloud Cover datasets. I’ve also added two more COADS datasets, SST and wind speed. The wind speed data shows a shift in 1945 but of the opposite sign.
Note: I am not the first blogger to note that the shift also occurs in the Marine Air Temperature data. In the ClimateAudit post Nature "Discovers" Another Climate Audit Finding, at comment 120, Fred Moolten writes, “SST measurements are only one means of recording temperature over the oceans. A separate method involves marine air temperature (NMAT) measurements that are less robust but generally match SST quite well. These, of course, are not subject to errors involving insulated or uninsulated buckets or engine intakes. If one examines NMAT records, they faithfully parallel the SST trends at most times during the twentieth century. During the 1940s, this is also true, but only partially. In particular, they exhibit the same dip as the SST measurements, implying that this dip cannot be fully explained by SST artefact. However, the SST dip slightly exceeds the NMAT dip, and it is this small excess, rather than the entire descending limb of temperature, that is most plausibly attributable to artefact. (In passing, one should note that land temperatures also exhibit a peak and decline, but to a much lesser extent).”
And though one might exist, I have found no mention of the shift in cloud cover data at the same time in any blogs.
Figure 1 illustrates Global SST Anomalies [HADISST] and Global Marine Air Temperature [MOHMAT4.3] data from January 1930 to December 1959. One would believe based on the similarities between the two datasets that they were used to infill one another during the war period when data collection was sparse. Note that the shifts in 1945/46 are of similar magnitudes.
The COADS Air Temperature Anomalies have a much greater upward surge in the early 1940s than the Global SST Anomalies [HADISST], Figure 2, with a much greater drop in 1945/46.
Ocean Cloud Cover, Figure 3, also shows an early 1940s surge. The 1945 downward shift Ocean Cloud is exaggerated.
COADS SST ANOMALY DATA
The COADS SST Anomaly dataset available through the KNMI Climate Explorer webpage appears to be the raw data prior to any adjustments. I say this because the COADS SST anomaly data still includes the anomalous offset in the data prior to 1941, which was adjusted for by Folland et al back in the 1990s. Refer to C. K. Folland and D. E. Parker, 1995, CORRECTION OF INSTRUMENTAL BIASES IN HISTORICAL SEA SURFACE TEMPERATURE DATA, Q.J.R. Meteorolol. Soc. 121, 319-367
These adjustments were further discussed by Steve McIntyre of ClimateAudit in 2005:
In looking at the comparison of HADISST Global SST anomalies and the COADS Global SST Anomalies, Figure 4, it’s very clear that a partial correction of the sudden shift in 1945 has already been made.
And that raises the question…
HAS THE HADLEY CENTRE ALREADY MADE THE ADJUSTMENTS FOR THE DISCONTINUITY NOTED IN THE THOMPSON ET AL PAPER?
The answer is no for the Hadley Centre’s HADSST2 version of SST anomalies. I had an older copy (my file was dated October 2007) of the HADSST2GL data, one that was published before the Thompson et al paper. Unfortunately, I do not have an older copy of the HADISST data that I’ve used as reference throughout this post. But it seems unlikely that the Hadley Centre would update one SST dataset and not the other. Figure 5 compares the October 2007 HADSST2GL data with the most recent update, February 2009, for the period of January 1930 to December 1959. I had to reduce the weighting of the February 2009 dataset (red curve) in order to show the October 2007 version of the dataset (the blue curve). There are identical.
COADS WIND SPEED ANOMALY DATA
So far the rises and falls in all of the datasets have been in the same direction (an increase in the early 1940s followed by a sharp decrease in the mid-1940s) and they have all occurred at approximately the same times. This would lead one to believe the all of the datasets may have been relied on to help infill the others during the period of World War II, intermixing them. If not, then the unlikely explanation would be, maybe (BIG MAYBE) there was a statistical quirk that was responsible for all of the shifts. Or those shifts actually occurred.
To confuse matters, there’s the COADS Wind Speed Anomaly dataset. Refer to Figure 6. Note that there is an increase in the Wind Speed anomaly data from January 1930 to December 1959, which is similar to all of the other datasets illustrated in this post. They all have positive trends from 1930 to 1959. But the shift in the Wind Speed anomaly data in the early 1940s is negative—wind speed drops during that time. Then in 1945 the Wind Speed anomaly increases sharply.
The opposing fall then rise is easily visible in the comparative graph of HADISST Global SST Anomaly data and COADS Global Wind Speed data, Figure 7. I’ve scaled and ranged the Wind Speed anomaly data for illustration purposes only.
The fall then rise in the Wind Speed Anomaly data does appear to lag the SST anomaly data, but why would it only happen during that time period. Curious.
Steve McIntyre’s posts and the comments on the subject of SST anomaly adjustments are quite insightful. The following are links to them:
There may be other posts at ClimateAudit about these adjustments.
The HADISST, the MOHMAT3.4, the COADS Air Temperature, the COADS Cloud Cover, the COADS SST, and the COADS Wind Speed datasets are available through the KNMI Climate Explorer website.
And the HADSST2GL data is available here: