>UPDATE – September 28, 2009
The post has been updated to include brief discussions of the other datasets in which the 1945 discontinuity appear.
For those still concerned about the upcoming adjustments to the Hadley Centre’s SST data, and because the methods used by Thomspon et al as discussed in Thompson et al (2009) – High-Tech Wiggle Matching Helps Illustrate El Nino-Induced Step Changes left significant ENSO residuals, I investigated the possibility that the discontinuity revealed in Thompson et al (2008) was the result of an unrecorded El Nino event in 1943/44 or 1944/45. It appears as though global temperatures are responding to an El Nino event in 1943 and 1944, with sharp rises and falls in Sea Surface Temperature (SST) and Land Surface Temperature (LST). This could then be the cause of the severe drop in 1945 (the discontinuity) if the global temperatures were then responding to the subsequent unrecorded La Nina event. But, using the Southern Oscillation Index (SOI) as a second reference, there is no evidence of an ENSO event during those years. But this assumes that the measurement of sea level pressure in Darwin and Tahiti was not impacted by the war years.
Keep in mind, though, that the discontinuity also exists in COADS and Hadley Centre’s Marine Air Temperature data (as does the appearance of an El Nino beforehand) and in the COADS Cloud Cover and Wind Speed data.
The Hadley Centre is rewriting the Global SST record to reflect the findings of Thompson et al (2008). In the 2008 paper “A large discontinuity in the mid-twentieth century in observed global-mean surface temperature”, Thompson et al write, “The discontinuity in global-mean surface temperatures in late 1945 is evident in the unfiltered global-mean time series, but its prominence and unique character are highlighted by the removal of the ENSO and COWL-related variability (Fig. 2).” They continue in the next paragraph, “The step in late 1945 does not appear to be related to any known physical phenomenon. No substantial volcanic eruptions were reported at the time, and the nuclear explosions over Hiroshima and Nagasaki are estimated to have had little effect on global-mean temperatures…” The discontinuity is illustrated in Figure 1.
Thompson et al (2009) “Identifying signatures of natural climate variability in time series of global-mean surface temperature: Methodology and Insights” reinforces the 1945 discontinuity and provide advanced notice of the Hadley Centre’s intent. They write, “THE SST DATA CORRECTED FOR INSTRUMENT CHANGES IN THE MID 20TH CENTURY ARE EXPECTED TO BECOME AVAILABLE IN 2009, and it will be interesting to see how the corrections affect the time history of global-mean temperatures, particularly in the middle part of the century.” [Emphasis added.]
But Thompson et al (2009) acknowledged that there was also a drop in land surface temperatures at that time. They wrote, “It is worth noting that the land data also exhibit a small drop in the mean ~1945, albeit much smaller than that found in the residual SST time series (Figure 10).”
A simultaneous drop in the land data? This leads to the question…
IS THERE A MISSING ENSO EVENT IN THE SST RECORD?
Thompson et al base their claim of a 1945 discontinuity in the Global SST dataset on the fact that, after they eliminated the effects of ENSO and COWL, there was no corresponding volcanic eruption to cause the drop in temperature. This assumes the volcano records and the SST records for the NINO regions are correct. It is difficult to imagine that an explosive volcanic eruption that would have lowered global temperatures 0.3 deg C could be missed, but can the same be said about ENSO events? Also, the methods used by Thompson et al (2008) to extract the ENSO signal from the global temperature record are similar to those in Thompson et al (2009), but as illustrated in Thompson et al (2009) – High-Tech Wiggle Matching Helps Illustrate El Nino-Induced Step Changes, they left significant ENSO residuals in their adjusted data. With this in mind, I resorted to raw HADISST Global and NINO3.4 SST anomaly data and to raw CRUTEM3 Land Surface Temperature anomaly data for this post.
Figure 2 illustrates the global SST anomalies from January 1935 to December 1974 for three datasets: HADISST, ERSST.v2, and ERSST.v3b. The data has been smoothed with a 13-month filter to remove the rough edges. Note how the discontinuity now appears to be a gradual decrease from late 1944 to late 1946. Scaled NINO3.4 SST anomalies have also been provided for comparison. All three global SST datasets appear as if they’re responding to an El Nino event when none exists in the NINO3.4 SST anomaly data. Also note that there is also little agreement between NINO3.4 SST anomalies and the global SST datasets between 1943 and 1950, while global SST anomalies do appear to respond to NINO3.4 SST anomalies at other times.
In Figure 3, HADISST Global SST anomaly data from January 1935 to December 1974 are compared to CRUTEM3 Land Surface Temperature data. Again, the data have been smoothed with a 13-month filter. And again, scaled HADISST NINO3.4 SST anomalies are provided for comparison. Global land surface temperatures also appear to be responding to an El Nino event around that time. But looking at other ENSO events, global land surface temperatures do exaggerate some ENSO events but not others. And looking at the land surface temperature data for 1938/39, there was also a similar anomalous rise and fall without an El Nino event.
THE SOI INDICATES NO EL NINO EVENT IN 1943 OR 1944
Figure 4 is a comparison chart of NINO3.4 SST anomalies and inverted and scaled Southern Oscillation Index (SOI) data, the surface pressure component of ENSO. The SOI represents the sea level pressure difference between Tahiti and Darwin Australia, and would, therefore, not be impacted by the transition between SST measurement types. This assumes that the sea level pressure measurements during those war years were accurate. However, if there were shifts in sea level pressure measurements during those years as there were in SST, the anomalous rises and falls in global SST and LST data may reflect El Nino events.
Seeing how poorly global SST anomalies track NINO3.4 SST anomalies in the 1940s, the upcoming modifications by the Hadley Centre could be a good thing. Maybe they’ll try to clean up the data from 1910 to 1940 next.
Will the Hadley Centre attempt to flatten out the appearance of a 1943/44 and/or 1944/45 El Nino in the Global SST data. That rise and fall also appears in the COADS and MOHMAT Marine Air Temperature datasets, and, as illustrated above, in the CRUTEM3 LST data. Global Ocean Cloud Cover and Wind Speed datasets also indicate anomalous variations during the war years, and cloud cover and wind speed are major drivers of SST. Refer to my posts The Large 1945 SST Discontinuity Also Appears in Cloud Cover and Marine Air Temperature Data and Part 2 of The Large SST Discontinuity Also Appears in Cloud Cover and Marine Air Temperature Data.
SST and LST anomaly data used in this post is available through the KNMI Climate Explorer: