>UPDATE (March 31, 2010): Also refer to the post GISS Acknowledges Addition of ERSST.v3b Data To Their GISTEMP “Options”
Like many people, as soon as one of the climate change blogs (most often Lucia’s The Blackboard) announces that GISS has posted their monthly GISTEMP global temperature anomalies, I visit the GISS Global Maps webpage to pull up the most recent map. I had stopped by the webpage a few times last week to look at maps of SST anomaly trends for a specific time periods, and there wasn’t anything out of the norm. But early this week I discovered that GISS has added another choice to the Ocean Data Source drop-down menu. Refer to Figure 1. They’ve added “NOAA/ER_v3b”. I don’t remember the option last week, and I don’t believe I would have overlooked it. Did they add it when they updated with their January 2010 data?
BUT GISS DOES NOT USE ERSST.v3b SST DATA
GISS uses HADISST SST data from January 1880 to November 1981 and then splices the NCDC’s OI.v2 SST data to it for the period after December 1981. That raises the question, Is GISS Changing SST Datasets? I found no mention of it on the GISS update page. There was no mention of it on GISTEMP homepage or the current analysis page. So the answer to that question is, dunno.
WHY ADD IT TO THE WEBSITE?
But if they weren’t planning to change SST datasets, why add the ERSST.v3b SST anomaly data to the webpage? I’m not going to speculate on all of the possible reasons for adding the ERSST.v3b SST data to the map-making page, but the only thing that makes sense to me is that they’re planning to switch datasets.
PROS FOR ALARMISTS IF GISS WERE TO CHANGE SST DATASETS
A better question, why would GISS change SST datasets? The OI.v2 SST data from the NCDC, presently being used by GISS after December 1981, is satellite based and is said to have a cool bias when compared to data based on ship and buoy readings. Figure 2 compares global SST anomalies for the ERSST.v3b and OI.v2 SST datasets from January 1982 to December 2009. The linear trend for the OI.v2 SST anomaly data is 0.095 deg C/decade, while the ERSST.v3b data shows a linear trend of 0.11 deg C/decade. If they change to the ERSST.v3b data, they’d add to the linear trend of the SST data. Global temperatures would rise even more.
Note: Recall that the original version of ERSST.v3 SST data included the satellite-based data, and that the NCDC went to great lengths to include that satellite data. In fact, a good portion of Smith et al (2008) Improvements to NOAA’s Historical Merged Land-Ocean Surface Temperature Analysis (1880-2006), dealt with the benefits of the satellite data. But the satellite data was deleted due the cold bias. The NCDC wrote in explanation in a November 14, 2008 attachment to their main ERSST.v3b webpage, “In the ERSST version 3 on this web page WE HAVE REMOVED SATELLITE DATA from ERSST and the merged product. The addition of satellite data caused problems for many of our users. Although, the satellite data were corrected with respect to the in situ data as described in reprint, there was a residual cold bias that remained as shown in Figure 4 there. The bias was strongest in the middle and high latitude Southern Hemisphere where in situ data are sparse. THE RESIDAL BIAS LED TO A MODEST DECREASE IN THE GLOBAL WARMING TREND AND MODIFIED GLOBAL ANNUAL TEMPERATURE RANKINGS.” [Caps added.] The link for that quote is here:
If we look at the pre-satellite data (January 1880 to December 1981), Figure 3, the linear trends of the HADISST (presently used by GISS) and the ERSST.v3b data are basically the same, at approximately 0.03 deg C/decade, or 0.3 deg C per century.
Note, however, prior to 1940, the variations in the ERSST.v3b SST anomaly data are much greater than the HADISST data. So alarmists who wanted to take advantage of that fact would begin their trend analysis at the commonly used start of the 20th Century. See Figure 4. The ERSST.v3b data would add about another 0.12 deg C/century.
CONS FOR ALARMISTS
The ERSST.v3b data would have some disadvantages for alarmists. The dip and rebound from the late 19th to the early 20th centuries is much greater with the ERSST.v3b data. As shown in Figure 5, there was little difference between the SST anomalies in the early 1880s and those in the 1970s.
The dip and rebound in the HADISST data presently used by GISS is not as significant. Refer to Figure 6.
In short, GISS would have to increase the effects of volcanic aerosols in their Model E to account for the greater variability of the ERSST.v3b SST data prior to 1940.
THE SPATIAL COVERAGE IS BASICALLY THE SAME
The presentation of ocean coverage for the two datasets is similar. This can be seen in the two linear trend maps created via the GISS map-making webpage, Figures 7 and 8. Though it’s not visible in those maps, for the pre-satellite period, the HADISST data is presented in 1 deg latitude and longitude grids, while the ERSST.v3b data is in 2 deg grids. For the satellite era, in areas with sea ice and in other locations where ship and buoy readings are sparse, the OI.v2 data has better coverage than the ERSST.v3b data.
A NOTE ON SST RECONSTRUCTIONS
Every time I see a map of SST anomalies, or long-term trends in SST anomalies, I remind myself that much of the data is infilled by the Hadley Centre and NCDC. Long-term SST data from the Hadley Centre and NCDC is based on COADS data. And as illustrated with the maps of typical (boreal winter) monthly (January) SST anomalies for 1910, 1930, 1950, and 1970, Figure 9, the coverage prior to the satellite era was poor and in many areas non-existent. In January 1910, the vast majority of the Pacific and all oceans south of the equator were unsampled. In time, coverage improved, but was pretty much limited to shipping lanes. As recent as January 1970, there are still very few readings south of 45S and for a good portion of the Southeast South Pacific.
AND THE BOTTOM LINE
Comparing the Global temperature anomalies for GISTEMP based on the ERSST.v3b SST data (Figure 10) and the present merged HADISST/OI.v2 data (Figure11), if GISS were to change SST datasets, the January 2010 Global Surface Temperature anomaly would rise 0.03 deg C.
And the rise in Global Surface Temperatures based on the linear trends from 1900 to 2009 would increase 0.1 deg C. Refer to Figures 12 and 13.
SOME CLOSING COMMENTS
Is GISS planning to switch SST datasets? I see no other reason for including the ERSST.v3b data in GISS map-making webpage. The switch would increase temperature trends but it would do so based only on the different methods used to reconstruct and measure the sea surface temperatures of the global oceans.
Note that the highest global ERSST.v3b SST anomaly at the peak of the 1997/98 El Nino is ~0.52 deg C in Figure 5, but is ~0.25 deg C in Figure 2. This is, of course, a result of the different base years used for the anomalies. The default base years used by the KNMI Climate Explorer are 1971 to 2000. If the period selected does not cover all of the years of the default base period, another period is used.
Note also, on the map of the ERSST.v3b data, Figure 8, some of the areas with the highest trends are the areas with the poorest coverage in the early decades. That will be the basis for a future post.
ERSST.v3b, HADISST, OI.v2 SST and COADS data are available through the KNMI Climate Explorer: