>CORRECTION: The original version of this post incorrectly stated that GISS used HADSST2 SST anomaly data for the period from 1880 to 1981. This was incorrect. I know GISS uses the HADISST dataset. Must be my age. I’ve corrected the error in the text. It does not change the overall message.
OPENING NOTE: This is not a post about the accuracy of the instrument temperature record. It also is not about the methods researchers employ to infill missing data or to reconstruct sea surface temperatures. This post is a complaint about the portrayal of compete global coverage of temperature measurements.
In his post “What NASA has to Say about Global Temperatures,” Brett Anderson at the AccuWeather Global Warming Blog provided a link to a NASA/Goddard Multimedia video “2009 Global Temperature Package: Year Tied as Second Hottest”. NASA/Goddard Multimedia describes the video as “Reporters package style video about the new 2009 global temperature data. Scientists at the Goddard Institute for Space Science found that 2009 was tied as the second hottest year ever recorded.” The scientist interviewed in the video is James Hansen of GISS.
Link to corresponding NASA/Goddard Space Flight Center webpage:
Link to transcript:
One image in the video prompted this post. Figure 1 is a screen cap of the map of Global Temperature anomalies for 1880. The image appears approximately one minute into the video as the narrator says, “And when looking back all the way to 1880, the year when precise temperature record keeping began…” I stopped the video there, so I’m unsure if the next image was the obligatory fire-engine red depiction of global temperatures in 2009.
Even though there was no temperature scale for the 1880 temperature anomaly map, I was not struck by the overwhelmingly cold portrayal of the temperature anomalies. What hit me was the total global coverage as the narrator said, “…1880, the year when precise temperature record keeping began…” NASA did not state that the image was a map of the precise global temperature anomalies for the year 1880. NASA did not state that the global coverage was complete. NASA implied both…at least to me they did. And someone who does not understand how sparse temperature measurements were in 1880, and in the decades that followed, would be led to believe that they were complete and “precise”.
Figures 2, 3, and 4 are global temperature anomaly maps available through the GISS website here:
Figure 2 shows global combined land+sea surface temperature anomalies for 1880 with the GISS standard 1200km smoothing. Coverage is far less complete than the map provided in the video. There are no readings in the Antarctic and for much of the Southern Ocean that surrounds that continent. The Arctic, Africa, and Asia show little to no available data. The only reason South America appears to have any data is the bleed-over of sea surface temperature anomalies onto land, an illusion.
A map of Land Surface Temperature anomalies for 1880 with 250 km smoothing, Figure 3, provides a better illustration of how poor the coverage of the instrument temperature record actually was in those early years. Note the carry-over of land surface temperature anomalies into the oceans. This is yet another illusion that gives the appearance of better global coverage, an illusion that is much greater with the standard GISS 1200 km smoothing (Refer to the update at the end of the post).
Other than the Arctic and Southern Oceans, the Sea Surface Temperature anomaly map for 1880 (250km smoothing) shows complete coverage. For the period of 1880 to 1981, GISS uses sea surface temperature data from the Hadley Centre (HADISST). The Hadley Centre infills data that is missing, and in 1880, there were large areas of the global oceans with no readings, as you shall see.
Sea surface temperature measurements prior to the satellite and buoys eras were primarily taken in shipping lanes. The map of other Hadley Centre SST dataset (HADSST2) in 1880, Figure 5, better reflects the areas where measurements were sparse. The Hadley Centre uses COADS Sea Surface Temperature data, adjusts it for known biases, and presents it in 5 degree grids.
The spatial coverage of the COADS Sea Surface Temperature data (2 degree grids) in 1880 is shown in Figure 6.
In Figure 7, I’ve shifted the COADS map to provide an uninterrupted illustration of the Pacific. Note the massive areas in the central and western Pacific where there were no readings in 1880.
The coverage of Sea Surface Temperature readings did not improve greatly over the decades. This can be seen in the collection of COADS SST data coverage maps available from the National Center for Atmospheric Research (NCAR), Figure 8.
In short, COADS provides its Sea Surface Temperature data in 2 degree grids, and as can be seen in Figure 7 above, there was very little coverage of the global oceans in 1880. For the HADISST data, the Hadley Centre infills vast amounts. In turn, this infilled HADISST data is used by GISS. Land Surface Temperature readings are no better in the early decades, as shown in Figure 3. The GISS 1200 km smoothing is a process that gives the illusion of better global coverage, but even with the 1200 km smoothing, land surface temperature readings are so incomplete in early years that there are major gaps in coverage over land masses, Figure 2.
Yet somehow the coverage of global temperature measurement in 1880 is shown to be complete in the NASA video. And as noted earlier, someone who does not understand how incomplete the instrument temperature record was in 1880, and in the decades that followed, would be led to believe that it was complete and “precise”.
UPDATE: Above, I did not present the GISS Land Surface Temperature map for 1880 with the 1200 km smoothing, the map that shows the illusion of more complete coverage presented by the additional GISS smoothing. Figure 9 is a gif animation that compares GISS land surface temperature and sea surface temperature maps for 1880 with 1200 km smoothing. The Land Surface Temperature map shows the bleed-over into the oceans that presents nothing meaningful. The carry-over into the oceans makes the coverage appear more complete, but it does not represent sea surface temperature. The comparative animation of LST and SST illustrates that the 1200 km bleed-over into the ocean bears no similarity to the SST readings.
The HADSST2 and COADS SST maps are available through the KNMI Climate Explorer:
The COADS coverage maps can be found at the bottom of the NCAR webpage here: