I’ve moved to WordPress. This post can now be found at Land Surface Temperature Comparison by Continent – CRUTEM, GISS, NCDC#######################
INITIAL COMMENT ABOUT FILTERING CORRECTION
I prepared a similar post less than a week ago, but the data sparseness before the 1950s bothered me. I caught my mistake: I hadn’t overridden KNMI’s default data-filter settings. I’m glad I found it, because the filtering reduced the resolution and made long-term comparisons difficult. Most data sets now run back in time from present to 1850 for the CRUTEM3 data and from present to 1880 for the GISS and NCDC data.
In the prior post, I spent too much time identifying which data supplier had the highest recent temperature. For the vast majority of the datasets, it’s CRUTEM.
I have deleted the earlier post.
The KNMI Climate Explorer website allows users to download time-series data of the three suppliers of land surface temperature (LST) anomaly data: CRUTEM, GISS, and NCDC. We’ve all seen comparisons of Global LST anomaly data similar to Figure 1.
Note 1: KNMI does not note if the GISS LST anomaly data they possess uses 250 or 1200 km smoothing, but in another data set, they identify the 250 km smoothing. Does it also apply to the LST anomaly data presented here? Unknown.
Note 2: All data in this post have been smoothed with a 37-month running-average filter. The KNMI LST anomaly data for each also have the same base years, 1971-2000, as noted above. This simplifies the comparison.
Note 3: I used the CRUTEM3 data. CRUTEM2 and CRUTEM2v are also available.
Note 4: The start year for the Southern Hemisphere, Africa, Antarctic, and South American datasets varies from the others, because they developed gaps in early years. To minimize my need to work with the data, I deleted any early data when more than a month separated it from later data. In most cases, this shortened the dataset by less than a year or two. The Antarctic data, however, begins in 1903.
Note 5: The start dates listed in the illustrations are those for the longest data set--CRUTEM.
BACK TO THE POST
Rarely do we see comparisons of the three data sets on a hemispheric basis. Refer to Figures 2 and 3.
And I’ve never seen this comparison before. In Figure 4, I’ve plotted Northern and Southern Hemisphere LST on one graph. What stands out remarkably well is when the Northern and Southern Hemisphere LSTs diverge from one another in recent years, following El Nino events.
In this post, I’ve also taken the comparisons one step farther. I’ve divided the globe by continents or as close to the individual continents as a coordinate-based system will allow. Refer to Figure 5 for the coordinates used and the areas of overlap. For the Arctic, I used 65N-90N, 180W-180E, and for the Antarctic, I used 90S-60S, 180W-180E. I also wanted a look at the hot spot in Siberia that seems to attract so much comment. Its coordinates are also shown in Figure 5.
Note 6: In the following, I provide brief descriptions of the data and some narrative, but will wait for future posts for further segmentations and comparisons.
The African LST anomalies, Figure 6, have a downward trend from the 1920s to the mid-70s. African LST anomalies then take a significant change in direction. They rise in four steps that appear to coincide with El Nino events. We’ll have to take a closer look in a future post.
We’re often told that the Antarctic has a high rate of warming over the past fifty years. In numerous prior posts, I’ve illustrated that the claims are not consistent with SST data, since SSTs for the Southern Ocean flattened in the 1980s and has been dropping like a stone since the 1990s. The ERSST.v3 and ERSST.v3b versions of the Southern Ocean SST data also show a significant drop in SST since 1880. Looking at the LST anomaly data for the Antarctic, Figure 7, reveals something else. There was a significant drop in LST anomalies in the late 1950s, so choosing 1960 as a starting date for Antarctic LST data could be described as cherry-picking. (I left the years covered by the Antarctic graph the same as the others for those who like to fix on specific years with their cursors and scroll down through the graphs.)
The Arctic comparison is shown in Figure 8. In many respects, the Arctic LST anomalies have the same characteristics as global and northern hemisphere LST anomalies. They’re simply exaggerated, or amplified. Arctic LST anomalies decrease from 1850 to the late 1800s and rise until the 1930s and 40s. There are multiyear cycles with a downward trend through the 1930s through 1960s. The rise then commences in what could be considered steps. Regardless, the majority of the rise occurs after the 1997/98 El Nino, the effects of which have lingered for many years in SST data.
The rise in Asian LST anomalies, Figure 9, is shown to be relatively flat until the mid-to-late 1970s, when they skyrocket. The great Pacific climate shift, heralding an increase in the frequency and magnitude of El Nino events, occurred then; the period of regular volcanic activity drew to an end near that time; and the AMO began its rise in the mid-1980s. Whatever it was, something shifted to cause that rise in Asian LST anomalies.
AUSTRALIA PLUS NEW ZEALAND
In Figure 10, there appear to have been two shifts in Australian LST anomalies over the data period. The first occurred in 1890, when LST anomalies dropped. They then rose approximately 0.1 to 0.2 deg C over the next 80 to 85 years. Australian LST anomalies then shifted up almost 0.3 deg C in the mid-to-late 1970s. The NCDC data divergence in 1980s and 90s is curious. It also hampers the visualization of the shift in the other two datasets.
The European data is illustrated in Figure 11. Between 1850 and the 1930s, European LST anomalies are relatively flat; that is, no major multi-decadal trends in either direction. Then there’s a change in the amplitude and frequency of the multiyear variations in the 1930s. That’s also when LST anomalies shift up (eyeballing it) approximately 0.5 deg C. They decline until the late 1980s, then take two major steps up.
The graph of North American LST anomalies, Figure 12, shows the typical four-period curve. LSTs decrease to the late 1800s, then rise until the 1940s, decrease until the 1970s, and then rise to present. These changes in trend should correspond roughly to those of Northern Hemisphere SSTs.
The South American LST anomaly curves, Figure 13, are unlike all others, most likely a result of the proximity to the Southern Ocean and the NINO areas. But I’ll save that comparison for an upcoming post.
http://i43.tinypic.com/9bgaph.jpg Figure 13
SIBERIAN HOT SPOT
Other than the major swing during the 1920s, the multiyear swings in Siberian Hot Spot LST anomalies, Figure 14, before 1960 were relatively small in comparison to the major variations from approximately 1960 to 1990. The smoothing masks the actual timing of those variations, so there’s no need to speculate. The year-to-year variations reduced in amplitude from the late 1990s to present, but LST anomalies rose significantly over that period.
But are the variations and rise in Siberian Hot Spot LST anomalies that unusual? Figure 15 is a comparative graph of European and Siberian Hot Spot LST anomalies from 1976 to present. It appears the Siberian Hot Spot LST anomalies follow many of the trends of the European LST anomalies. In this comparison, the Siberian Hot Spot LST anomalies do not appear that unusual.
A QUICK SHORT-TERM COMPARISON OF CONTINENTAL LST
Figure 16 is a busy comparative graph of LST anomalies and their linear trends for the continents from January 1976 to November 2008. (Why 1976? It was the year of the Great Pacific Climate Shift.) The European LST linear trend is approximately 0.45 deg C per decade, while the South American linear trend is approximately 0.075 deg C per decade and the Australian LST linear trend is approximately 0.11 deg C per decade or less than 25% of the European trend. The Antarctic trend is just slightly higher than that of Australia.
The land surface temperature data used in this post is available through The Royal Netherlands Meteorological Institute (KNMI) Climate Explorer website:http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere