No surprise there.
But there are also periods when reported shorter-term global warming and global cooling have been decreased.
This post discusses changes to a global surface temperature dataset from the Goddard Institute of Space Studies (GISS) that is based solely on land-surface air temperature data. We’re going to compare the global surface temperature anomalies from the 1987 Hansen and Lebedeff paper Global Trends of Measured Surface Air Temperature to the current version of the same dataset from GISS, their meteorological station-based data, a.k.a. “dTs”.
Because we’re discussing Hansen and Lebedeff (1987), we’ll also take a look at their analysis of the impacts of the heat island effect on land-based global surface temperature data.
IMPORTANT: While the dTs global surface temperature dataset presented in this post is not the “official” dataset from GISS, the comparisons will provide us with rough idea of how land surface air temperature data have changed in 30 years: a 1987 vintage edition versus the current generation.
We recently presented and discussed the differences between “raw” global surface temperature and the end products from data suppliers. See the posts:
- Do the Adjustments to Sea Surface Temperature Data Lower the Global Warming Rate?
- UPDATED: Do the Adjustments to Land Surface Temperature Data Increase the Reported Global Warming Rate?
- Do the Adjustments to the Global Land+Ocean Surface Temperature Data Always Decrease the Reported Global Warming Rate?
In the two posts above that included land surface air temperature data, I provided an initial note:
If you’re expecting the adjustments to the land surface temperature data to be something similar to those presented by Steve Goddard at RealScience, you’re going to be disappointed. Steve Goddard often compares older presentations of global land+ocean data to new presentations so that we see the change in data from a decade or two ago to now. Example here from the April 8, 2016 post here. But, in this post, we’re comparing recent “raw” land surface temperature data to the current “adjusted” data, which is another topic entirely.
This post is more like the comparisons at RealScience.
The GISS surface air temperature data presented in this post is the version that excludes ocean-based (sea surface) temperature data, where GISS extends land surface air temperature data from islands and continental land masses out over the oceans. See Figure 1, which is the temperature change map (based on local linear trends) for the period of 1880 to 1985 from Hansen and Lebedeff (1987).
Keep in mind, because GISS extends data out over the oceans this is also not truly land-only data. That is especially true at high latitudes of the Northern Hemisphere where polar amplification (both positive and negative) will impact the results. See the trend maps in Plate 2 of Hansen and Lebedeff (1987).
The data presented in this post is not the “official” often-reported Land-Ocean Temperature Index (LOTI) data from GISS.
As noted on the GISS Surface Temperature Analysis (GISTEMP) webpage:
Note: LOTI provides a more realistic representation of the global mean trends than dTs below; it slightly underestimates warming or cooling trends, since the much larger heat capacity of water compared to air causes a slower and diminished reaction to changes; dTs on the other hand overestimates trends, since it disregards most of the dampening effects of the oceans that cover about two thirds of the earth’s surface.
The data included in Hansen and Lebedeff (1987) run from 1880 to 1985. See their Table 1. So the global long-term data presented in this post will end in 1985. The source of the current version of the GISS dTs data is here.
Hansen and Lebedeff (1987) also broke the data down into three periods (warming from 1880-1940, cooling from 1940-1965, and warming from 1965-1985) so comparisons will also be provided for those periods.
With the exception of three graphs (Figures 3, 9 and 11), the temperature anomalies are referenced to standard GISS base years of 1951 to 1980. I’m showing differences between datasets in Figures 3 and 9, so I’ve used the full term of the data (1880-1985) for anomalies so not to skew the results. And in Figure 11 the data have been shifted to zero the trend lines at the start of the graph to help highlight the different warming rates between the current and 1987 versions of the GISS dTs data.
LONG-TERM TREND COMPARISON
Figure 2 includes the current version of the annual GISS dTs global land-based surface temperature anomalies for the period of 1880 to 1985 and the original version from Hansen and Lebedeff (1987). The newer land-only data have a noticeably higher warming rate than the original data from Hansen and Lebedeff (1987). Based on the linear trends, the original data show a reported global warming of +0.54 deg C from 1880 to 1985, while for the newer data the reported global warming is +0.73 deg C, almost 0.2 deg C higher.
And for those interested, Figure 3 shows the difference between the original GISS dTs data and the current version…with the original subtracted from the current. As noted earlier, I’ve used the full term of the data (1880 to 1985) for the base years for anomalies so not to skew the results.
The changes to the land surface temperature data decreased the reported warming from 1880 to the mid-1930s, but there was an even greater increase in the reported warming from the mid-1930s to 1985 as a result of the changes…thus the increase in full-term warming.
THE THREE PERIODS SELECTED BY HANSEN AND LEBEDEFF (1987)
I noted above that Hansen and Lebedeff (1987) also broke the data down into three periods: a warming period from 1880-1940, a global cooling period from 1940-1965, and a warming one from 1965-1985. Referring to their Figure 6, they wrote:
The smoothed global temperature increases by about 0 .5 deg C between 1880 and 1940, decreases by about 0.2 deg C between 1940 and 1965, and increases by about 0.3 deg C between 1965 and 1980.
The choice of 1965 for a breakpoint is odd. It follows the 1963/64 eruption of Bali’s Mount Agung, and according to climate models (see illustration here) there was a severe temporary downtick in global surface temperatures at that time as a result of the volcanic eruption. If a skeptic was to choose that year as a breakpoint, someone would claim it was cherry-picked. Note also in Figure 2 how the impact of the 1976 Pacific Climate Shift stands out like a sore thumb. Regardless, we’ll use 1965 as a breakpoint for consistency with Hansen and Lebedeff (1987).
EARLY WARMING PERIOD: 1880 TO 1940
For the period of 1880 to 1940, Figure 4, the current GISS dTs data show a noticeably lower warming rate than the original data from Hansen and Lebedeff (1987). Based on the linear trend, the reported global temperature increase from 1880 to 1940 was +0.58 deg C, but with the current data, the reported global land-based surface temperature increase was +0.48 deg C, about 0.1 deg C less.
MID-20TH CENTURY COOLING PERIOD: 1940 TO 1965
Figure 5 compares the current and original versions of the GISS land-based global temperature (dTs) anomaly data for the global cooling period of 1940 to 1965. There is a striking difference in the cooling rates. During the cooling period of 1940 to 1965, the current version of the GISS dTs data show only a slight cooling of -0.04 deg C based on the linear trend, but the original data showed a much more pronounced cooling of -0.17 deg C.
LATE WARMING PERIOD: 1965 TO 1985
It should come as no surprise that the current version of the GISS land surface-based global temperature (dTs) anomaly data has a noticeably higher warming rate than the original version for the period of 1965 to 1985. See Figure 6.
Based on the linear trends, over that short 21-year period of 1965 to 1985, the reported increase in global temperatures was +0.42 deg C for the original dTs data, but with the current data, the increase was +0.52 deg C, about 0.1 deg C more. That’s a chunk over a short 21-year period.
HANSEN AND LEBEDEFF (1987) ON THE HEAT ISLAND EFFECT
I thought you might be interested in the Hansen and Lebedeff (1987) discussion of the heat island effect. They write:
An additonal [sic] issue or uncertainty about the derived global temperature change is the following: How much of the change is a result of the growth of urban heat island effects? There is abundant evidence that the growth or development of urban areas is a significant contributor to local temperature trends [Mitchell, 1953; Landsberg, 1981; Cayan and Douglas, 1984; Karl, 1985; Kukla et al., 1986]. We obtained an estimate of the magnitude of urban influence on the global temperature change of the past century by eliminating from the data set all stations associated with population centers which had more than 100,000 people in 1970. The usefulness of the test is based on the assumption that even though the urban heat island effect exists for all city sizes, the effect generally increases with population this assumption is supported by empirical studies, e.g., Mitchell . We used Table E of Davis  to identify population centers exceeding 100,000 people. Elimination of all stations within these population centers reduced the number of stations by about one third.
Removal of the city data reduced the magnitude of the global and hemispheric warmings, as illustrated in Figure 13. For example, the global temperature change in the past century was reduced from 0.7 deg to 0.6 deg C, where these numbers represent the difference between the mean 1980-1985 temperature and the mean 1880-1885 temperature. We subjectively estimate that complete correction for urban heat island effects should not reduce the global warming in the past century, defined as the temperature difference between 1980-1985 and 1880-1885, to less than about 0.5 deg C.
Basically, Hansen and Lebedeff (1987) have guesstimated that a “complete correction for urban heat island effects” for the period of 1880 to 1985 would show that the heat island effect increased global warming roughly 0.2 deg C above 0.5 deg C, or phrased differently, that roughly 0.2 deg C of the reported 0.7 deg C global warming from 1880 to 1985 was due to heat island effect.
My Figure 7 is Figure 13 from Hansen and Lebedeff (1987).
Note how correcting for the heat island effect would have increased, not decreased, the global cooling from 1940 to 1965. Also note that the correction for the heat island effect would decrease the warming rate from 1880 to 1940, which is consistent with the decrease shown in Figure 4. Due to the similarities of the “all stations” and “excludes cities” data from 1965 to 1985, there is basically no change in trend during this period due to heat island effect according to Hansen and Lebedeff (1987).
For my Figure 8, using the x-y coordinate feature of MS Paint, I’ve replicated the smoothed “excludes cities” data from Hansen and Lebedeff’s Figure 13, and included the 5-year running mean of the Hansen and Lebedeff global dTs data. Also shown in the replica are the linear trends (1880 to 1985) for the “all stations” and “excludes cities” data based on their 5-year running means.
Based on the linear trends, for the period of 1880 to 1985, the reported increase in global temperatures was +0.57 deg C for the “all stations” dTs data from Hansen and Lebedeff (1987), but with the “excludes cities” data, the increase was +0.44 deg C, or about +0.13 deg C of the +0.57 reported global warming for that period was due to heat island effect. Then again, Hansen and Lebedeff believed the heat island effect, if accounted for with a “complete correction”, would have an even greater impact.
In Figure 9, I’ve subtracted the “excludes cities” data from the “all stations” data to show the possible impacts of the heat island effect on global land surface only temperature data based on Hansen and Lebedeff (1987). So not to skew the results, as noted earlier, the anomalies were referenced to the full term of the data.
The largest permanent uptick occurred from about 1908 to the early-1910s (immediately before World War I). There were then multidecadal variations until the mid-1940s, after which the heat island effect grew at a relatively consistent rate until the late 1960s. From the late 1960s to the late 1970s, the difference between the “all stations” and “excludes cities” data decreases slightly, before accelerating until the end of the data. Sadly, the data end there.
Much has changed since 1987 with how GISS determines global mean temperature anomalies. Data from many new land surface stations have been added, (sea surface temperature data have been included for their Land-Ocean Temperature Index) and new methods were developed to account for anticipated biases. GISS currently uses land surface air temperature (and sea surface temperature) data from NOAA, which have also been adjusted.
GISS recently added a History of GISTEMP webpage. See it for a more-detailed discussion of the changes to the GISS global temperature data.
That history webpage from GISS includes a graph (their Figure 3) of how their Meteorological Station Data, a.k.a. “dTs” data, smoothed with 5-year running means filters, have changed with time. That graph is interactive, allowing you to highlight the vintage of the data and zoom in. But a version with all of the colors intensified is available here. I’ve included it as my Figure 10. The various generations of the dTs data are also available from GISS here as zipped text and .csv files.
GISS writes regarding their Figure 3 on that history webpage:
For historical reasons we also maintain a calculation of the anomalies that would result if one only used the meteorological station data. This estimate is not affected by issues in ocean data processing, but because the land is warming faster than the ocean, it has a larger trend than the land-ocean index that is now our standard product. That too has been remarkably stable over the years:
Figure 10 (Color-Intensified Version of Figure 3 from GISS History Webpage)
Clearly, the 2016 version has the highest long-term warming rate.
“[R]emarkably stable”? Figure 11 is similar to Figure 2, comparing the current version of the GISS land surface-based global temperature (dTs) anomaly data and the original Hansen and Lebedeff (1987) version for the period of 1880 to 1985. But in Figure 11, I’ve shifted the data so that the trend lines intersect with zero at 1880. That helps to highlight the different warming rates between the current and 1987 versions of the GISS dTs data. Sorry, GISS, that doesn’t appear to be “remarkably stable” to me. We must have different definitions of “remarkably stable”.
In this post we examined the differences between the current version and the original version of the land surface-based global surface temperature anomaly data presented in the 1987 paper by Hansen and Lebedeff. As noted in the opening, the dTs global surface temperature dataset presented in this post is not the “official” dataset from GISS, but the comparisons provided us with rough idea of how land surface air temperature data have changed in 30 years: 1987 versus now.
Between the 1987 version and the current version, adjustments to the GISS Meteorological Station Data, a.k.a. “dTs”, have increased the long-term warming rate from 1880 to 1985 (Figures 2 and 11) and increased the short-term warming rate from 1965 to 1985 (Figure 6).
Between the 1987 and current versions, the adjustments to the GISS dTs data decreased the warming rate from 1880 to 1940 (Figure 4) and decreased the cooling rate from 1940 to 1965 (Figure 5).
We also discussed the initial look by Hansen and Lebedeff (1987) on the impacts of heat island effect on land-based global surface temperatures.
Next we’ll look at the many versions of the GISS Land-Ocean Temperature Index. (Thanks, Gavin, for adding that GISS History webpage with all of the links to past generations of the GISS data.)