The best context for understanding decadal temperature changes comes from the world’s sea surface temperatures (SST), for several reasons:
- The ocean covers 71% of the globe and drives average temperatures;
- SSTs have a constant water content, (unlike air temperatures), so give a better reading of heat content variations;
- A major El Nino was the dominant climate feature in recent years.
HadSST is generally regarded as the best of the global SST data sets, and so the temperature story here comes from that source. Previously I used HadSST3 for these reports, but Hadley Centre has made HadSST4 the priority, and v.3 will no longer be updated. HadSST4 is the same as v.3, except that the older data from ship water intake was re-estimated to be generally lower temperatures than shown in v.3. The effect is that v.4 has lower average anomalies for the baseline period 1961-1990, thereby showing higher current anomalies than v.3. This analysis concerns more recent time periods and depends on very similar differentials as those from v.3 despite higher absolute anomaly values in v.4. More on what distinguishes HadSST3 and 4 from other SST products at the end. The user guide for HadSST4 is here.
The Current Context
The 2021 year end report below showed rapid cooling in all regions. The anomalies then continued in 2022 to remain well below the mean since 2015. This Global Cooling was also evident in the UAH Land and Ocean air temperatures (Still No Global Warming, Milder March Land and Sea).
The chart below shows SST monthly anomalies as reported in HadSST4 starting in 2015 through March 2022. A global cooling pattern is seen clearly in the Tropics since its peak in 2016, joined by NH and SH cycling downward since 2016.
Note that higher temps in 2015 and 2016 were first of all due to a sharp rise in Tropical SST, beginning in March 2015, peaking in January 2016, and steadily declining back below its beginning level. Secondly, the Northern Hemisphere added three bumps on the shoulders of Tropical warming, with peaks in August of each year. A fourth NH bump was lower and peaked in September 2018. As noted above, a fifth peak in August 2019 and a sixth August 2020 exceeded the four previous upward bumps in NH.
After three straight Spring 2020 months of cooling led by the tropics and SH, NH spiked in the summer, along with smaller bumps elsewhere. Then temps everywhere dropped for six months, hitting bottom in February 2021. All regions were well below the Global Mean since 2015, matching the cold of 2018, and lower than January 2015. Then spring and summer 2021 brought more temperate waters and a July return to the mean anomaly since 2015. After an upward bump in August, the 2021 yearend Global temp anomaly dropped below the mean, driven by sharp declines in the Tropics and NH.
Now in 2022 all regions remain cool. In April 2022 NH warmed slightly, offset by cooling in SH and the Tropics, so the Global anomaly remained unchanged and lower than the mean for this period.
A longer view of SSTs
The graph above is noisy, but the density is needed to see the seasonal patterns in the oceanic fluctuations. Previous posts focused on the rise and fall of the last El Nino starting in 2015. This post adds a longer view, encompassing the significant 1998 El Nino and since. The color schemes are retained for Global, Tropics, NH and SH anomalies. Despite the longer time frame, I have kept the monthly data (rather than yearly averages) because of interesting shifts between January and
July.1995 is a reasonable (ENSO neutral) starting point prior to the first El Nino. The sharp Tropical rise peaking in 1998 is dominant in the record, starting Jan. ’97 to pull up SSTs uniformly before returning to the same level Jan. ’99. For the next 2 years, the Tropics stayed down, and the world’s oceans held steady around 0.5C above 1961 to 1990 average.
Then comes a steady rise over two years to a lesser peak Jan. 2003, but again uniformly pulling all oceans up around 0.5C. Something changes at this point, with more hemispheric divergence than before. Over the 4 years until Jan 2007, the Tropics go through ups and downs, NH a series of ups and SH mostly downs. As a result the Global average fluctuates around that same 0.5C, which also turns out to be the average for the entire record since 1995.
2007 stands out with a sharp drop in temperatures so that Jan.08 matches the low in Jan. ’99, but starting from a lower high. The oceans all decline as well, until temps build peaking in 2010.
Now again a different pattern appears. The Tropics cool sharply to Jan 11, then rise steadily for 4 years to Jan 15, at which point the most recent major El Nino takes off. But this time in contrast to ’97-’99, the Northern Hemisphere produces peaks every summer pulling up the Global average. In fact, these NH peaks appear every July starting in 2003, growing stronger to produce 3 massive highs in 2014, 15 and 16. NH July 2017 was only slightly lower, and a fifth NH peak still lower in Sept. 2018.
The highest summer NH peaks came in 2019 and 2020, only this time the Tropics and SH are offsetting rather adding to the warming. (Note: these are high anomalies on top of the highest absolute temps in the NH.) Since 2014 SH has played a moderating role, offsetting the NH warming pulses. After September 2020 temps dropped off down until February 2021, then all regions rose to bring the global anomaly above the mean since 1995 June 2021 backed down before warming again slightly in July and August 2021, then cooling slightly in September. The present level compares with 2014.
What to make of all this? The patterns suggest that in addition to El Ninos in the Pacific driving the Tropic SSTs, something else is going on in the NH. The obvious culprit is the North Atlantic, since I have seen this sort of pulsing before. After reading some papers by David Dilley, I confirmed his observation of Atlantic pulses into the Arctic every 8 to 10 years.
But the peaks coming nearly every summer in HadSST require a different picture. Let’s look at August, the hottest month in the North Atlantic from the Kaplan dataset.
The AMO Index is from from Kaplan SST v2, the unaltered and not detrended dataset. By definition, the data are monthly average SSTs interpolated to a 5×5 grid over the North Atlantic basically 0 to 70N. The graph shows August warming began after 1992 up to 1998, with a series of matching years since, including 2020, dropping down in 2021. Because the N. Atlantic has partnered with the Pacific ENSO recently, let’s take a closer look at some AMO years in the last 2 decades.
This graph shows monthly AMO temps for some important years. The Peak years were 1998, 2010 and 2016, with the latter emphasized as the most recent. The other years show lesser warming, with 2007 emphasized as the coolest in the last 20 years. Note the red 2018 line is at the bottom of all these tracks. The heavy blue line shows that 2022 started warm, but in March went below all the tracks and in April remains near the bottom.
The oceans are driving the warming this century. SSTs took a step up with the 1998 El Nino and have stayed there with help from the North Atlantic, and more recently the Pacific northern “Blob.” The ocean surfaces are releasing a lot of energy, warming the air, but eventually will have a cooling effect. The decline after 1937 was rapid by comparison, so one wonders: How long can the oceans keep this up? If the pattern of recent years continues, NH SST anomalies may rise slightly in coming months, but once again, ENSO which has weakened will probably determine the outcome.
Footnote: Why Rely on HadSST4
HadSST is distinguished from other SST products because HadCRU (Hadley Climatic Research Unit) does not engage in SST interpolation, i.e. infilling estimated anomalies into grid cells lacking sufficient sampling in a given month. From reading the documentation and from queries to Met Office, this is their procedure.
HadSST4 imports data from gridcells containing ocean, excluding land cells. From past records, they have calculated daily and monthly average readings for each grid cell for the period 1961 to 1990. Those temperatures form the baseline from which anomalies are calculated.
In a given month, each gridcell with sufficient sampling is averaged for the month and then the baseline value for that cell and that month is subtracted, resulting in the monthly anomaly for that cell. All cells with monthly anomalies are averaged to produce global, hemispheric and tropical anomalies for the month, based on the cells in those locations. For example, Tropics averages include ocean grid cells lying between latitudes 20N and 20S.
Gridcells lacking sufficient sampling that month are left out of the averaging, and the uncertainty from such missing data is estimated. IMO that is more reasonable than inventing data to infill. And it seems that the Global Drifter Array displayed in the top image is providing more uniform coverage of the oceans than in the past.
USS Pearl Harbor deploys Global Drifter Buoys in Pacific Ocean