Exaggerated claims of unusual temperatures
Summer 2023 saw a great deal of media fuss about excessive temperatures using plots such as figure 1, from the website Climate Reanalyzer. This display shows that the temperature on 4 July 2023 was 17.2°C, 1°C higher than the 1979-2000 average. This has been interpreted as corresponding to an extremely unlikely probability of 1/3.5 million, indicating imminent danger for humans. A very disturbing event in this climate alarmism was a speech by the Directeur General of the United Nations, António Guterres, saying in July 2023: ‘”The era of global warming has ended, the era of global boiling has arrived”.
Figure 1. Display of Global Mean Surface Temperature (GMST) from the Climate Reanalyzer website, showing daily temperature for individual years 1979 to 2023. The solid line is for 2023, the orange curve is 2022 and the black dashed curves are 1979-2000 average and 2 sigma ranges.
The basic problem with this type of interpretation is that it ignores the long-term trend of increasing GMST which is currently about 0.2° per decade. The analyses presented on this post take this long-term trend into account to better understand the short-term fluctuations.
Temperature fluctuations presented as residuals
The analyses here, use temperature data from the database: ERA5 monthly averaged data on single levels from 1940 to present and data between 1979 to 2023. Data is downloaded from the data base giving 2-metre temperatures (T2M), which covers land and sea, and sea surface temperature (SST) covering only sea. For some analyses the 2-metre temperature is used just over land and compared with the SST temperatures for the sea. The basic method in the analyses here, is to use residual temperatures rather than raw temperatures or temperature anomalies.
The temperature residuals are obtained, for a particular region and particular months of the year, by making a fit with a simple line to the raw temperature data. The “residuals” are calculated as the difference between the raw data and the fit. These temperature residuals are different from “anomalies” where data at a certain time have a subtraction of constant value from an average of data over a reference period. The advantage of residuals over anomalies is that the long-term trend of temperature increase is subtracted leaving just the fluctuations whereas in anomalies the fluctuations are compounded with a constant from the mean of the reference years. With residuals, as demonstrated later, the study of correlations of temperature fluctuations with other effects is facilitated.
Figure 2 (left) shows the 2-metre temperature averaged over the whole globe for the month of July from 1979 to 2023 together with a fitted line having a slope 0.19°C/decade. Figure 2 (right) shows the difference between the data and fit corresponding to the lefthand plot.
Figure 2. Left: Global Mean Surface Temperature for the month of July with a linear fit (dotted red) to data 1979-2023. Right: Residual of fit (difference data-fit).
From the lefthand plot, it can be seen that July 2023 has the highest global temperature in the plot range, at 16.9°C which is 0.30°C higher than the next highest temperature. With the long-term trend of temperature removed in the righthand plot, the value for 2023 is still the highest, at 0.35°C, but now it is only 0.12°C higher than the next highest value. The distribution of the residuals has a standard deviation (called sigma in the following) of 0.14°C. With this, the residual values for July 2023 is higher than the mean by 2.5 sigma corresponding to a probability of 0.5%. Any analysis of rare probability must take into account how many samples of data were looked at to find an anomaly. For 2023 only July was anomalous to a significant extent in 7 months. The probability that 1/7 months has a 2.6 sigma fluctuation is 3.3%. This probability is vastly more likely than the claims made from the Climate Analyzer plot of figure 1.
To see where this July 2023 temperature fluctuation occurs, an approximation to the global residual distribution is made in figure 3, which instead of the subtraction of a fit from the year 2023, an average over the previous 5 years 2018-2022 is taken and subtracted. This is the method of temperature anomalies but with a reference period much closer to the target data than is usual, to avoid large errors due to the trend.
Figure 3: For the month of July, 2023, temperature anomalies in different regions around the globe. Here the anomaly is an approximate residual as the difference between year 2023 and the previous 5-year average.
It can be seen that the fluctuations vary in sign around the globe and have a relative small granularity indicating local weather effects. The Antarctic and Arctic regions have very large fluctuations. Outside the polar regions, the largest positive excesses for July 2023 are offshore from South America and onshore and offshore in Canada.
The 2-metre temperature residuals for the whole globe are shown in figure 4 for all months of the year from 1979 to 2023. In this plot it can be seen that the July 2023 excess of 0.35°C is and August 2023 0.39 °C not the largest if all months are considered with January 1998 having a residual of 0.50 °C December 2015 having a residual on 0.45°C. Hence nothing so exceptional is happening (yet) for 2023.
Figure 4. 2-metre temperature residuals for whole globe, for all months of years 1979-2023. The points are residuals for every month and the line is a moving average over 3 months.
Figure 5 shows the distribution of sea surface temperature anomalies with the same method used for figure 3, together with 2-metre temperature over land. The land image shows again the large fluctuations over the polar regions. For simplicity in what follows the polar regions will be excluded from the study.
Figure 5. Temperature anomalies for the month of July, calculated as the difference of year 2023 and the previous 5-year average. Left sea surface temperatures, latitudes 60°S to 60°N ; Right 2-metre temperatures plotted over land, for all latitudes.
Figure 6 shows fits and residuals for SST akin to figures 2 but only latitudes 60°S to 60°N. Here the largest July residual is +0.26°C in 2023.
Figure 6. Sea Surface Temperature for the month of July with a linear fit to data 1979-2023, for latitude 60° S to 60°N. Left: temperatures, right residuals.
The spatial distribution of residuals is studied in figure 7 for five ocean regions, with residual plots for July 1979-2023 and detailed residuals in 2.5° x 2.5° latitude x longitude bins for July 2023. For these plots, the Summary Table, at the bottom of the page, summarizes the residuals for the five ocean regions. The North Atlantic has the largest fluctuation with a residual of 0.71 °C with the standard deviation of all months since 1979 at 0.20 °C, this gives a significance of 3.5 sigma and so is a rare event.
Figure 7. Distribution of SST temperature residuals in different oceans for the month of July. From top to bottom: South Pacific; North Atlantic; South Atlantic; North Pacific; Indian Ocean. Left: residuals averaged over region vs. year; right residuals for July 2023, approximated as differences 2023 – average 2018-2022 in 2.5° bins in latitude and longitude.
Figure 8 gives the 2-metre land temperature residuals which shows that although the July residual is +0.32°C it now is not a remarkable value compared to several earlier years. Figure 9 compares these land temperature residuals with sea temperature residuals for each month of the year. It can be seen that the month to month fluctuations for land regions are much larger than for the sea regions. Figure 10 plots distributions of these residuals to make this more quantitative. The standard deviations of the distribution for sea is 0.10°C while for land it is 0.32°C, showing the sea temperatures are much more stable month to month and year to year than the land temperatures.
Figure 8. Temperature (left) and temperature residual from fit (right) over land for July and latitudes 60° S to 60° N
Figure 9. Temperature residuals over land (brown) and sea (blue) for each month of the year 2010 to 2023
Figure 10. Distributions of temperature residuals for each month 1979 to 2023 for latitude 60°S to 60°N. Left: Sea; Right: Land.
The July and August 2023 residuals for the land continent regions are added to the residuals for ocean regions in the Summary Table ( at the bottom of the page).
Causes of sea surface temperature fluctuations
Two ocean circulation phenomena the El Niño-Southern Oscillation and the North Atlantic Oscillation cause sea surface temperature fluctuations. The analysis in this section, studies the correlations of the temperature residuals time-series and measures of these two ocean oscillation phenomena, indicating both contribute to the July 2023 high temperatures.
El Niño-Southern Oscillation (ENSO)
The ENSO (El Niño/La Niña) effect is characterized by the Oceanic Niño Index (ONI) described at ENSO. Figure 11 shows the temperature residual in the “ENSO 3.4” region of the Pacific Ocean (5°S to 5°N, 120°W to 170°W) where the ONI index is calculated. The plot shows the temperature residuals together with bars indicating when the temperature exceeds the ONI definition limits of +0.5°C and -0.5°C.
Figure 11. Sea surface temperature residuals for all months of year 1979-2023 in the ENSO 3.4 region of the Pacific Ocean. The red and purple lines indicate when the residuals exceed the ONI definition of +0.5C° and -0.5°C. The residuals are plotted with a moving average over 3 months.
The SST residuals over the latitude region 60°S to 60° N are shown in figure 12 compared to the ONI index for El Niño/La Niña episodes. The plot shows clear correlations between high temperature residuals with El Niño episodes and low residuals with La Niña episodes.
Figure 12. SST residuals for all months in year from 1979 to 2023, for latitudes 60°S to 60°N. The plot uses a moving average over 3 months to smooth the data. The bars top and bottom indicate when the ONI index exceeds +-0.5°C. The highlighted areas indicate the El Niño (red) and La Niña (purple) episodes lasting more than 7 months.
As well as affecting ocean temperature, ENSO also affects land temperatures. Quantitative measures of the correlations are given in figure 13 which plots the temperature residuals in the South Pacific and in South America against the ENSO 3.4 region residuals.
Figure 13. Correlation of temperatures residuals in two regions, sea and land, against residuals in the ENSO 3.4 region. Left: South Pacific Ocean region; Right: South America land regions. Each point is one month for years 1979 to 2023 with the residuals smoothed with a moving average over 3 months. The correlation is demonstrated by the clustering around the diagonal and quantified by the correlation coefficient.
The Pearson Correlation Coefficient of the distributions of figure 13 are 0.68 for the South Pacific Ocean and 0.60 for the South America continent. The Summary Table (at the bottom of the page) summarizes these correlations for all oceans and continents. From this table it is clear that the influence of ENSO is limited to the Pacific Ocean and the Indian Ocean, with also an influence on the South America land region.
North Atlantic Oscillation (NAO)
The North Atlantic Oscillation affects temperatures in the North Atlantic Ocean. The effect is quantified by the NAO Index which is published monthly by NOAA (North Atlantic Oscillation, noaa.gov) and this is plotted for recent years 2010-2023 in figure 14.
Figure 14. The North Atlantic Oscillation (NAO) index, points are monthly NAO values.
This NAO index is compared in figure 15 (left) with residuals in the North Atlantic.
Figure 15. Left: SST residual for North Atlantic (blue) and NAO index (green). The monthly data for both is smoothed with a moving average over 12 months. Note the negative value of the NAO index is plotted to more clearly see the correlation where the temperature is high when the NAO index is large negative. Right: The spatial distribution of SST anomalies for July 2023 in the North Atlantic, using the method used in figure 3.
The anomaly distribution in figure 15 (right) shows the high temperature residuals are in two regions, further figure 3 shows some regions of Africa have high temperature anomalies in July 2023. Figure 16 (left) shows the correlation of the North Atlantic region with the NAO index and figure 16 (right) shows the correlation of the land region of Africa and the Middle-East, indicating that the high temperature in parts of Africa are indeed caused by NAO.
Figure 16. Correlation of temperatures residuals in two regions, sea and land, against the NAO index. Left: North Atlantic Ocean region; Right: Africa and Middle-East land regions. Each point is one month for years 1979 to 2023 with the residuals smoothed with a moving average over 3 months. The correlations are negative because the temperature is high in these regions when the NAO index is negative.
The NAO effect in the North Atlantic which moves circulation systems around in the ocean according to the NAO index and leads to complex variations in temperatures. The temperature anomalies are positive and negative in different regions at different times as seen in, as in figure 15 (right) for July 2023 and the temperature correlations with NAO regions are corresponding different in different regions. Figure 17 gives three examples of correlations in different regions of about 10° x 10° of latitude x longitude. These three regions have correlation coefficients with the NAO index with different signs as follows: North West ( -0.40), Middle West ( +0.58) and South East ( -0.59). The average over the whole of the North Atlantic gives a correlation of the temperature residual with the NAO index of -0.51.
Figure 17. Left: Correlation of SST temperature residuals with NAO index for in three regions of North Atlantic. One point per month 1979 to 2023 with smoothing from moving average over 12 months. Right: Temperature anomalies for July 2023 (as figure 3) for the North Atlantic regions corresponding to the correlation plots.
NAO correlation coefficients for all oceans and continents are given in the Summary Table below.
Other contributions to temperature rises
Berkeley Earth has made the cartoon in figure 18 to illustrate the various contributions to the factors contributing to the recent temperature rises.
Figure 18. Illustration of contributions to temperature rise from Berkeley Earth.
Beyond AGW and ENSO/NAO there are three smaller causes of temperature increases. Solar activity quantified by the number of sunspots has an 11 year cycles and year 2023 is on a rising trend of activity which likely causes a small temperature rise. Estimates have been made for the eruption of the undersea volcano Hunga Tonga in January 2022 which injected large amounts of water vapour into the Stratosphere/ The third effect is the reduction of SO2 emissions from marine shipping due to new pollution regulations.
The five effects in figure 18 have magnitudes as follows:
- Man-made Global Warming: about 0.2°C/decade averaged over globe.
- Ocean Oscillations ENSO and NAO: up to about ±0.2°C for a year or two.
- Solar Activity: debatable but maybe ±0.03°C in 11 year cycles.
- Hunga Tonga volcano: approximately +0.04 °C for several years.
- Marine shipping reduction of emissions of SO2 : approximately +0.02°C.
Summary and Conclusions
Summary of data
The analyses described on this page have studied the time-series of 2-metre and sea surface temperature from the ERA5 database from 1979-2023 with the objective to understand the magnitude and causes of temperature excursions in the summer of 2023. The page uses residual to quantify the excursions rather than the usual anomalies for reasons described extensively above. The Summary Table below gives the temperature in summer 2023 together with the results of a correlation analysis between the residuals and two ocean circulation oscillations, ENSO and NAO.
Summary Table. For regions of the globes as indicated, a summary of temperature residuals for summer 2023 and correlations of temperature residual from 1979 to 2023 with known natural weather fluctuations. The correlation coefficients are with the ocean oscillation ENSO and NAO as described above. For the residuals, the significance is given in terms of the standard deviation of the residual distribution for the 535 months fitted in the date range (e. g. figure 10). Significant residuals and also significant correlations are highlighted.
Figure 19 collects some of the results already presented on this page, but with a shorter time range to make clearer the changes in 2023. The top window shows land and see residuals indicating a slow significant rise in sea temperature residuals starting in Jan 2023 and a more rapid rise in land temperature. As already demonstrated in figure 10, the land residuals have 3 times the magnitude of month-to-month fluctuations as sea fluctuations, so these land residual changes are not very significant. The bottom window shows the ENSO 3.4 region residuals and NAO indices, both of which are increasing in 2023 (note that the negative of the NAO index is plotted because this is what correlated with increasing temperatures). With the clear evidence of correlations of these indices with temperature as shown above, this demonstrates that these effects are the cause of the 2023 temperature increases.
Figure 19. Summary of data presented in this post. The last points are August 2023. Top: Temperature residuals over global land regions (brown) and global sea regions (blue) both for latitude 60°S to 60° N. Bottom: Temperature residual in ENSO 3.4 region of Pacific used to define ONI (red) and negative value of NAO Index (Green).
The summer 2023 temperature fluctuations are mainly due to ENSO and NAO phenomena in the oceans. Both these phenomena give peaks in temperature every few years but with an irregular periodicity. A coincidence of these two effects gives a larger than typical upward fluctuation in 2023. The “Global Boiling” response was misleading propaganda from the UN DG who has great global responsibility and who ought to know better.
Links to Similar Studies
This guest post is copied from the website of the author, John Carr, climate-and-hope.net , which contains information and data analyses related to climate change and electricity generation.