We now have updates from both the GISS and NOAA surface temperature records for August.
Both show another warm month.
Second warmest August in the GISS record, third warmest for NOAA.
There isn't much of an actual difference between the two.
Comparison of August Anomalies between GISS and NOAA.
This has made vary little difference to either the current average, or my predictions for 2017.
For GISS the average for 2017 up to August is 0.93 °C, compared with 0.94 °C up to July.
The prediction for 2017 is 0.91 ± 0.07°C, unchanged from the prediction in July apart from slightly more confidence.
For NOAA the average for 2017 up to August is 0.88 °C, compared with 0.89 °C up to July.
The prediction for 2017 is 0.86 ± 0.07°C, compared with a prediction of 0.87 ± 0.08;deg;C in July.
The differences in predictions for GISS and NOAA are due to rounding to two decimal places. In both cases the predictions dropped by around 0.001 °C.
Annual Observations and Forecast for 2017 for GISS Surface Temperatures, based on data up to August 2017.
Annual Observations and Forecast for 2017 for GISS Surface Temperatures, based on data up to August 2017.
The slight narrowing of the prediction intervals has firmed by the probabilities a bit, and it looks increasingly likely that 2017 will be 2nd warmest in the GISS record, and 3rd warmest in NOAA's.
Below are the tables showing the probability, based on data up to August, of each data set finishing at the specified position, and the change from the predictions made based on data from July.
This post is part of my ongoing global temperature analysis and prediction series for 2017.
RSS have now released the figures for August and it was another warm month.
At 0.71°C, it was the warmest August in the RSS 4 record, and the warmest monthly anomaly since September 2016.
(Incidentally, the old RSS 3.3 version of the data has August at 0.55°C,
making it the second warmest August in that data set, only beaten by August 1998 at 0.57°C.)
Here's the entire RSS 4 series, with 12 month rolling average.
RSS 4 TLT Anomalies in °C compared with 1981 - 2010 base period.
Blue - linear trend.
Red - 12 month moving average.
The trend since the start in 1979 is currently 1.85 °C / Century.
Updated Forecast for 2017
The average anomaly for the year 2017 up to August is 0.60°C, compared to 0.58°C for July.
My forecast for 2017 is 0.58 ± 0.07°C, compared to last month's forecast of 0.57 ± 0.08°C.
Annual temperatures for RSS 4, and forecast for 2017 based on data up to August 2017.
This slight increase edges the forecast ahead of 1998.
This means that there is now a better than 50% chance of 2017 being the second warmest year in the RSS 4 data set.
This table lists the probability of 2017 finishing in each position:
Rank
Year
Anomaly
Probability
Change
1st
2016
0.74 °C
0.0%
+0.0%
2nd
1998
0.58 °C
55.2%
+13.6%
3rd
2010
0.56 °C
20.6%
+0.8%
4th
2015
0.54 °C
15.4%
-4.2%
5th
2005
0.42 °C
8.8%
-10.2%
6th
2014
0.41 °C
0.0%
+0.0%
2017 is most likely to finish in 2nd place, but there is a reasonable chance that it could finish as low as 5th.
Citations
Analysis for this blog post used R.
R Core Team (2015). R: A language and environment for statistical computing.
R Foundation for Statistical Computing, Vienna, Austria. URL
https://www.R-project.org/.
Graphics produced with the GGPlot package
H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New
York, 2009.
I'm going to try to update some temperatures as they come in (No promises).
Dr Roy Spencer has announced the value for his UAH 6, and it's a rise, up to 0.41 °C, compared with the 1981 - 2010 base period.
This makes it the 3rd warmest August in the UAH 6 data set, only slightly cooler than August last year,
and 0.11 °C cooler than August in 1997.
To put this into some context this graph compares temperatures around the two major El Niños of 1997 and 2016.
What this shows is that whilst the El Niño spikes were very similar in terms of absolute temperature, the years preceding the 2016 El Niño were warmer than the years preceding that of 1998, and so far the year following 2016 has been warmer.
The Forecast
The increase in temperature has pushed the average temperature for 2017 to 0.32 °C, up from 0.30 °C in July.
The forecast reflects this with a rise from 0.30 ± 0.08 in July to 0.32 ± 0.06 °C.
Annual temperatures for UAH 6, and forecast for 2017 based on data up to August 2017.
UAH 6 - Probability of 2017 Ranking
Rank
Year
Anomaly
Probability
Change
1st
2016
0.51 °C
0.0%
+0.0%
2nd
1998
0.48 °C
0.0%
+0.0%
3rd
2010
0.33 °C
30.4%
+11.0%
4th
2015
0.26 °C
63.8%
-1.4%
5th
2002
0.22 °C
5.6%
-8.7%
6th
2005
0.20 °C
0.2%
-0.6%
7th
2003
0.19 °C
0.0%
-0.2%
8th
2014
0.18 °C
0.0%
+0.0%
This slight rise in the forecast has edged UAH 6 close to the third place set in 2010.
As a result, whilst it's still most likely that 2017 will be 4th warmest, the chances of finishing 3rd have increased somewhat.
Analysis of UAH 6
I'm going to look in some detail at the trend of UAH 6 and what it means for Christopher Monckton's Great Pause.
I'm using UAH 6 for this, as it is the closest current data set to Monckton's preferred RSS 3.3 set.
This is still being published but has now been superseded by version 4, which shows rather more warming.
To be clear, the only reason I am using UAH 6 is that it shows by far the least warming of any data set.
First of all lets look at UAH 6 over its lifetime.
UAH 6 TLT Anomalies in °C compared with 1981 - 2010 base period.
Blue - linear trend.
Red - 12 month moving average.
The peak in 2016 is very similar to that of 1998, but as mentioned above the years before and after have been much warmer.
This can be more clearly seen by looking at a 2 year rolling average.
UAH 6 TLT Anomalies in °C compared with 1981 - 2010 base period.
Blue - linear trend.
Red - 24 month moving average.
UAH 6 continues to set new records averaged over the last 24 months, and is currently 0.15°C warmer than the high point in 1998.
For a clearer picture of the climate, here's the same graph using 20 year averages.
UAH 6 TLT Anomalies in °C compared with 1981 - 2010 base period.
Blue - linear trend.
Red - 20 year moving average.
Whatever the truth behind the Pause, it has had very little impact on the long term trend.
The trend over the entire life of UAH 6 is currently 1.25°C / Century (a lot less than in other data sets).
The trend up to 1997 was just 0.93 °C / Century. A year later it had risen to 1.64 °C / Century.
None of these trends are significantly different from each other.
UAH 6 TLT showing different trend periods.
Blue - trend to August 2017.
Green - trend to 1998.
Red - trend to 1999.
Christopher Monckton's Great Pause
A reminder that Christopher Monckton's Great Pause, which he used to annotate every month in Watt's Up With That, was a ludicrous example of cherry-picking, in which he would take one data set (RSS 3.3), work out the trend starting at each month up to the current month, and then choose the earliest starting month yielding a negative trend.
He would then claim that this represented the start of the Great Pause (his capitalization), and tell his readers exactly how many months it had lasted.
It didn't matter that the actual start would move backwards and forwards each month.
Using Monckton's definition there is currently no Pause.
Or at least not one lasting more than 5 years - taking the strictest definition there has been a pause since May 2015, but I would hope even Monckton has more sense than to talk about a 2 year pause.
Limiting ourselves to trends of at least 5 years, the closes to Monckton's pause is to cherry pick the starting month that gives the slowest rate of warming.
This is December 1997, with a trend of 0.58 °C / Century.
UAH 6 monthly anomalies - with trend lines for the periods before and after December 1997.
But we can just as easily cherry-pick starting dates that will give extremely fast rates of warming.
For example, since April 2005, UAH 6 has been warming at over 2 °C / Century.
Since September 2006, it has been warming at over 3 °C / Century,
and since August 2007, the rate of warming has been more than 4 °C / Century.
Of course all of these steep rises rely on the warm El Niño last year, just as the pause relies on starting with a very warm El Ni˜o.
And none of these Great Warmings are statistically significant, any more than Monckton's Great Pause was.
And, as with the Pause the lines don't meet up.
UAH 6 monthly anomalies - with trend lines for the periods before and after August 2007.
When Will the Pause Return?
Curiously, Christopher Monckton asked the same question back in January, but neglected to even attempt an answer.
Whilst it might have been possible for a brief return in a year or so, if there had been a strong La Niña this year, the fact that temperatures have continued to be so high throughout this year makes any return in the next couple of years highly unlikely,
and unless something very dramatic happens I doubt we will ever see a return to a pause stretching back to before 1998.
To simplify things lets use annual averages, rather than monthly.
Now, assume 2017 does finish up within my prediction interval, but at the low end, i.e. around 0.26 °C, what would it take for the Great Pause to return by the end of 2018?
For this to happen, 2018 would have to average -0.22°C.
That would be a pretty phenomenal drop in temperature in a single year.
It would require 2018 to be the coldest year since 1992.
Alternatively, if 2018 was to drop to -0.1 °C, equal to the coldest single year during the Great Pause, and then stay at that level for the next few years, we still wouldn't see the return of the Pause until 2020.
In other words, for there to be a return of Monckton's Pause anytime in the next few years would require a period of cold unlike anything seen during the last 25 years.
Citations
Analysis for this blog post used R.
R Core Team (2015). R: A language and environment for statistical computing.
R Foundation for Statistical Computing, Vienna, Austria. URL
https://www.R-project.org/.
Graphics produced with the GGPlot package
H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New
York, 2009.
Here are my predictions for 2017, updated for data up to July.
First the usual disclaimer, all these predictions are my own, based on simple statistical inference.
They are not based on climate models or forecasts.
They are just a bit of fun, and a chance for me to mess about with R.
See here for the June predictions.
The main headline from July is that nothing much has changed.
The predictions for all data sets are the same as they were for June (to the nearest hundredth of a degree).
The confidence in these predictions has increased, so the probabilities of the likely positions will have changed.
The follow shows the details for each of the main data sets.
For each set there is a description of the expected value for 2017, followed by a graph showing the previous annual anomalies starting in 1979, and the 95% prediction interval for 2017.
Finally there is a table listing the current warmest years for the set along with the probability of 2017 finishing at each position, and the change since the June predictions.
All anomalies are compared to the base period used the the providers of each data set, so should not be compared with each other.
The Predictions
UAH 6
The anomaly for UAH 6 up to July is 0.30 °C.
The predicted anomaly for 2017 is 0.30 ± 0.07 °C. This would make 2017 the 4th warmest year on record.
Rank
Year
Anomaly
Percent
Change
1st
2016
0.51 °C
0.0%
+0.0%
2nd
1998
0.48 °C
0.0%
+0.0%
3rd
2010
0.33 °C
19.4%
-3.3%
4th
2015
0.26 °C
65.2%
+9.9%
5th
2002
0.22 °C
14.3%
-4.3%
6th
2005
0.20 °C
0.8%
-1.2%
7th
2003
0.19 °C
0.2%
-0.5%
8th
2014
0.18 °C
0.0%
-0.3%
RSS 4
The anomaly for RSS 4 up to July is 0.58 °C.
The predicted anomaly for 2017 is 0.57 ± 0.08 °C. This would make 2017 the 3rd warmest year on record.
Rank
Year
Anomaly
Percent
Change
1st
2016
0.74 °C
0.0%
-0.1%
2nd
1998
0.58 °C
41.7%
-0.3%
3rd
2010
0.56 °C
21.2%
+3.0%
4th
2015
0.54 °C
18.2%
+1.7%
5th
2005
0.42 °C
18.9%
-4.2%
6th
2014
0.41 °C
0.0%
+0.0%
7th
2003
0.39 °C
0.0%
-0.1%
8th
2002
0.38 °C
0.0%
+0.0%
GISSTemp
The anomaly for GISSTemp up to July is 0.94 °C.
The predicted anomaly for 2017 is 0.91 ± 0.08 °C. This would make 2017 the 2nd warmest year on record.
Rank
Year
Anomaly
Percent
Change
1st
2016
1.00 °C
2.3%
-1.0%
2nd
2015
0.87 °C
83.8%
+3.9%
3rd
2014
0.73 °C
13.9%
-2.9%
4th
2010
0.70 °C
0.0%
+0.0%
5th
2005
0.67 °C
0.0%
+0.0%
6th
2007
0.64 °C
0.0%
+0.0%
7th
2013
0.64 °C
0.0%
+0.0%
8th
2009
0.63 °C
0.0%
+0.0%
NOAA
The anomaly for NOAA up to July is 0.90 °C.
The predicted anomaly for 2017 is 0.87 ± 0.08 °C. This would make 2017 the 3rd warmest year on record.
Rank
Year
Anomaly
Percent
Change
1st
2016
0.95 °C
2.7%
-1.7%
2nd
2015
0.91 °C
15.2%
-1.3%
3rd
2014
0.75 °C
82.0%
+3.2%
4th
2010
0.70 °C
0.1%
-0.2%
5th
2013
0.67 °C
0.0%
+0.0%
6th
2005
0.66 °C
0.0%
+0.0%
7th
2009
0.64 °C
0.0%
+0.0%
8th
1998
0.63 °C
0.0%
+0.0%
HadCRUT
The anomaly for HadCRUT up to July is 0.73 °C.
The expected anomaly for 2017 is 0.72 ± 0.08 °C. This would make 2017 the 3rd warmest year on record.
Rank
Year
Anomaly
Percent
Change
1st
2016
0.77 °C
7.5%
-4.8%
2nd
2015
0.76 °C
5.7%
-1.0%
3rd
2014
0.58 °C
86.7%
+5.9%
4th
2010
0.56 °C
0.1%
-0.1%
5th
2005
0.55 °C
0.0%
+0.0%
6th
1998
0.54 °C
0.0%
+0.0%
7th
2013
0.51 °C
0.0%
+0.0%
8th
2003
0.51 °C
0.0%
+0.0%
Summary
This bar chart shows the estimated probabilities for each data set.
Conclusion
The two satellite sets are most likely to finish 4th warmest or higher,
but with a reasonable chance of finishing 5th.
For UAH 6, there's a slim chance it could finish 6th or lower.
For RSS 4 there is a reasonable chance of beating 1998 to finish 2nd.
The surface data sets are almost certain to finish at least 3rd warmest.
Hadley and NOAA are both very likely to be 3rd, but GISSTemp is most likely to finish 2nd.
The differences between the data sets has more to do with the relative position of 2015,
than with the temperature for 2017.
What all sets continue to show is that on the basis of the first 7 months, 2017 is very likely to be another very warm year.
There is no sign of a return to any sort of pause.
