Some Notes
This is an update to my ongoing look at forecasting the likelihood of 2016 setting a record for hottest calendar year in different data sets. It includes the last two surface data sets for August, and the satellite sets for September. See here for the previous update.
There are a few complications to mention first:
- HadCRUT has been updated to version 4.5.0.0. The changes are small, but the last couple of years have gained around 0.02 ° C compared to version 4.4.0.0. Details here
-
RSS 3.3 has also had some adjustments, though there is no official mention of this. Again the changes are very small, similar to those for HadCRUT. Needless to say this has caused claims of disbelief at Watt's Up With That, the site that objected to satellite-deniers.
Hot Whopper has more information about the changes.
- For some reason the data file for UAH beta 6 has not yet been updated, though 5.6 has. I'm using the values Roy Spencer reported in his blog, for now.
Although the changes to HadCRUT and RSS were very small and have little impact on any global trend, they might have an impact on the question of a record year. All my charts are based on the current data, so will show retrospective changes. There have undoubtedly been other updates to data, but so far I haven't been testing for changes. I'm only mentioning the RSS change as WUWT made such a fuss about it.
What's Happened Since the Last Post
NOAA and HadCRUT values for August have been released. Both show an increase in anomaly since July, but smaller than the increases for BEST and GISTEMP.
All the satellite data for September has been released (though UAH beta 6 only unofficially). All show an increase in anomaly from August. UAH beta 6 only increased by 0.01 °C, and this only because the August figure was reduced slightly. All others showed a more substantial increase.
Current State of Predictions
Here are my usual 2 graphs showing how the forecast for 2016 has progressed over the year.
HadCRUT and RSS 3.3 both increased their probability of beating their respective records, whilst the probability for UAH beta 6 has continued to decrease - down to 71%. All others are close to certain to beat their records.
More About UAH beta 6
Whatever the merits of the UAH beta 6 set, and noting it is still in beta and details have yet to be published, it is the set most likely to fail to set a record. My guess is that it's an even split as to whether it does set a record, but most likely it will be very close to 1998, and could well be called a statistical tie.
Here's what UAH beta 6 looks like for 2016 and 1998. The dotted line shows the needed average anomaly for the last three months of the year, for 2016 to equal 1998.
The last three months of 2016 will need to be slightly warmer than the last three months of 1998 to beat the record, but 1998 had a very cold November.
However UAH beta 6 has already beaten the record for annual temperatures - if you define annual as 12 months rather than a calendar year. The last 12 months narrowly beat the previous peak in 1998 by 0.03 °C. It seem likely this will be the peak for the current 12 month rolling average as October 2015 was quite warm.
This has also been the warmest 2 year period in UAH history. The last 24 months have been 0.09 °C warmer than any 24 consecutive moves before 2016), and may continue to increase for the next month or two.
In fact you can take any whole number of years and UAH beta 6 is currently the warmest its been over that time frame.
No comments:
Post a Comment