HadCrut & NOAA October Temperature Anomalies: Lower than September.
Both NOAA/NCDC and Hadley have posted their surface temperature anomalies for October 2009. Both indicate temperature anomalies dropped relative to September. (Recall that GISSTemp reported 0.66C for both Sept and Oct.) The temperature anomalies reported by HadCrut and NCDC for the more recent three months are:
- HadCrut : 0.548 C, 0.459 C, 0.437 C for August, September and October respectively.
- NOAA/NCDC: 0.6043 C, 0.613C, and 0.5745 C
I’ve set all three groups anomalies to a common baseline and plotted below; October value are circled.
Figure 1: Global Surface Temperature Anomalies Since 1990
As you can see, since Jan 2001, which I use as my preferred start date for comparing models to data, the least squares trends from NOAA/NCDC and HadCrut happen to be negative. However, this is not statistically significant. However, the difference between the multi-model mean trend and the observation continues to be statistically significant at p=95% when we correct for lag-1 autocorrelation and apply a bi-variate t-test.
Currently, it’s interesting to notice that Hadley and NOAA report anomalies that are close in value when reported at a common baseline. GISS is warm relative to the other two reporting agencies. I’ve created a plot showing trends since 2000 and 2001, highlighting the current levels using arrows:
Figure 2: Trends since 2000 and 2001.
More later… I have emails to read! 
Comments Closed: If you would like them re-opened, Contact Lucia
Comments
Fred Nieuwenhuis (Comment#23795) November 19th, 2009 at 7:09 pm
Again, note that GIS, HADCRUT and NOAA do not have Canadian data. See note from NOAA:
Please Note: This preliminary assessment was made before data from Canada was made available. Therefore, some of the ranks presented in this report could change slightly when Canadian data is included in the final record. The inclusion of the missing data may cause modest changes to global-scale ranks. The assessment for Northern Hemisphere land temperatures could change by several ranks.
tetris (Comment#23831) November 19th, 2009 at 10:20 pm
Interesting as we are experiencing an El Nino.
DG (Comment#23838) November 19th, 2009 at 10:27 pm
Previous HadCRUT temps keep getting adjusting upward month after month.
lucia (Comment#23839) November 19th, 2009 at 10:28 pm
tetris–
Yes. That said, temperature could still go up. It’s been a warm November in Illinois.
vg (Comment#23853) November 20th, 2009 at 12:41 am
refer to previous post to assess credibility of ANYTHING from British met Office
DG (Comment#23900) November 20th, 2009 at 9:09 am
Files for HadCRUT data are dated thru 3-9-2009.
http://www.cru.uea.ac.uk/cru/data/temperature/
crosspatch (Comment#23924) November 20th, 2009 at 12:16 pm
I will agree that the first half of the month has seemed warm in the continental US but I am seeing indications that the second half of the month is going to be chilly. Though the chill might start a bit too late to influence the monthly mean very much.
tetris (Comment#23987) November 20th, 2009 at 5:40 pm
Lucia,
I know weather is only weather, but West coast temps here in BC have been well below the norm with the exception of late September-early October, and there are new snow records being established both in terms of quantity and early in the season.
Gaz (Comment#25781) December 2nd, 2009 at 9:35 pm
“..Jan 2001, which I use as my preferred start date…”
Ha ha bloody ha.
Why not use last Wednesday?
lucia (Comment#25812) December 3rd, 2009 at 8:06 am
Gaz–
The AR4 projections are based on a set of Scenarios that were published in late 2000 (roughly November, if memory serves me right.) January 2001 is the first month of the year after those SRES were defined. I prefer to test projections or predictions made before the events could be observed. So, January 2001 is the first date I consider reasonable. Earlier dates permit comparison to hindcasts because both the temperatures and the forcings had already occurred.
If you read my blog you will see that I regularly post comparisons beginning in many earlier years. For example, in this post, you can see I show a graph beginning in 1990 as well as 2001.