## Thursday, December 17, 2015

### Warmest November Yet, 2nd-Warmest Month Ever

Both NASA GISS and NOAA have posted their November temperature anomalies in the last few days, and both are searing.

GISS found November to be +1.05°C above the 1951-1980 baseline period, and NOAA found it to be +0.97°C above their 1981-2010 baseline.

I should bring those to a common baseline. Maybe later.

Both are guaranteed to have 2015 as the warmest year in their records, unless a huge meteroid slams into Earth, like, tomorrow. (I  hope it's not before my flight home.) The GISS anomaly for December needs only to be greater than -0.38°C for 2015 to be a record, and NOAA needs it greater than -0.76°C. As my niece says, easy peasy.

JoeT said...

As I wrote below, last year NASA gave the probability that 2014 was the warmest year at 38%, while NOAA said 48%.

Assuming that December has the same anomaly as November, I estimate that NASA will put the probability of 2015 being the warmest year at 94%, NOAA at 97%.

David Appell said...

I tried to do that calculation last January, but got different numbers:

http://davidappell.blogspot.com/2015/01/probability-2014-is-warmest-year-66.html

I haven't seen GISS's actual calculation anywhere, just the result.

David Appell said...

PS: Joe, how did you do your estimates?

JoeT said...

If you're looking at 2 normal distributions with say T1 > T2 both with a standard deviation a, the probability of T1 being higher than T2 is given by

0.5*(1 + erf((T1 - T2)/sqrt(4*a*a)))

So for example using the NASA data, and assuming the December anomaly is the same as November then T1 = 0.86C for 2015 and T2 = 0.74 for 2014. I'm almost certain that 1 sigma for NASA is 0.05 C. That alone says that for NASA, the probability that 2015 is greater than 2014 is .5(1 + erf(0.12/0.1)) = 95.5%.

For multiple temperatures, I just brute force it setting up 1000 trials of random guassian distributions. For NASA, I just used the top 6 warmest years but even that was overkill. It lowered the probability from 95.5% to 94%.

I'm not as certain as to the 1 sigma uncertainty for NOAA, but just assumed it was the same as for NASA.