HadCrut posted their December temperature NH&SH anomaly: 0.251C, down from Novembers 0.455C. I’ll be posting a year end summary a bit later, but for now I’d like to show the 10 year trend based on monthly data:
As some know, I have been, and plan to continue to watch trends since January 2001, because that’s that corresponds the beginning of a year after the SRES used to create projections in the AR4 were frozen. It will, of course, make some people angry for me to notice that using Hadley, the trend since 2001 closes the year in negative territory, just as it did in 2009, 2008 and 2007. That said, the trend computed since 2000 is slightly positive: 0.0004 C/year, ad was positive for the past three years. For whatever reason, HadCrut is effectively trendless during the beginning of the year. ( Other observing groups show some warming.)
No Record in the Annual Average
Many of us have been watching for records in the annual average. In the final months of the year, few anticipated that HadCRUT would show any record; it didn’t. The 12 month lagging average for temperatures from assorted metrics all rebaselined to the 1980-2000 average are shown below:
Temperatures themselves fall below the multi-model mean, but are within ±1SD of the multi-model mean from models driven using A1B of the SRES. When interpreting this, bear in mind, rebaselining forces averages to match between 1980-2000.
What will I be monitoring for during early 2011?
I always watch to see if observed temperatures will manage to pierce the multi-model mean. If the multi-model mean is unbiased, the temperature should do so. I’ll continue to do so.
Looking at the graph, and knowing La Nina persists, I’ll be watching to see if the annual average temperature falls below the 1SD range of the models as it did in 2007 before temperature rise when ENSO re-establishes itself (which it eventually must.)
Lucia,
“Statistically insignificant” global warming? CO2 driven and man-made we are told.
As per your calculations above, we are now talking about a 2000-2010 “slightly positive” trend of 0.0004C/year…
The French have a rather poetic take on things like this: enculer les mouches. Relative to that, we are at this point dealing with humping bacteria and viruses…..
I respect your like for mathematical and physical accuracy, but this type of attention to minute detail is tantamount to falling off the “pico“ cliff. 🙂
tetris–
That happens to be the trend for this period with HadCrut. The trend since 2001 based on NOAA is 0.0013C/year since 2001 and 0.008 c/year since 2000, both positive.
I pointed out the one negative trend because, as a matter of formalism, that trend remains negative. I anticipate warning to return, and I anticipate that we’ll be seeing a positive trends since 2001 by 2020. But currently, trend for HadCrut since 2001 remains negative and if we report them all… well.. that one is negative.
If the anomaly is 0.251C, why is the data point on your graph under 0.1? Do you move the baseline?
Chris, She really get’s down on statistics which means she’s a complete loon or totally crazy.
Chris–
Yes. I always show data and model results using the 1980-2000 baseline. I note the baseline on the graph.
Thanks for all your great work Lucia.
I think now is the right time for a general post explaining (in common language) why AR4 projections begin in 2001; the Hadcrut reference frame in AR4; the relevance of 0.2C projections; and how it’s statistically falsified.
Thanks
Also, can you please explain the difference between OLS and MEI corrected trend? I assume OLS is the linear trend with which to compare to projections.
It could be interesting to get a look at all non-overlapping decadal trends from the various data sets (RSS, UAH, GISS, HADCRUT) and look at the distribution. E.g. according to woodfortrees, UAH and RSS both had negative trends for 1980-1990, though the longer trend is clearly up.
1) I continue to think it would be better and statistically more honest to calculate a trend from peak to peak, that is from 1998 to 2010
2) In 1980 the different curves are within 0.4°C of each other; in 2010 they are within 0.8°. But we know that nobody cheats, do we not?
3) Yopu migzh want to corretc the title
P. Gosselin says the trend this century is negative for Hadcrut.
http://notrickszone.com/wp-content/uploads/2011/01/Hadcrut.gif
DonB– HadCrut posts several series! Using NH&SH, I also get a negative trend since 2001. Traditionally, “1” is the first year of the century (well… unless it’s bce. Then the century runs from -100 to -1. There is no year zero. You go from -1 “before the christian era” and +1 after, skipping past zero).
So, yes, the trend for this century is negative.
Alexej
I am aware you prefer the method the statistically unconventional method of calculating “peak to peak”. That method requires you to select which peak (or peaks) you prefer, and then compute “peak to peak”.It seems you have settled on starting with the peak in 1998. Why pick that peak? Why not pick the previous peak around 1995? Or 1990? Do you have some standard method for picking the 1998 peak?
I noticed the model results have become flat since ~2008 instead of following the oft quoted 0.2 C/decade linear trend. What mechanism in the model is predicting the “flat-line” trend over the last few years?
Crashex– The 0.2 C/decade is the average for the model trend over the first few decades. The model mean is averaged over a number of models, each with between 1-5 runs. That mean retains noise because the number of models is not infinite and individual runs contain noise. It happens that, computed over very short periods, the multi-model mean falls below 0.2C/decade and for very short periods, it rises above 0.2 C/decade. There is no physical significance to this. It’s the noise that imperfectly averaged as a result of the finite number of models.
If I plotted the annual avearged mm-mean over a longer period you’d see the positive trend resume. If I plotted monthly values, your eye would better see the noise.
There are benefits and short comings to plotting 12 year annual averaged data on monthly intervals. One of the shortcomings it that smoothing tricks the eye into thinking there is less noise than exists, and so over interpreting the “flattening” in that curve. Once over interpreted, people are included to want to explain it– but it’s residual noise.
Lucia, rather than base your estimates on dubious calculations, why not use the historical record Don Easterbrook has constructed for the last 10,500 years? I think you could make the case much easier by pointing out that it has been warmer over the last 10,500 years 86.6% of the time.
Of course, I am predicting sustained reduction in GAT for the next 20-30yrs, based on falling oceanic temperatures. I also contend that abnormally low co2 levels cannot lead to much warming and believe that co2 forcing is over estimated. I also think radiation is escaping into space much more rapidly than the models predict, as Richard Lindzen has done.
An additional question I have for Lucia is what if instead of warming, we had cooling at the same level, the total inverse. Also, what would be worse, if GAT dropped 3 F or if it rose 3 F?
“There is no physical significance to this.”
That is a good summary of what can be expected of Climate models.
Thanks
Hi Lucia,
It just occurred to me that you might be interested in a thread
http://climateaudit.org/2011/01/04/the-c-e-t/#comment-251929 which I am currently involved in. It’s on testing predictions of HadCRUT3 increases, and it should be right up your street – please feel free to comment. I probably won’t have time to comment there again until tomorrow.
Cheers,
Rich.