GISS Temp is out for December! Back in September, I decided to start comparing the cached version to the fresh version each month. Here’s a comparison of the cached version and the fresh version.
As I sip my morning coffee it appears to me that:
- December’s temperature anomaly was 0.14 C lower than November’s anomaly as reported in the fresh update.
- November’s temperature anomaly rose 0.01C during December.
- The 2006 Jan-Dec annual average GMST anomaly got warmer; 2005 and 2001 Dec-Nov annual average also got warmer. (I made a mistake and circled the 2007 as well. Because didn’t save the file where I circle, I’m leaving that. 🙁 )
- Other months may also have changed.
Readers may recall that I posted a variety of GISSTemp version in November. As I’d posted them, I wanted to see how the Jan 14 cached version had changed compared to the Nov. 12 version. Here are screen shots circa Nov 2008:
Out of curiosity, I circled temperatures in the Jan 14 cache that differ from those in the Nov. 12 version. You can scroll up to see.
What I can observe is quite a few monthly avearage surface temperature differ in the Nov. 12 records and the Jan. 12 record. I got tired of scanning and circling. I may have made mistakes.
For those wondering why I am a bit curious about all this: Well, there are the ordinary reasons. We all sort of wonder vaguely how long it takes GISS Temp anomalies to stabilize. More specifically, I have in the past mentioned an important caveat to any and all analyses: The results depend on the data being accurate. When the data change, this can affect our current conclusions. For this reason, I like to have a feeling how much the already reported GISSTemp anomalies tend to change from month to month; this lets me judge whether the updates in past data are likely to affect the result of any hypothesis test.
Currently, based on changes we are seeing: The updates in past data can change the result of a month’s test slightly. Every now and then, if we repeated last months test updating that data with the corrections (but not including the actual additional month), the slight difference in trend and residuals to a fit will result in a different outcome to a hypothesis test at a chosen confidence level. However, the differences due to updates are not large. (In contrast, future data could affect the trend and the level of confidence in any hypothesis test. )
Anyway feel free to wax philosophical about this data. When you see HadCrut update, let me know!
UpdateLink to data here.
“The results depend on the data being accurate” That’s the crux of it. One cannot count on GHCN data on being up to date, the degree of adjustment, homogenization (sp?), extrapolation, interpolation that happens between the “raw” and GISTEMP GMST value.
Fred–
Also, the fact that some of this data isn’t stable. Similar to the other agencies, GISS doesn’t keep formal archives of every update (or if they do, those aren’t available on the web). This means that any formal write up needs to state day the data were downloaded and the author needs to personally archive that specific version. Trends, estimated confidence intervals etc. will always depend on the data set used.
Hansen has done a write-up on the temps of 2008.
It almost reads as though Hansen is turning a corner although I’m probably reading too much into it.
Near the end, he almost says the GHG warming trend is now reduced to 0.15C per decade.
“4) Greenhouse gases: annual growth rate of climate forcing by long-lived greenhouse gases (GHGs) slowed from a peak close to 0.05 W/m2 per year around 1980-85 to about 0.035 W/m2 in recent years due to slowdown of CH4 and CFC growth rates [Reference 6]. Resumed methane growth, if it continued in 2008 as in 2007, adds about 0.005 W/m2. From climate models and empirical analyses, this GHG forcing trend translates into a mean warming rate of ~0.15°C per decade.”
http://www.columbia.edu/~jeh1/mailings/2009/20090113_Temperature.pdf
Well… there ya go! When I have looked at it. +0.15 C/per decade is not outside the range of what is consistent with data.
Lucia:
I think Bill Illis’ quote has “~0.15” – that is, about +0.15
Perhaps you have taken his number as a negative (reading the tilde as a minus sign).
Something does not compute.
CO2 forcings have been increasing quite rapidly over the last decade. If these rapid rates are slower than the IPCC models projected then that suggests the IPCC BAU GHG forcings are nonsense.
It would not surprise me since the forcings in Hansen’s original Scenario A had no connection with reality but he presented them as plausible in order to make things look scary.
Brent– I read Bill’s comment as a positive. I meant to type two dashes –0.15C/decade meaning +0.15C/decade.
Internal variability (weather) can result in excursions from the “underlying trend”. If the underlying trend is 0.15 C/decade, getting some negative 8 year trends in not necessarily unexpected. If the underlying trend is 0.2 C/decade, it seems unexpected.
Of course, I’ll now check 0.15 C/decade. But you now have a climatologist giving a projection below that published by the IPCC. He states his reason, which for all I know, may be correct. We’ll see what others say. That said, this new estimate of 0.15C/century, appeared now after cold weather set in. So, it’s great evidence of predictive ability of large teams of climate scientists acting collectively to provide their best estimate.
Data integrity says it all.
.
I’ve been looking at the uber-large positive anomalies in Siberia and in Antarctica on the GISS maps recently. In both of those areas, there’s a large gradient in the anomalies.
.
UAH doesn’t show that gradient. If the gradient is real, UAH should show it as well. The magnitudes may not match, but the gradient should be present in both GISS and UAH. It is not.
.
So I started downloading the daily files from GHCN for Siberian stations in the middle of the “hot” spot (uh-oh . . . is it okay to use that term? haha 🙂 ) and also the Antarctica ones.
.
On first glance, there’s a problem with the data: Several of the Siberian stations have more than 200 data points missing each year since about the year 2000. Looking at Olenek specifically, nearly twice as many MINs are missing as the MAXs. I’m not quite sure how GHCN interpolates when data is missing yet . . . so working on that . . . but the raw mean is not right. Nice thing about the Siberian ones is that the GISS homogenization doesn’t seem to change them at all (data set 0 and 2 are identical for Olenek). The stations are too sparse.
.
In Antarctica, the problem is a bit different. There’s very little data from the stations I looked at so far to set a baseline. So I think the GISS zero point for Antarctica may be incorrect.
.
Anyway, I’m going to finish going through all the stuff and hopefully in a couple of weeks I’ll have some pretty graphs to show.
.
More fun with GISS temps, I suppose. Haha.
Brent–
To futher clarify. If I use “The Method Of Santer 17” using data from Jan 200-2008, the GISS 2-sigma uncertainty for GISS temp says the data could be consistent with trends bewteen -0.194 C/decade to +0.174 C/decade. 2 sigma is nominally the ±95% confidence range (though it depends on how many degrees of freedom one has.) So, Hansen’s new estimate of +0.15 C/decade would fall in that range.
You can hunt around for different start years for the analysis, but I pretty much stick with Jan 2001.
For the same period, using the same method, HadCrut is consistent with 0.0304 C/decade to -0.264 C/decade. So, Hansen’s 0.15 C/decades would not be consistent with HadCrut for that period.
Caveat:
This numbers change with different choice of start year.
The will change when I add December.
Hansen’s new prediction did not “forecast” or “project” what happened during that period. However, it might be read to suggest that someone who believes models do work when driven by correct forcings, is thinking that maybe the “about 2C/century” does not apply right now.
Can someone post the GISSTEMP URL, for the tabular data? I seem to have lost the address.
thanks,
Hi Les:
Giss link to data here.
Hadcrut: http://hadobs.metoffice.com/hadcrut3/diagnostics/global/nh+sh/monthly
The ~0.15C per decade could always be a typo or it could be just a specific subset of GHG forcing that he is talking about – he is not always clear.
The statement also includes the phrase “From climate models and empirical analysis” which could mean he is just using an approximate number somewhere in between the high numbers of the models and the lower numbers of actual observations to date.
Lucia; You are a mind reader. I lost HadCRUT too.
I still have UAH and RSS, so 2 out 4 ain’t bad….
Lucia,
This graph from an early draft version of the AR4 (didn’t make it into the final version for one reason or another) might help clarify Hansen’s point:
As you can see, CO2-eq emissions have been roughly constant from the late 1980s to present due to the dramatic reduction in ODP gas emissions (due to the Montreal Protocol). That said, now that virtually all the CFCs and HFCs are gone, emissions are still increasing rapidly.
Lucia (re: #8692)
Thanks for clarifying the glyphs!
Speaking of GISS, Lubos has some interesting statistical evidence that GISSTemp data may be manufactured…..
http://motls.blogspot.com/
Pity I am traveling. I can see Tamino, Schmidt and Romm all blowing a gasket on this….
Zeke–
I’m not sure the graph clarifies Hansen’s point. To what extent was the slowdown in CH4 and CFC emissions since 1980 already known in 2001 when the published the SRES?
If this slow down due to the Montreal protocol or any other treaty (or any orther reason) was correctly captured in the SRES, then the slow down in emission cannot be the cause of the projections overshooting the future observations.
If, on the other hand, the slow down in these emissions was unexpected and not incorporated into the SRES, then that could result in model projections being ending up too high. This would result from incorrect SRES.
I can’t tell how reality compares to the SRES based on Hansen’s letter, or from your figure.
Zeke,
Your graph is emissions. I think this graph shows things more clearly.
http://www.esrl.noaa.gov/gmd/aggi/aggi_2008.fig4.png
This shows GHG forcings from actual atmospheric concentrations. Although Hansen is right in that the rate of growth seems to have peaked in the 80s, the actual rate of increase for total GHG forcings is still pretty darn constant since 1990 or so, and it doesn’t look like CFCs (the main components of “ODPs” in your graph) can account for any significant surprising decrease in the rate of forcings increase since 1990.
Thanks John, thats exactly what I was looking for. A few quick calculations, and we get:
Average annual percent increase in radiative forcing by year range:
1980-1984: 2.272
1985-1989: 2.328
1990-1994: 1.308
1995-1999: 1.51
2000-2004: 1.202
2005-2007: 1.15
So yep, a big decrease in the rate of forcing growth between 1988 and 1992 (largely due to the CFC phaseout) and relatively constant rate of growth since. Interesting.
(note the y axis of the graph should be percent increase, not absolute w/m^2… oops!)
Zeke,
Nice graph! One can indeed argue a step-change in rate since ~1990.
However, I’ve gone back and tried to figure out Hansen’s intent with the quote Bill gave above. At first, I thought he was speaking in the context of his “Scenarios” which were made in 1988. But when I actually opened the link and tried to discern the context, it is clear that he’s using the forcings in a longer discussion to explain why he expects record temperatures in the next 1-2 years.
I really don’t get what his item “(4)” has to do with his 1-2 year prediction/projection, other than he seems to be implying that he expects the recent “spike” in CH4 concentrations to continue.
Odd how the “spike” in methane is treated, considering that the “dip” in temperature over the last year or so is considered insignificant.
http://www.esrl.noaa.gov/gmd/aggi/aggi_2008.fig2.png
http://rankexploits.com/musings/wp-content/uploads/2009/01/rss-dec.jpg
Okay, there is no typo and no particular subset being pulled out here by Hansen.
GISS repeated Hansen’s conclusions on its GISSTemp page today.
“Greenhouse gases: Annual growth rate of climate forcing by long-lived greenhouse gases (GHGs) slowed from a peak close to 0.05 W/m2 per year around 1980-85 to about 0.035 W/m2 in recent years due to slowdown of CH4 and CFC growth rates [ref. 6]. Resumed methane growth, if it continued in 2008 as in 2007, adds about 0.005 W/m2. From climate models and empirical analyses, this GHG forcing trend translates into a mean warming rate of ~0.15°C per decade. ”
There is definitely a change in the wind.
http://data.giss.nasa.gov/gistemp/2008/
Bill Illis (Comment#8746)
Interesting, Lucia’s graph shows a slope of .162C/decade in her GISS Temp Jan 2001(0)-Dec 2008 post for 1980 to present.
Hi Lucia,
I have an excel sheet that would solve the problem of circling differences, as it does it automatically, by shading differences in one color or another depending on the direction of the change. You only need to copy and paste the raw text data of the two GISS datasets that you want to compare, as they were presented in the link you provide. It is very quick and easy, as the excel sheet was created precisely for this purpose.
If you are interested, tell me and I will send it to you. However, if you are analizing how long it takes GISS temperatures to stabilize, I can already answer that question: NEVER. With the new data release in January including the December anomaly, other temperatures as far as November 1886 (that’s the 19th century) have experienced changes. In total, 35 monthly averages have changed, most of them before 1954! The longest period without any changes is 1954-1976, right in the middle of the 1950-1980 period whose average GISS uses for a zero anomaly. 4 out of 5 of the monthly data changes before 1900 warm the temperature; all of the 17 monthly data changes between 1900 and 1954 cool the temperatures; and 9 out of 13 of the monthly data changes after 1976 warm the temperature.
The excel sheet I created can be now downloaded:
http://www.elsideron.com/GISS_Trends.xls