The Blackboard

With the recent cold in the Chicago area (14 deg below “normal” at night) I harken for warming.

Jack– That Roy! Try as I might, I can’t manage to scoop him on UAH! Darn him to heck!

Hmm. For a while there I was beginning to think the May anomaly for both UAH and RSS would be ridiculously high!

Guess I focus too much on the 3,300 ft channel.

It is interesting that the decrease since 2001 is so steep. But it won’t mean a whole lot unless it continues for another 21 years.

KW–

But it won’t mean a whole lot unless it continues for another 21 years.

Well, that depends whot you mean by “a whole lot”. The short term trend is sufficiently far from 2 C/century to indicate that the current short term weather would not be consistent with “weather noise” super-imposed on trend of “about 2 C/century”. “about 2 C/century is what is projected for the surface during the early part of the 21st century, and generally, one expects the warming in the troposphere to be a bit faster. (But, the IPCC didn’t specifically state any projections for the troposphere.)

So we can say the data appear inconsistent with warming at that rate. Would this meet the definition of saying “a whole lot”? Or is it saying only a little?

What we can’t say is that according to tests using 2001 as a start date, the trend is inconsistent with noise superimposed on a warming trend– but that warming trend must be lower than “about 2 C/century”. Of course, you could argue about fitting trends starting in 2001. But, that year give intermediate results relative to starting in 2000 or 2002, and it happens to correspond to publication of the SRES used to drive the projections.

Am I the only one who is concerned that temperatures are not declining more (or the rate of warming declining more dramatically) during a period of solar inactivity? Am I missing something basic here?

re: #14154

I’ve been wondering the same thing

Well, it’s been cold enough here so that I haven’t had to turn my AC on yet. I’m not sure I have been able to hold out this far in the year before. My goal was to leave it off the entire month of May, and it happened. No we’ll see how far into June I we get.

And for those who are worried “the temperatures” arent falling fast enough, I say… huh? You guys take these numbers seriously enough that you can tell that something is wrong?

Andrew

Tom and Eric: This is a controversial subject. Lief over at WUWT, who is a solar expert, has stated that there does not appear to be a strong correlation between TSI (total solar irradiance) and global temperature. The TSI in recent times has varied only +- 0.1% and such a small variance would need to be magnified hugely by an as yet unknown mechanism to explain temperatures. There are of course well known plots of sunspots versus temperature which are comepelling at first glance. I believe they do not hold up well to analysis, taking onto account time lags, etc. Then there is Svensmark’s work about CGR’s and clouds and solar activity which does posit a strong influence. Anyway, it seems that the jury is still out on this topic. I personally believe in a meaningful solar influence but this is just based on intuition and not on published research. And finally, you must take into account time lags. If the sun did dim we might have to wait a while for cooling, as the oceans hold a lot of heat. Just my 2 cents

Tom, Eric, to important points to remember about this are that the response of the climate system to high frequency oscillations like the solar cycle are highly damped, and delayed on top of that. So it would take a sustained period of unchanging solar activity for the full effect of the change up until that point to manifest-it’s the same with response to just about anything else. This is a well recognized climate property. The second point is that many factors are at play at any given time determining the anomalies. A coming El Nino will warm us up whether cycle 24 flounders or not.

Anyway, just what expectation are you comparing the offsetting of warming to? Way out there predictions of the next ice age? That’s unfair. Mainstream climate models? Can’t be, they didn’t predict the amount of offsetting that occurred (not surprising given their failure to duplicate the IPCC acknowledged solar cycle amplitude of .1 C). You must be careful. Me thinks you are comparing it to your own prejudices, no offense intended good sirs.

Tom Fuller/ Erik–
Some blog visitors have suggested that the sun is the cause of the lack of warming. The IPCC AR4 literature review sections does discuss the expected variation of temperatures from peak to trough of the solar cycle and one paper suggests it’s about 0.1C (I think.) But others dispute this possibility (and even scoff that the effect could be that large.)

Unfortunately, the model projections going forward after 2001 took out the variable solar cycle. But I don’t know if the modelers set the sun stuck near “high” (as it was near the beginning of the century) or immediately dropped it to “average”. So, that makes it difficult to interpret the mis-match between models and projections.

But let’s assume hypothetically the modelers set the sun to “high” marching forward. In that case, some of the current mis-match between models and observations may be due to going from the top to the bottom of the solar cycle. However, if we estimate for the magnitude of the effect, assume the 0.1C is true and just add it to the observations, the observed trends still fall the model projections. (That said, last time I did that, the models were no longer “outside ±95%” for tests starting in 2001 after the speculative correction was applied. I could re-run it. I haven’t checked for tests since 1979, but for that case, the sun issue is, oddly, not particularly relevant because the historic variations add to the estimate of the ‘weather noise’ and so the uncertainty intervals already account for that.)

So, that was a long answer. But, with regard to saying models appear to be over predicting warming, I’m not too worried about the sun. (Though, I admit I could ultimately be shown to have been wrong about that issue.) But if someone is decreeing cooling, then, definitely, we need to be aware that the sun is going to pick up soon, and part of the negative observed trend may be due to the sun.

I downloaded the solar radiation at the top of the atmosphere data (variable: rsdt). I processed some of it a while back to get an idea of what it looks like. It has an annual cycle due to the Earth’s elliptical orbit. I looked at the data for 20c3m and sresa1b and I did not see anything that resembled the actually observational data. I just gave it a quick look over. I don’t want to recount too much just from memory of what I found. But I can take a more thorough look later in the week. Right now I’m processing the variable ta which has atmospheric temperatures at 17 layers so I can construct synthetic MSU temps.

Chad– I vote for constructing synthetic MSU temps. Everyone wants those. On the solar issue, it might be useful for someone who is curious about the influence of the sun on short term trends to ask one of the public-face of climatology, aka blogging climate modelers (e.g. Gavin) their opinion about the role of the sun in this. My impression is “they” don’t think that explains away any current discrepancy (but then, that’s because I think “they” think the current discrepancy doesn’t need to be explained away. I think “they” are wrong about that. )

I’d ask, but a modeler, but it’s not really my question.

Still, I can find this by googling:

This is not to say that there is no solar influence on climate change, only that establishing such a link is more difficult then many assume. What is generally required is a consistent signal over a number of cycles (either the 11 year sunspot cycle or more long term variations), similar effects if the timeseries are split, and sufficient true degrees of freedom that the connection is significant and that it explains a non-negligible fraction of the variance. These are actually quite stiff hurdles and so the number of links that survive this filter are quite small. In some rough order of certainty we can consider that the 11 year solar cycle impacts on the following are well accepted: stratospheric ozone, cosmogenic isotope production, upper atmospheric geopotential heights, stratospheric temperatures and (slightly less certain and with small magnitudes ~0.1 deg C) tropospheric and ocean temperatures. More marginal are impacts on wintertime tropospheric circulation (like the NAO). It is also clear that if there really was a big signal in the data, it would have been found by now. The very fact that we are still arguing about statisitical significance implies that whatever signal there is, is small.

(Italics mine.)

Overall, I read that article to suggest that Gavin is inclined to think variability of the earth’s surface temperature over the solar cycle is rather small and that possibly even the estimate of 0.1 deg C may be too high. Of course, I may be misinterpreting that, but over all, the tenor of that 2005 blog post conveys the impression that he thinks whatever the magnitude of the variations due to the solar cycle, it is sufficiently small so as to be buried in the noise.

KW (Comment#14152)

Guess I focus too much on the 3,300 ft channel.

I got just as mislead this month by Channel 05 (14400 ft). It’s my excuse for getting trounced in my prediction at Niche Modeling this month. Of all the recent months, I thought this month would be up for sure.

David Stockwell has mercifully removed the guesses.

http://landshape.org/enm/guess.....-may-2009/

There are two unavoidable truths about the Earth’s Climate: 1) The main heat transport mechanism is the oceans and their characteristic oscillation modes are defined by the boundaries of land and the gravity field (hence why GOCE is flying in space to map this).
2) The atmosphere can best be modelled as a coupled radiative-convective ‘blanket’ not a greenhouse. CO2 and H20 will radiate at all heights throughout the atmosphere where they are found but the heat transport and lapse rate can only be adequately accounted for if convection is introduced.
With these two things in mind, there will be coupling effects due to inherent frequencies and excited frequencies from external variations, like the Sun. The Earth’s Climate appears to exhibit the same interesting characteristics as seen in things like plasmas or even electronic noise; there is a degree of universality about the behaviour. The key is that to understand the effect of external variations you need to at least know something about the underlying frequency dynamics and boundaries of them (either caused by land interfaces or tropopause and sea surface in teh atmosphere) before you introduce solar effects or cosmic rays effectively. For example, If someone can produce a viable mechanism for the onset of El Nino i.e. why the strengthening of the trade winds beyond a certain threshold, then a better understanding of climate will follow.

it appears the 1988 climate model predictions made by Jim Hanson are falling apart and can no longer be acheivable. It’s been 20+ years since Jim published the 3 global temperature scenarios and all 3 predictions are higher than current global temperatures. Why do we still rely on human models? It’s becoming a joke. Does anyone really think we will be able to predict the global temperatures in the next 100 years?

[URL=http://img13.imageshack.us/my.php?image=scenariochart.jpg][IMG]http://img13.imageshack.us/img13/863/scenariochart.jpg[/IMG][/URL]

I think it’s hard to follow UAH monthly given the known annual cycle in their anomaly. We’d better wait for a fix.

Ahab–
What known annual cycle? How large is it?

Anyway, even if it exists, why is this necessarily important? The real underlying data has an annual cycle. That’s taken care of by the anomaly method. Even if a small amount of the honest to goodness physically based, annual cycle remains in the anomalies, this would appear as noise. Noise of any sort (whether “weather” or “measurement”) widens the computed uncertainty intervals. So, if there is a small one, but we already consider the uncertainty intervals when interpreting the meaning of the data, why should we ignore data for that reason?

So… I’d be interested in reading anything you have showing that the world at large knows there is an annual cycle in UAH, but I still use it. If, on the other hand, you are aware of a slow drift, that would bother me; if established as absolutely known, I wouldn’t use the temperature series.

@Lucia
Deep Climate just posted on this issue and he’s not the only one.
Also, it’s a pretty regular pattern (on the anomaly, not absolute temperature itself) of almost 0.3 °C peak to peak on average from about year 2000; then it’s bias, not noise. There’s no reason to belive it cancels out on average.
I’m not saying we should reject UAH, we all keep using it at least because there are not that many satellite data set. But we should keep this issue in mind.

Ahab–
Atmoz posted on this a long time ago. I think I commented on it then. I’m not convinced from anything I read the magnitude of the differences found is anything more than what we would expect to arise as a result of the uncertainty in determining the annual cycle by computing it during a finite period “A” and then freezing those errors into all future anomalies. This ‘error’ would easily become statistically significant in all periods outside the one where the anomaly was defined if you forget that there was uncertainty in computing the annual cycle in the first place. I commented on this when Atmoz first noted the issue.

Also, Tamino posted on this and nothing about what he wrote changed my mind because he didn’t happen to consider the freezing in of the uncertainty associated with the initial determiniation of the annual cycle in both RSS and UAH. I guess Deep Climate has now.

BTW: If Deep climate found a peak to peak difference on the average difference between RSS and UAH has a peak to peak difference of 0.3C, he made a mistake.

Everyone makes mistakes sometimes, but it’s not usual for him to make one of that particular nature. Do you have a link to what he found and said that caused you to think he found a “it’s a pretty regular pattern (on the anomaly, not absolute temperature itself) of almost 0.3 °C peak to peak on average from about year 2000″

I’m pretty sure that the “annual cycle” is in the differences between RSS and UAH, related to their differing diurnal corrections. In this regard, given the use of AQUA by UAH, I’d bet my hat that the problem is with RSS and NOAA 15.

There are also papers in the literature which suggest that UAH is superior in that regard, for instance:
http://www.agu.org/pubs/crossr.....8864.shtml

The upward trend since 1979 is caused mainly by the 10 warm months in 1998. Those points are in the right half of the diagram. And they will stay in the right half for quite some years. We won’t see the upward trend dissappear until 2019 when we have more points after the 1998-peak than from the 1979 – 1998 period.
With the peak in the left hand of the diagram, it will cause a downward trend … I guess.

It’s not often I have to oppose a skeptical notion, but in point of fact, when you account for the influence of El Nino and even volcanoes, there is still a tropospheric temperature trend. Sorry.

If it makes you feel better, it is still a fairly modest trend.

Andrew_FL–
Yes. No matter how much filtering, correcting for el nino etc. one does, a positive trend remains over the lifetime of the measurements.

I avoid correcting for El Nino (including 1998) because I’m not confident it can be done properly. If one is going to do it, they should also include an estimate of the uncertainty associated with the correction for El Nino. That’s quite difficult to estimate, particularly since you need to first decide the lag associated with the El Nino index that best explains the later variations in the earth’s surface temperature.

I’d rather just retain ENSO type noise as unexplained and permit that to be reflected in the uncertainty intervals. Others can do differently to learn something that interests them, and what they are trying to infer from the data.

I’m currently working on such an analysis, and that is one of the issues that has been bothering me. Both the temperatures and the “explanitory” factors are uncertain (they have errors) and I am uncertain how to deal with that. Do I combine them? Then my results are so uncertain as to be worthless. Do I ignore them? Then I perilously risk getting criticized for overstating the certainty. Do I reduce them? How?

Up till now, I’ve just said “well, you probably aren’t going to publish or anything, so just do what you want”-which seems to be the standard rule when “accounting” for such effects in the literature!

There is an annual cycle in the Earth’s average temperature. The Earth’s orbit is not a perfect circle. It moves farthest away in July and is closest in January. Total Solar Irradiance received at Earth distance changes considerably between these two dates +/- 10 watts/metre^2.

In terms of the solar cycle, the troposphere should be affected a little more than the surface. The solar cycle impact is estimated to be:

+/- 0.05C at the surface;
+/- 0.1C in the troposhere;
+/-0.2C in the stratosphere.

Bill Illis-On what basis do you halve the surface effect relative to the troposphere? I would think that, since (correct me if I’m wrong) tropospheric effects are supposed to be 1.2Xsurface effects, it would be more like

.1 Troposphere

.083 Surface

Andrew_FL:

Both the temperatures and the “explanitory” factors are uncertain (they have errors) and I am uncertain how to deal with that. Do I combine them? Then my results are so uncertain as to be worthless. Do I ignore them? Then I perilously risk getting criticized for overstating the certainty. Do I reduce them? How?

Sometimes there is just not enough data to reduce the uncertainty as low as you’d like it to be.

If you tried to determine the average height of American men by picking 10 at random and measuring them, you will get an estimate of the uncertainty in your determination of the average height. No amount of correcting for geographical location, income, educational level, immigration status or ethnic group can drive that uncertainty to zero. Those factors may matter, and you may be able to prove they affect height given a large enough sample. But you can’t get perfect estimates of their effect. So, when you correct, you would need to account for the uncertainty of your estimate of their effect on your final result.

In undergraduate engineering labs, we would 1) estimate the standard error in parameters associated with each effect, 2) linearize the effect of the “correction” based on the magnitude of the parameters us and then use the chain rule to find the estimate of the standard error in the ‘correction’ and 3) using sum of the squares of errors, add this to the uncertainty in the average height we after “correcting” the average height for any effect.

This method gives an estimate of the extra error associated with the uncertainty in the magnitude of the parameter used to “correct” for an effect. It’s imperfect and even then, it only works if the things “corrected” for weren’t discovered by data mining until you found statistically significant correlations between anything. (And obviously, it’s never worth correcting for effects that whose statistical significance you can’t demonstrate somehow.)

Also, when doing this, you have to consider the uncertainty in every choice. So, for example, if you are going to correct for ENSO, you probably picked some number of months lag between ENSO and the best fit to surface temperature. Let’s say you got 3 months. Then you got a parameter for the correction and you can estimate the uncertainty in that parameter based on your fit to ENSO You picked that based on the fit between surface temperature and ENSO at that lag.

But… what if the “prefect” fit isn’t 3 months, but 4? How would that affect your estimate for the trend? Well… you should, add that uncertainty into your estimate!

In any case, sometimes, if you follow the formalism, you conclude that your estimate is more uncertain after correcting. If you don’t follow the formalism, you will often trick yourself thinking you can correct until nearly all uncertainty disappears. You can’t. All you’ve done is over fit.

(Real statisticians have other ways to test for over fitting using specific statistical tests. But the method described above is useful for keeping you out of trouble quite often. Plus, it has advantages that you can use it before you collect data using simple estimates of likely standard errors based on stated precision of instruments in sales catalogs. It really helps the experimentalist figure out whether the are better off improving the precision of say, a temperature measurements as opposed to pressure measurements, estimating how much data to take, how to space data points etc. It’s a good discipline afterwards, but it’s a very important thing to do before hand even if you aren’t going to “correct” your data afterwards.)

Hmm. For all of us who like to look at every anomaly released each month, it would be cool if someone created a rating system (e.g. 1-10 or A B C D E F) for them all. I wonder what people, modelers, and scientists beleive the best all around measures are for a global anomaly.

My take is that satellites would perform better in regards to coverage. But problems with drift, etc. would make me question if they are indeed most precise. Land measurements with or without urban adjustments just don’t have the coverage…i.e. the oceans and the poles.

KW–
Everyone disagrees which is best. Also, there are people who are sniffing around for any and every possible problem with either UAH (because it shows the smallest long term trends) and NOAA (because it shows the largest.) Either will scream bloody murder at over the flimsiest possible evidence.

Then, there are people who are more balanced and are aware of problems. My take is that it’s best to look at all of them and bear in mind that there are some structural uncertainties. But it’s worth nothing that qualitatively, all are in agreement on a few specific things. There is a long term warming trend. Since 2001, best fit trends are statistically indistinguisahble from zero. Under the assumption of AR1 noise, all are statistically distinguishable from “about 2 C/century”.

So, the squabbles about the surface stations or possible problems with the sattelite don’t affect these more modest claims.

Andrew_FL

I think we have been misstating the solar cycle influence for the surface temps. The temp changes by 0.1C through a solar cycle so it should properly be stated as +/- 0.05C (sometimes people including myself write it out as +/-0.1C when it is only half that).

In terms of the different temp effects on different levels of the atmosphere, these estimates come from Judith Lean, who is one of the foremost solar experts. Here is nice short presentation recorded on youtube (it covers a lot of climate science in a fairly short time).

http://www.youtube.com/watch?v=OOMsQcEN1Bg

Bill–
The citation in the IPCC is not Judith Lean. But her values correspond to the IPCC reference. Not everyone agrees with Judith. As far as I can determine, that estimate is in the upper range of what various experts consider plausible.

Bill- I always thought the estimate of solar cycle effect came from Douglass and Calder:

http://arxiv.org/ftp/physics/p.....411002.pdf

Which, unless I misread, finds +/-.05 in the troposphere as opposed to the surface.

I am somewhat suprised by how little influence solar rad has on temperatures.

Before any human influence, how on earth did temperatures rise and fall so wildly? Was it not somehow related to the sun? Or does the heating and its lag somehow magically get affected mainly by the oceans, atmospheric gases, and cloud interactions without the significant need for the suns warming at all? I mean…where the heck else would it come from? Geothermic heat from the earth?!

Lucia: I wonder what your take is on this:
http://arctic-roos.org/observa.....-in-arctic
note the 2 lower graphs. This has been up for 3 days since F13 and f17 channels were “corrected”

KW-the effect of the 11 year solar cycle is .1-the effects of longer cycles may be larger. If those cycles exist. And right now, we don’t know if they do. Some think so, but others don’t. Which is why I’ve taken an interest in Roy Spencer’s work on climate’s intrinsic variability.

One of my favourite things to do is to determine the area under the curve for TSI or SSN charts(always using Leif’s data!!) to determine the “activity” of the sun. If you do this you will find that the 19th century has ~10% more “activity” than the 18th; the 20th ~17% more than the 19th. Also the two halfs of the 20th century are almost equal (the later being slightly more!!)

If you do the same for Volcano DVI (using Mann’s weighted DVI), the 19th century is, I don’t have the numbers in front of me, 60% (at least maybe more) greater than the 18th and the 20th even less than the 18th (~30%)

Surely this would have some effect on the earth’s average global temperature with the 20th century likely to be the warmest.

Also, it has been pointed out by Bob Carter, Bob Tisdale and others that the El Nino of 1998 was and anomaly who’s effects has not been dissapated. If you take the UAH trend from Dec 1978 to June 1997 the straight line trend is 0.036/Decade. If you take the trend from 1989 to Nov 2008 the trend is 0.132 /Decade. I’m just waiting patiently to see what the temperatures do in the next few years. Maybe they will go down, revert to the very slight upward trend of before the El Nino or go through the roof!!

VG–
I don’t have any particular take on those figures about the arctic ice. It looks like they say the ice is near average. It looks like they are having some instrumentation difficulties.

Is there something I’m missing?

Steve Hempell

Do you have any data on the sum/accumulation of solar activity that you could show us.

I agree that increasing/decreasing solar activity has to have an accumulating impact (although the length of time for that accumulation may in fact be quite short – perhaps just days or maybe as much as 35 days – unless the oceans enter the picture and then we could be talking about years again.)

One would need to match up the accumulation period with a logical/physical reservoir that would be capable of holding, in essence, photons of EM radiation from the Sun for that specific period. Technically, the average period of time that a photon (more accurately the energy represented by a photon) from the Sun stays in the Earth system is only 18 hours.

Yes but remember, the median, or in this case the mean, isn’t the message.

VG (Comment#14185)

It’s getting a little hinky out there as far as sea ice goes

nsidc

shows a major drop down to almost 2007 levels but ROOS is showing close to average. NSIDC is supposed to have had sat problems. They’re is an algorithm change (at least at JAXA which the one I’ve been following) as of the start of Jun. I don’t know what the status is on the other sites. Not sure who to believe.

I venture to suggest that perhaps the NSIDC satellite (as well as others who use same) is tending to fail and this shows up as an exaggerated ice loss every time. Maybe for the first time, NORSEX has realized this and adjusted all ice data (even 2008 when there was a massive down correction done in November 08 me thinks, which now seems to have been re-corrected to its original tract), to correct level. Just from eyeing the actual ice, this seems a more likely scenario. I put to you, that the current NSDC ice is also showing a failing satellite signal when compared to what you can actually see (the ice) for example at CT. Complete speculation on my part BTW! Re Lucia’s comment re if NORSEX and other adjustments are now correct, nearly all ice bets last year would have been off/wrong?.

Bill Illis:

I’m not sure what you are asking for. I do have 2 Excel worksheets containing some charts which I used to get my figures. They are horribly messy however, but I could attempt to clean them up for you and make them understandable. Can I post them here somehow? Might take me a few days to do.

The DVI figures were somewhat off. I used both Mann and Lamb/Mitchell to determine the DVI figures. Bob Tisdale objected to my using Mann.(posted this on WUWT too;RSS Chart)

The figures are
19th > 18th by 62% (Mann) 82% (Lamb/Mitchell)
20th< 18th by 6% (Mann) 33% (Lamb/Mitchell)

Steve: You wrote, “Bob Tisdale objected to my using Mann.(posted this on WUWT too;RSS Chart)”

Actually I cautioned you about using the Mann DVI data since MBH exaggerated the recent decades (as usual), as discussed in my post here:
http://bobtisdale.blogspot.com.....index.html

Regards

Steve: Note that MBH use the term WEIGHTED in the title of their DVI data.

Are you having a summer ice extent minima competition this year Lucia?

Regards

Andy

AndyW–
I can’t until I remedy my deficiency with regards to mailing brownies! So… I better do that. (Blush!)

This is a fair site and i hope note has been taken of my previous post re arctic ice. Once again
http://arctic-roos.org/observa.....-in-arctic
was re-adjusted (from channel 13 to 17 )from LAST YEAR (when a massive “down adjustment” was made) It is now showing that a considerable part of 2008 was NORMAL (within 1SD) and is now showing that 2009 is complete NORMAL. Excuse the CAPS, Sorry… I do feel a bit over the top on this one.

VG
I think this is a very interesting development. The Norwegian data uses a 1979-2007 [not 1979-2000] average which makes the information all the more relevant.

I have argued before that looking back in a few years it will be clear that 2007 was the outlier in the stats. NASA and others have shown that the 2007 “unprecedented” low ice levels were caused predominantly by wind patterns which pushed the ice out to warmer waters.

Meanwhile, in the face of this and other relevant data, we still have the alarmist propagandists [Joe Romm comes to mind] telling the world that the Arctic is melting at an accelerating rate and polar bears are on the verge of extinction..

And yes, maybe Lucia should consider running another ice extent minima contest.

“Update: Roy Spencer leaked his April UAH temperature anomaly; it’s 0.04C, down from 0.09C.”

Lucia, you meant of course “his May UAH”

Tetris: Re Romm etc.. One thing I noticed more recently was a divergence in the UNISYS V NOAA SST graphs. I’ve been harping on and on about the differences between UNISYS and NOAAA SST. Finally someone has noticed. We now know why.
from icecap:
“Based on the coming El Nino he (ROMM) hints at upcoming disappointment for climate realists with respect to arctic ice and warmer global temperatures for 2009 and for the decade. Of course he used the bogus NOAA temperatures which have taken the lead in being the most contaminated and exaggerated through station dropout globally, no adjustment for urbanization, a purposeful adjustment up of sea surface temperature warming (compare UNISYS with NOAA satellite), and bad siting (Anthony Watts has identified only 10% of the 948 United States stations meet government’s own standards for siting”.from time to time”

Thank god for those honest Scandinavians!

Tetris, Lucia: NANSEN now has confirmed that NH ice was and is now near normal
http://wattsupwiththat.com/200.....#more-8273
thanks to WUWT for following this up

VG–
Wow! That’s cool. I think I’m going to have to remedy my deficiency so I can run a contest again. What with the instrument issues, predicting the reported outcome at the end of the season could be a real cr*p shoot! That actually makes it more fun. Sort of like predicting coin tosses!

Can someone cross-check my spreadsheet skills? Here are the slopes I get for each of the following 4 global temperature datasets…

Hadley – March 1997 to April 2009 = -0.0049 C / year
GISS – December 2000 to April 2009 = -0.0097 C / year
UAH – May 1997 to May 2009 = -0.0102 C / year
RSS – March 1997 to April 2009 = -0.0042 C / year

As per usual, GISS seems to be the outlier. I actually calculated monthly slope from the monthly series, and multiplied by 12.

Walter-
It’s unnecessary to multiply by 12. Here’s what I got,

Hadley: -4.0849e-004 ± 0.0055 °C/year
GISS: -8.1072e-004 ± 0.0099 °C/year
UAH: -8.3675e-004 ± 0.0086 °C/year
RSS: -3.5320e-004 ± 0.0079 °C/year

All the figures you report are right, but 12 times too large. The UAH figure is a bit too high. For the UAH data that my script downloads, there is no data for May 2009.

Thanks. I was focussing on the fact that I used monthly data. I forgot that my X axis was already converted to years and fractions, so I was already dealing with years. Good thing I asked for a 2nd opinion before making a fool of myself.

Roy Spencer releases his data a bit early at http://www.drroyspencer.com/category/blogarticle/ as a blog article every month. See the June 4th blog entry for the first 5 months of 2009.

Has anyone noticed that the May 2009 anomaly is almost identical to the first May in the series. Clearly that is cherry picking but can we calculate the chances of that happening with an upward trend of 0.2ºC/decade and the observed “weather” noise. I don´t play cards – or horses – so I am not very good at working out the odds.

Jorge– Doing no math at all, we can easily say the probability is less than 1/2!

Lucia –

You are playing with me.

It is strange how these coincidences happen though. There was the Hansen testimony case where the temps 20 years on were actually lower.

I am not sure whether it is the colour changes or the straight lines confusing me but the data since 2000-01 seem to look different to the earlier data. There seems to be less scatter and less cycling. I wonder if the climate has moved towards a new attractor.

The original idea was that the warming trend would emerge unmistakably from the weather noise but the noise seems to be of larger amplitude and lower frequency than expected.

It looks like we will have to wait a lot longer before we find out if climate models have captured enough real world physics to be able to make a useful prediction.

Jorge-
To create both trends in EXCEL, the post 2001 data are superimposed on the 1980-now data. I chose similar colors and outlined. So yes, the later data look just a little different.

Yes. I’m toying with your question. To compute a probability that the “May 2009 anomaly is almost identical to the first May in the series” we would have to define the numerical value that defines “almost identical”. The computation could be done…. but I’m doing other things.

The fact is, all specific outcomes can be defined into being rare if we impose enough conditions. If we like games we can adopt baseball player announcer statistics. (Excample: Did you know this batter has the highest batting average for a left-hander batting in the third inning this May.)

I think we actually do sometimes do that in climate-blog-wars. We read May is the “Nth warmest” ever, the “nth coolest” this decade, the current RSS trend is the lowest 101 month trend since the trend beginning in the first two months of operation. etc. (The current RSS trend really is the lowest 101 month since the 101 trend beginning in Feb 1979. I was curious because someone was asking about historical trends on a thread elswhere… so I looked.)

Thanks Lucia –

“The fact is, all specific outcomes can be defined into being rare if we impose enough conditions.”

Yes, I understand this. It seems to be inherent in the way frequentist statistics works. You have to define outcomes and take care in defining the class to which the outcomes belong. As you increase the number of conditions attached to the class one has fewer and fewer occurrences to play with.

I am convinced that most people who have not studied statistics think you can use use the methods to say that hypothesis A is x times more likely to be true than hypothesis B. It is my understanding that frequentists refuse to make any such statements. For example, if an observed outcome would happen 90% of the time if A were true and 10% of the time if B were true, we can´t claim A is ten times more likely to be true than B.

It is a shame because I think that is exactly what most of us want to know!

Jorge–
I think people who do Bayesian analysis do sometimes answer “A is X times more likely than B”. But, of course, then need to select two hypotheses to test against each other. In the case of climate, would you want to test “The current underlying trend is 2C/century vs. The currently underlying trend is 1.5 C/century?” That is: Is the AR4 more likely to be right than the TAR?

Anyway, there are all sorts of mis-conceptions about what these tests mean “out there”.

Well… in the end, El Nino will come. Then, El Nino will end. I suspect the current result is not the 1 in 20 outlier, but who knows?

Jorge and Lucia,

Zorita has written an interesting paper (How unusual is the recent series of warm years?) that provides a framework for these types of analyses – see http://coast.gkss.de/G/Mitarbe.....pub.htm#08

Check also the comment by Burger and the response for a discussion on pitfalls.

UAH is “officially” out:
2009 1 0.30
2009 2 0.35
2009 3 0.21
2009 4 0.09
2009 5 0.05

Official UAH for May is out:
http://vortex.nsstc.uah.edu/da.....uahncdc.lt

Hi dribble –

Thanks for the links. I have had a quick read through but will need more time to take it in.

[...] note these remain consistent with “about 2C/century”. We can also note that people like ahab Ahab and, evidently, Deep Climate, have more confidence in RSS, which indicates the 2C/century [...]