主播大秀

It will be fascinating to see if global temperatures continue to rise from the recent La Nina minimum. After all the Met Office said a year or so ago that half the years between 2010 and 2015 would be hotter than the hottest year 1998, based on their forecasts. That would mean a sharp, sustained rise from here.

Yes, odd that UAH chose to change its anomaly base period midway through December 2010 - during the last month of the joint warmest J-D on its record. This had the effect of raising the 'baseline', thus making it appear as though the recent warming wasn't all that significant after all.

If you keep 'raising the bar' like this in any trend then you tend to 'dampen' the apparent rate of change.

What changing the base line can't do though is alter the linear temperature trend, which as Paul points out, now sits at +0.14 C per decade since 1979, up from +0.13 C. This fact is overlooked in Dr Spencer's blog post: . This is also odd, because he made a specific point of mentioning this when the trend fell from +0.14 to +0.13 about 18 months ago.

Also, you will never see a linear trend line added to the temperature anomaly graph in Dr Spencer's blog; though he is very keen to add these to any temperature series that appears to show a decline.

So - a rise of 0.14C per decade, 1.4C in a 100 years - big deal! I don't suppose it will stop the hysteria and demands for money, money, money for useless Canute type projects!

newdwr54
regarding your little dig at UAH changing its baseline, on the previous thread you stated @15
"While there are fluctuations in every temperature set, 30 data points is usually considered to be sufficient to identify an overall trend in a series and determine its statistical significance over time."

It was my understanding that this was explicitly why the UAH baseline changed. That other agencies prefer to stick with their existing baselines is up to them.

I still find it "odd" that 0.3mm a year has been retrospectively added to all sea level data points. The fact that 6 months ago, sea level rise had slowed from around 3.1mm to around 2.7mm was presumably a coincidence. Ooh look now - its back to just over 3mm a year. And the justification for this? Something to do with sinking ocean floors and increasing ocean basin. Fantastic, so sea level measurement now has a volumetric constant added. We're now measuring sea level from the bottom up!

When the UAH baseline was changed, it was stated that it was to bring it in line with other series which used a 30 year period, such as NOAA:
"we have just switched from a 20 year base period (1979 鈥� 1998) to a more traditional 30 year base period (1981-2010) like that NOAA uses for climate 鈥渘ormals鈥�."
Unfortunately they seem to have overlooked the fact that NCDC/NOAA actually uses the 100 year base period, 1901-2000. Personally I think that a standard base period should be introduced for all of the series, and the 20th century would probably be a good standard. The plethora of different base periods only causes confusion (I sometimes think that is the intention), and even the IPCC seems to use different base periods in it's reports. I suspect that UAH changed to the 30 year base period because they actually had the amount of data to support it.

newdwr54, I have discussed before (not sure if it was with you), that a single month's anomaly is never going to make a significant difference to the linear trend over the entire UAH series, at least to 2 decimal places. In order to see changes in the trend, it is necessary to look at more decimals or calculate it over shorter periods, such as the rolling 10 year period I use. In the case of UAH, this month's figure has reduced the 10 year linear trend from +0.0286c/ decade to +0.0269c/decade. Along with other series, the UAH 10 year linear trend has been declining since reaching a peak of about 0.43c/decade around June 2002 and at the present rate will be negative by early next year, although in that respect it is behind the other series.

Personally, I don't think the last two month's UAH figures necessarily represent a return to an upward trend. In the previous post El Nino phase, the decline wasn't constant and there were several ups and downs before much lower figures were reached. I also don't think this month's figure necessarily implies an increase in this month's HadCRUT3 anomaly.

Is anyone else having problems with posts?
Mine seem to get stuck and I have to refresh the screen to see if it has been posted. Maybe my computer needs a re-boot.

yes QV - I had the same issue

I am not sure if this post will all go through at once, so I may have to finish it in another post:

In my previous evaluation of the accuracy of the individual models in IPCC scenario A1B, using the cumulative Mean Absolute Deviations, there was an error in the spreadsheet formulas which resulted in each year's anomaly being compared with the NEXT years' multi model mean when calculating the MAD.
As a result, the individual model cumulative absolute deviations were incorrect. Although the differences in most cases were minor, they did make
a difference to the relative accuracy of the models.
Based on the correct calculation of absolute deviations, the 5 "worst" and 5 "best" models were as follows, with overall absolute deviations in brackets.

5 "WORST" MODELS:

cccma_cgcm3_1_t63 (0.296c)
cccma_cgcm3_1 (0.253c)
cnrm_cm3 (0.243c)
miroc3_2_midres (0.148c)
iap_fgoals1_0_g (0.138c)


5 "BEST" MODELS:

ncar_pcm1 (0.057c)
mri_cgcm2_3_2a (0.071c)
giss_aom (0.079c)
ukmo_hadgem1 (0.081c)
ncar_ccsm3_0 (0.086c)

In the case of the above, all of the differences from the previous figures are less than 0.01c, and there is no change in which models are the 5 "worst" and the 5 "best".

I should also point out, in case it wasn't clear in my original posts, that the MAD figures are based on the absolute deviations of individual model annual anomalies from the Multi-Model mean anomalies for that year. This is because the annual anomalies are themselves based on the predicted absolute temperature for each year, minus the mean absolute predicted temperature for 1980-99, for each model.

To put it another way, the anomalies quoted are relative to the individual model predictions for 1980-90, rather than actual measured temperatures. I personally don't find this method of calculating the anomalies entirely satisfactory, but it is apparently the method used in calculating the anomalies shown in IPCC graphs, such as fig. 10.4 in AR4, chapter 10. I have actually had this method confirmed by Gerald A. Meehl, one of the coordinating lead authors for AR4 chapter 10, (after a series of confusing e-mails).

I can't help thinking that this may in part account for the apparent inaccuracy of the model anomalies relative to actual anomalies. Having thought about it for a while, and having produced some simulated random model outputs, is seems logical that to the extent that model predictions of absolute temperature differ from actual absolute temperatures, the model anomalies are likely to differ mo

As I suspected, here is the last part again:

I can't help thinking that this may in part account for the apparent inaccuracy of the model anomalies relative to actual anomalies. Having thought about it for a while, and having produced some simulated random model outputs, is seems logical that to the extent that model predictions of absolute temperature differ from actual absolute temperatures, the model anomalies are likely to differ more from actual HadCRUT3 anomalies, than from the MMM anomaly figures. Indeed, this does seem to be the case, which suggests that at least some of the models may be less accurate than I have stated.

I would be grateful to see any constructive comments on the above from sceptics/non-sceptics alike.

lateintheday,

Thanks for the feedback.
I didn't want to re-boot as it sometimes takes hours.

How does this correlate to CO2 levels over the same period?

11 - it correlates very poorly. It however correlates well with ENSO plus the gradual escape from the Little Ice Age. We've been warming since 1600 (something which the Mann-made up warming believers try to wipe from history to support their greentech investments).

I thought I would have a look at how the UAH linear trend, over the entire series, had changed since the end of 2010, at a higher resolution than given in the data files. The trends per decade, to the end of the dates shown were:

DEC 2010 = 0.141328
JAN 2011 = 0.140210
FEB 2011 = 0.139057
MAR 2011 = 0.137531
APR 2011 = 0.137056
MAY 2011 = 0.136645

So, when you calculate the trend to 6 decimal places, it shows that it has been falling over the last 6 months (probably longer), a fact which is not revealed in the figure to 2 decimal places, and if the current rate of decline continues, it will fall to 0.13c again by October 2011, when rounded to 2 decimal places. Obviously, this is further confirmation that temperatures are declining, at least over the short-term.
I suspect that when the trend changed from 0.14c to 0.13c, it was, in reality a much smaller change, at the point where the figure was rounded down.
As the series grows in length, it will become increasingly difficult to change the trend over the entire series. All we are likely to see are changes from 0.14c to 0.13c or vice versa, due to rounding, which tells us very little about how the temperature trend is really changing. A more precise calculation or a trend over a shorter period, would be preferable to detect any changes in the trend, either positive or negative.

4. lateintheday wrote:

"Ooh look now - its back to just over 3mm a year."

As far as I know current sea level rise is always given with a fairly wide margin error due to inherent uncertainties in measurement. Last figure I saw from NOAA was 2.9 卤 0.4 mm/yr, which is still a considerable rate of increase.

QV thanks for that explanation.
A couple of things spring to mind which you may be able to answer.
Firstly, while as you have shown above, it's important to consider the effect of third and fourth (and beyond) decimal places on the headline trend figures, what is the accepted real world accuracy of current measured temps in terms of decimal places?

Secondly, you say "As the series grows in length, it will become increasingly difficult to change the trend over the entire series". Roughly how long might it be before a significant change would show? For arguments sake alone, let's say that global cooling sets in at a similar rate to the warming trend 1980-2000, so that by 2030, we're back to 1980 temps. The period 1980 - 2030 period would look like a sine wave. In that (rather unlikely) scenario, at what point would we see a shift to negative in the 30 year moving averages trend?

5. QuaesoVeritas wrote:

"When the UAH baseline was changed, it was stated that it was to bring it in line with other series which used a 30 year period, such as NOAA"

I accept that, but NOAA's 30 year anomaly baseline period ended over 20 years ago and it refers to a set covering over 130 years; about 23% of its total data. This base line gives a much clearer indication of the change in temperatures over time, both before and after the anomaly period.

The UAH anomaly period ends less than a year ago and covers 94% of its entire data series. It is nearly impossible to glean an accurate indication of changes in temperature over time from this base line value alone. The 'bar' has effectively been raised to incorporate nearly all the temperature fluctuations within the measurement period. This defeats the purpose of having an anomaly base line at all in my view.

A trend line added to any temperature graph reveals the true trend in temperatures over time, regardless of the base line anomaly used. it is notable that Dr Spencer rarely, if ever, adds a trend line to the UAH data set, at least not on his blog page. We can only speculate as to why that might be.

I also agree with the fact that a rise of 0.01 C in a 10-year trend is insignificant; I was merely lamenting the lack of obvious reference to it in Dr Spencer's blog, given that he made particular mention of the 'fall' of 0.01 C in the same trend about 18 months ago.

#16. - newdwr54 wrote:
I accept that, but NOAA's 30 year anomaly baseline period ended over 20 years ago and it refers to a set covering over 130 years; about 23% of its total data. This base line gives a much clearer indication of the change in temperatures over time, both before and after the anomaly period."
I'm not sure why you say that the NOAA 30 year anomaly baseline ended over 20 years ago. As I pointed out, the NCDC/NOAA baseline is 100 years, 1901-2000.
Are you thinking of the NASA/GISS baseline which is is 1951-80. or the HadCRUT3, which is 1961-90?
Having said that, I agree that the new UAH baseline makes the anomalies look smaller, although I honestly don't think that is why they chose that period. In reality, once the various anomalies are adjusted to the same base period, they are all pretty much the same over the same period of time, but the different base lines cause confusion, hence my argument that a standard base line is required. Even though satellite figures obviously don't exist for 1901-2000, they could be implied from the other series anomalies.
I don't recall exactly what Dr Spencer said about the trend falling from 0.14c to 0.13c 18 months ago, but if he attached any significance to that, I think he was foolish, since he must have known that the "real" fall was much less than that.

lateintheday,
I'm afraid that you may think I know more than I do!
I am not sure what the accuracy limits of individual temperature anomalies are. Most series seem to publish figures to three decimals, but I suspect that the limits are much less than that, possibly less than 2 decimals. NCDC/NOAA actually publish them to 8 decimal places! I also don't know what the implications of that is for the accuracy of trend figures. However, I don't think that the accuracy to over 2 decimal places is spurious, since the changes in the trend figures appear to be consistent and not entirely random.
In relation to your second question, again I don't know, but I would suspect that if the decline pattern was symetrical with the rise pattern, it would take at least another 30 years before a declining trend would show. As you say, that seems unlikely and temperatures are likely to start rising again before a negative trend is seen over a 30 year period.
I should have added in my post that of course, the period over which the trend is calculated is critical. There is a trend of about 0.14c in the UAH data because it starts in 1979. If you go further back, e.g. to 1950, using HadCRUT3, the trend is only about 0.12c/decade, which perhaps isn't what you might expect. This is because anomalies were relatively flat between 1950 and 1980. In fact the longer you go back to calculate a trend in a data series, generally, the lower the trend. The overall trend in HadCRUT3 since 1850 is only about 0.042c/decade. So at the moment, it is in the interests of "warmists" to quote trends over shorter, more recent periods of time.

13. QuaesoVeritas:

With respect, by plucking six consecutive values out of a ten year trend you are cherry picking. All you have there is a six month trend within a longer underlying trend, which is of little or no significance in a longer time scale.

For instance, we can select numerous recent six month running temperature values that show far faster rising trends than that of the over-riding decadal trend. (See any six month period ending Jan, Feb or Mar 2010 for example. These all have trends well above 0.13 C, the decadal trend within which they fall.) However it is equally illegitimate to do that.

Short term variations are poor indicators for long term trends, which is why it is long term trends that are the usual rule of thumb measure in temperature series.

17. QuaesoVeritas:

Sorry, I must be confusing NOAA with HadCRUT. But the underlying principle is the same. It looks like we agree on a common baseline for all values.

Nice to agree for once!

#19. - newdwr54 wrote:
"With respect, by plucking six consecutive values out of a ten year trend you are cherry picking. All you have there is a six month trend within a longer underlying trend, which is of little or no significance in a longer time scale."
I chose the last six months (not just any six months), to show how the trend has changed in detail over the latest figures. It wasn't my intention to prove anything about what was happening over a longer time scale.

"Short term variations are poor indicators for long term trends, which is why it is long term trends that are the usual rule of thumb measure in temperature series."
I agree, but equally, long-term trends tell us nothing about how things are changing on a shorter timescale, e.g. over 10 years or allow us to identify possible causes or even make predictions about the future.
That is why I calculate the the rolling trends over periods of 10, 20 and 50 years, which do show the trend is not a simple continuous one, but a complex cyclical one, which may repeat over time.
I believe that calculting such trends tells us more about what is happening than a simple single trend over a long period.

#20. - newdwr54 wrote:
"Sorry, I must be confusing NOAA with HadCRUT. But the underlying principle is the same. It looks like we agree on a common baseline for all values."
No need to apologise, it seems that Dr. Spencer did the same thing!

It is interesting that according to the NCDC/NOAA "State of the Climate" National Overview, the overall temperature for the U.S. in May was "slightly cooler" than the 20th century average.

Looking it in more detail, however it turns out that "slightly cooler" was 0.6c below normal, which I wouldn't call "slightly lower".
Note that in April, the "highlights" mentioned that it was the seventh warmest April on record and 0.59c above the average.
No mention of May's ranking in the highlights, but further inspection reveals that it was actually the 25th coldest, or 92nd warmest on record.
I don't mention this because it has much relevance in terms of global temperatures, but merely to demonstrate the presentational bias that is evident in these reports.

"Note that in April, the "highlights" mentioned that it was the seventh warmest April on record and 0.59c above the average."

that's from the global overview.

I see the AMSU temperature Ch05 is rocketting up
We are almost at 2010 temperatures for June

Once again to ensure that our forcasts remain on record I post our predictions made for 2011 at the start of the year


Met Office +0.44
SmokingDeepThroat +0.39
quake +0.36
ukpahonta +0.35
Gadgetfriend +0.30
NeilHamp +0.27
QuaesoVeritas +0.25 (revised)
millinia +0.24
Joe Bastardi +0.2
Ken Sharples +0.18
LabMunkey +0.18
nibor25 +0.15

As mentioned in my previous post AMSU temperatures are shooting up for June
The La Nina is showing some signs of returning to neutral conditions
Perhaps the year end is heading for a warm surprise
The next few months will be critical
Comment number

#24. - quake wrote:
"that's from the global overview."
Oops!
Thanks for pointing that out.

#25. - NeilHamp wrote:
"I see the AMSU temperature Ch05 is rocketting up
We are almost at 2010 temperatures for June"
Yes a big jump on the 8th, although last year the anomaly was fairly flat at this time. There were similar jumps in April but they were not sustained.
Actually the graph looks a bit odd. The scale at the bottom doesn't look as if it is aligned properly with the data. I can't quite work out why, but the scale gets progressively out of line during the course of the year so by the end there is a month missing. They seem to be messing about with the site, so maybe it has been corrupted.
The data file also seems to have changed as it doesn't seem to have the average figure in it any more and all of the other files seem to have changed to AQUA.

NeilHamp
I seem to remember that you have in the past, been good enough to provide a further set of figures along the lines of . . temperatures for the remaining 7 months must average ??? in order to meet ??? predictions.
Any chance of this again?

#26. - NeilHamp wrote:
"Once again to ensure that our forcasts remain on record I post our predictions made for 2011 at the start of the year"
"As mentioned in my previous post AMSU temperatures are shooting up for June
The La Nina is showing some signs of returning to neutral conditions
Perhaps the year end is heading for a warm surprise
The next few months will be critical"

Yes, I am beginning to regret having changed my estimate now.
I think I may have been misled by the unusually low figures at the time, which caused me to think that the cycle was progressing faster than it was.
However, I still think we are in for a downturn, it's just a question of the precise timing.

#29. - lateintheday wrote:
"NeilHamp
I seem to remember that you have in the past, been good enough to provide a further set of figures along the lines of . . temperatures for the remaining 7 months must average ??? in order to meet ??? predictions.
Any chance of this again?"
In my case, the average has to be about 0.225c for the remainder of the year.
I must admit, it seems unlikely!

NeilHamp,
I can work the others out quite quickly if you want.
If anyone will believe my figures that is!

QV -I'll keep my fingers crossed for your estimate. ENSO forecasts seem to have stabilised over the last few months to non event status. Before that they were all over the place, jumping from double dip one month to el nino the next.
Last I heard, JB was still ruling out el nino for this year and still minded towards a possible return to la nina territory later this year. Tallbloke seems to think SSTs will remain low which should in turn, keep temps down.
You may feel like your sitting on a pair of eights, but it could still be the best hand.

Regarding the AMSU graphs on the amsutemps website, I have looked at this in more detail and it appears that although the scale goes to end December, the 2010 data, or any other year, only go to about the end of November.
However, while the figures are not in the correct place relative to the scale, they do seem to be in relation to other years.

Lateintheday,
Thanks, at least I have the consolation of knowing that my estimate will be more accurate than that of J.B., unless temperatures really fall!
And he has more than me riding on it.

The Met.Office have forcast +0.44 for 2011
So far this year HadCRUT3 temperatures have been
Jan.0.21
Feb.0.26
Mar.0.32
Apr.0.40
May.(0.45) estimated from UAH May figure

For Met.Office forcast to be achieved this will require the last seven months of 2011 to average +0.52.
The last 7 months of 2010 ("the 3rd hottest year") averaged +0.44
If this were repeated in 2011 the final year end average would be +0.39
This would make Smokingdeepthroat our competition winner

Gavin Schmidt at RealClimate forecast in Jan:

"I predict that 2011 will not be quite as warm as 2010, but it will still rank easily amongst the top ten warmest years of the historical record." :

Is this looking like a bold statement?

newdwr54 writes:

"Is this looking like a bold statement?"

I thought it was (the "easily" part), and mentioned this in the comments.



The SOI was the largest positive value in a 60-year period (indicating a very strong la Nina), and I thought anomalies might rival the 2007-2008 peak, but that hasn't happened. One can see this in the UAH data, and also in the surface record. Temperatures are unusually warm for a la Nina of this magnitude, but then again, there's that additional CO2 thing.



The GISS annual value will have to be 0.48 by the end of the year to crack the top 10, and it's already tracking that, despite some strong la Nina influence. A weakened la Nina should be enough to keep it there or higher.

One thing that Paul's last paragraph seems to imply is that the satellite and surface records are directly comparable with a simple baseline adjustment - that they should track each other closely on a monthly basis. They don't. Temperatures near the surface are less sensitive to large ENSO fluctuations than the entirety of the lower troposphere, as measured by satellites. UAH surpassed GISS for most of 2010. Now GISS is running ahead of UAH. So when someone is making a prediction, the distinction between surface and TLT records is important. I'd be more inclined to bet 2011 would be a top 10 year in the surface record during a strong la Nina than the TLT satellite record.

#37. - newdwr54 wrote:
"Gavin Schmidt at RealClimate forecast in Jan:

"I predict that 2011 will not be quite as warm as 2010, but it will still rank easily amongst the top ten warmest years of the historical record."

Is this looking like a bold statement? "

It probably didn't seem like a bold statement in January, but it looked very unlikely
after the first three months.
If the May HadCRUT3 increases at the same rate as UAH and RSS, it would be about 0.42c (this is a bit lower than Neil H's estimate) and if it stayed that way for the rest of the year, the annual figure would be 0.38c, which would make it the 11th warmest, It would need to be 0.4c or over to get into the top 10, based on UKMO annual figures.
As Neil says, to achieve that, it would need to be warmer than the end of 2010, but remember, temperatures had already started to cool by December.

I have just noticed the word "easily" in Gavin Schmidt's prediction.
Does that not imply top 5?
If so, it would have to be over 0.456c
Certainly better than 8th, which would require over 0.427c

#38 - MarkB2020 wrote:
"One thing that Paul's last paragraph seems to imply is that the satellite and surface records are directly comparable with a simple baseline adjustment - that they should track each other closely on a monthly basis. They don't. Temperatures near the surface are less sensitive to large ENSO fluctuations than the entirety of the lower troposphere, as measured by satellites. UAH surpassed GISS for most of 2010. Now GISS is running ahead of UAH. So when someone is making a prediction, the distinction between surface and TLT records is important. I'd be more inclined to bet 2011 would be a top 10 year in the surface record during a strong la Nina than the TLT satellite record."
I don't think that P.H.'s last para necessarily implies that satellite and surface anomalies should track each other on a monthly basis. However, they do tend to track each other over longer periods. They may get out of step, but usually catch up with one another. As you say yourself, UAH was above GISS last year, but this year it's the other way around. The point is, all of the anomalies are relative to different periods and you HAVE to do the adjustment in order to really know which IS the highest. Without the adjustment, UAH will always look much lower than GISS or HadCRUT3.

Just out of interest, I thought I would work out what HadCRUT3 temperatures have to be for the remainder of the year, in order for individual predictions to be correct. These figures are from the end of April, so include no estimate for May.
Met Office 0.51c
SmokingDeepThroat 0.435c
quake 0.39c
ukpahonta 0.375c
Gadgetfriend 0.30c
NeilHamp 0.255c
QuaesoVeritas 0.225c
millinia 0.21c
Joe Bastardi 0.15c
Ken Sharples 0.12c
LabMunkey 0.12c
nibor25 0.075c

I would say that as things stand, ukpahonta, quake and SmokingDeepThroat stand the best chance of being correct.

thanks for that- is interesting.

Thanks for that QV and NeilHamp
Looked the latest ensemble forecasts for ENSO this morning and it offers some hope for those with the lower predictions.



"Is this looking like a bold statement"

I wouldn't say it was particularly bold, though I'm pleased he felt prepared to put his name to an actual verifiable (in our lifetime) prediction. He knows that while year to year fluctuations can be significant, the chances of the temps dropping out of the longer term range are quite small. The trend has been more or less flat for around 10 - 12 years, with ENSO events not being sustained or powerful enough to make a difference. Also, he would've expected that the la nina would fade (which it has) and perhaps be offset to some extent by a rise in solar activity towards its predicted max around 2013. While this rise has been slower and weaker than he may have expected so far, it will eventually show up and add to the temps.

Both skeptics and AGW proponents will expect a much 'bolder' statement as soon as the next el nino appears on the horizon, since it will more than likely occur at or around solar max. Although, from recent discussions on this board, its possible that the highest temp may occur if an el nino comes a year or two after the max.

Shock...I see 主播大秀 and Phil Jones are on the case again.

/news/science-environment-13719510

QV,

ukpahonta 0.375c should be 0.35C

Wouldn't want to generate any concerns about positive bias in the blog records now would we, hehe!

nibor 25
thanks for the link . . . it's very typical I agree.

And to sum up, the article reads:

鈥淐ritics alleged the emails showed CRU scientists and others attempting to subvert the usual processes of science, and of manipulating data in order to paint an unfounded picture of globally rising temperatures.
Subsequent enquiries found the scientists and their institutions did fall short of best practice in areas such as routine use of professional statisticians and response to Freedom of Information requests, but found no case to answer on the charges of manipulation.
Since then, nothing has emerged through mainstream science to challenge the IPCC鈥檚 basic picture of a world warming through greenhouse gas emissions.鈥�

Now, compare and contrast . . . .

From: Phil Jones
To: 鈥淢ichael E. Mann鈥�
Subject: HIGHLY CONFIDENTIAL
Date: Thu Jul 8 16:30:16 2004
Mike,
鈥︹赌︹赌︹赌︹赌︹赌︹赌︹赌︹赌�..
鈥淚 can鈥檛 see either of these papers being in the next IPCC report. Kevin and I will keep them out somehow 鈥� even if we have to redefine what the peer-review literature is !
Cheers
笔丑颈濒鈥�


It鈥檚 difficult to see how anything could emerge through 鈥榤ainstream鈥� science with an attitude like that.

#46. - ukpahonta wrote:
"ukpahonta 0.375c should be 0.35C "

I think that was your original prediction.
The figure of 0.375c is the average anomaly required from now on to hit your prediction.
A figure of 0.35c from now on would result in a figure of 0.33c by the end of the year.
Sorry if I didn't make that clear in my original post.

#45. nibor25 wrote:

"Shock...I see 主播大秀 and Phil Jones are on the case again."

So if I understand this correctly, when the figure was over 15 years it wasn't statistically significant, but over 16 years it is? But all it means is that there is only a 5% probability that it occurred by chance. This is just statistical sleight of hand, designed to impress the uninitiated. I suppose that by the end of next year, it will be 100% certain it didn't happen by chance?
I wonder how the figure of a 0.19c rise over the period 1995 - 2010 is arrived at?
The linear trend over the 16 years = 0.012225c /year x 16 = 0.1955, so if that is how they worked it out, I am surprised they didn't call it a rise of 0.2c.
Can anyone remind me why the period of 1995-2009 was originally chosen in the first place?

QV,
Appologies I should take more care in reading posts.

Wasn't there a quote associated to Phil Jones in 2009 along the lines of 'no significant warming in the last 15 years' that the blogosphere picked upon and gave him grief over?
This is probably his substantial reply to that.

"Can anyone remind me why the period of 1995-2009 was originally chosen in the first place?"

Maybe because they didn't dare start in 1998!
(actually, I've no idea)

#50. - ukpahonta wrote:
"Appologies I should take more care in reading posts."
No problem. Not reading things properly is a fault of my own!

"Wasn't there a quote associated to Phil Jones in 2009 along the lines of 'no significant warming in the last 15 years' that the blogosphere picked upon and gave him grief over?
This is probably his substantial reply to that."
The problem is, the original is STILL true!
The significant warming is over 16 years.
In reality, nothing seems to have changed except the period over which the warming is being measured.
Can someone with a better knowledge of statistics than I, please clarify
if it is ever possible to have statistically significant warming (or cooling), over a 15 year period?

Looks like the La Nina may be coming to an end
If so my estimate of +0.25 by year end looks a trifle cool
But previous La Ninas have occasionally had a sting in their tale
NOAA's site showing the Multivariant ENSO Index (MEI) is worth checking out



Scroll down to the third graph to see how the 2010/11 La Nina compares to similar historic events. For some reason they are late with the June update, but we can expect a rise in the Index similar to 73-75. The year end temperatures will depend on what happens next.

Will the La Nina re-establish itself like 73-75 or will it dissappear like 88-90?
If we see a pattern like 73-75 then temperatures will start to cool later this year. We can only wait and see.

The 主播大秀 headline obfuscates "significant" with "statistically significant", as do some comments here. A large positive value with a 2 SD range exceeding that value could be considered the former but not the latter. Parts of the denialosphere of course dumbed it down further to "no warming".

With 2010 in the books, the period now passes 95% confidence. This is in the HadCrut data, which has an anomalously smaller trend compared with GISS and NCDC, and does not include the rapid warming trend in the Arctic. All 3 datasets have shown a signficant trend since 1995. All 3 also show a statistically significant trend since 1995, although 2010 data was required for HadCrut to pass the 95% confidence level.



Lastly, note that there is a pending upward correction to SST that will primarily affect the last decade.

#54. - MarkB2020 wrote:
"With 2010 in the books, the period now passes 95% confidence."
By "the period", I presume you mean 1995-2010, so I am sure that you will agree that it is still accurate to say the there was no statistically significant warming over the years 1995-2009. The fact that "the period" has been extended to 16 years, therefore makes no difference to the previous situation.
Would the period 1996-2010 show any statistically significant warming using HadCRUT3?

"Lastly, note that there is a pending upward correction to SST that will primarily affect the last decade."

Oh for goodness sake!

Here is an interesting analysis of Jones' claim of statistical significance:



Perhaps the professor could show his method?

#57. - RobWansbeck wrote:
"Here is an interesting analysis of Jones' claim of statistical significance:

Perhaps the professor could show his method?"
Thanks for posting that link.
While I don't pretend to have the statistical knowledge to understand much of what is being said, I somehow doubt if Richard Black does either, although I may be doing him a disservice. The point is that in the absence of such knowledge, the general public has to take what was said in the article at face value, but when someone who does have the statistical knowledge looks at it, it turns out not to be as clear cut as was implied in the article. For one reason or another, R.B. didn't apparently question Jones's methodology, and just accepted it as gospel.
It is interesting that the analysis was done on the two CRU datasets, i.e. HadCRUT3 and HadCRUT3v, and that using the latter, there apparently was statistcal significance. Someone in one of the responses said that HadCRUT3v was the "trend which most HadCRUT people use". I am not sure about that. I have never been able to find out how the variance adjusted data is arrived at, but the monthly values in the file are NOT the same as those used by the UKMO, while those in the HadCRUT3 file are. Moreover, in his "Information Sheet no. 1", updated January 2011, which concluded that 2010 was the equal third warmest year on record, Jones appears to use HadCRUT3, not HadCRUT3v. So to use the HadCRUT3v series to "prove" statistical significance would surely be an example of "cherry picking".
The information sheet can found on the link below and links in the sheet point to HadCRUT3, not HadCRUT3v.

The MEI has just been updated

Scroll down to the third graph on the site for a historical comparison of recent La Nina's

There has been a very rapid rise in the MEI confirming that the La Nina is coming to an end.