Officially, John Adams wins the argument: our official bird is a bald eagle. Clearly, Thomas Jefferson’s choice of a dove has never applied to the US. But, at least in November, the turkey is the national bird.

While Jim listened to the speaker, I looked for photo-ops. I found too many to show.

Duke, had taken special pains with his attire for this 4th of July and agreed to pose for this picture. Naturally, I petted him and talked to his human companions who were happy to learn I would post Duke’s image on the web, thereby ensuring his fame as a tax-protest dog. 

I strolled to the west edge of the crowd snapping examples of patriotic outfits. I asked this very friendly woman to pose. I forgot to ask her name, but I can report she had a little notebook and was taking notes. Fellow blogger maybe?

Notice the sweater?

I too was wearing a sweater. It was not warm out. Luckily, it hadn’t started to rain (again).

I milled around and checked out the literature. It began to rain, but I stuck around trying to collect as much literature as I could. Friendly people like the damp woman to the right gave me leaflets. I learned about:

• July 12: Rockin’ for the Troops featureing CSI: NY’s Gary Sinise and the Lt. Dan Band taking place at Catigny Park in Wheaton. Tickets $25. Park opens at noon, concert at 4pm.
• September 12: “Liberty Trail of 9-12″, a “…non-partisan coalition event conceived to create as much public awareness as possible of conservative ideals and values through the messages of the participating groups.” Learn more at Liberty Trail 9-12.
• I was given leaflets for Campaign for Liberty and Restoring American Liberty. ( I threw the latter away based on the stands on social issues.)

Various people spoke and sang. Given the tendency of the Illinois Republican Party to constantly bring out tired old re-treads, I was happy to see the tea party featured new face: John Arrington who has just thrown his hat into the race for US Senate. He was given about 3 minutes to speak so I obviously didn’t learn much. Around minute 2 I remembers my camera can take movies and recorded a very blurry film of his closing remarks. Here he is:

After he spoke, I identified his wife by partly on the basis of her JohnArrington for US Senate t-shirt. She was very friendly, so I asked her a few question about John before they rushed off to another tea party in Palatine. I learned that John earned a degree at Harvard, has been an alderman (for Harvey Illinois, I think….) where he battled corruption, and that that he supports more limited government. Using the information about Harvey Illinois, I googled and found this court case levied against the mayor of Harvey by several aldermen, including John Arrington. I’ll be able to learn more about the candidate when John Arrington’s web site comes online on Wednesday.

Do you want to know his position on Waxman-Markey? I was just about to ask that when some other taller, more vocal voter dashed up and began to give Arrington advice about dealing with the Illinois Republican Party. (I have no idea if the advise was good or bad.) But, I did tell John Arrington my readers would want to know his his position of Waxman-Markey. He suggested I email and he’d provide me a thoughtful answer. Meanwhile, if you want a better look at the man who may be our next US Senator, here he is:

I wonder how many people will show up? I’m going to go down and take photos of the Lisle Tea Party.

Of course, I’ll also swing by and take pictures of the the giant hot air balloons. right next to Riedy’s (on Maple avenue near the Jewel grocery store.)

Figure 1: UAH through June 2009

My quick scan indicates VG may have posted the closest specific guess in comments of +0.015C. El Niño conditions are underway way, so I bet UAH will rebound in July.

Update: No… ZerOth beat VG with 0.10C! I’ll have to read more carefully to see if anyone beat Zer0th.

Hat tip: Roy Spencer who calculated it and Andrew_FL who posted a comment while I was at the hairdresser.

The blog owner (i.e. I) can select which languages to offer her (i. e. my) blog in. Then users can click to obtain a sort of translated page. Based entirely on preconceived notions of where climate-blog addicts live, I picked a number of European languages, then threw Hindi into the mix.

If you click a language you can actually understand, the results can be sort of humorous. Below, I have posted the translation into French of the first paragraph from yesterday’s post:

J’ai été bummed que je ne peux jamais scoop Roy Spencer sur les anomalies de température UAH. J’ai secoué mon cerveau à essayer de comprendre comment je peux sortir avec une sorte de “première” lecture de la température de mois. J’ai examiné un certain nombre de stratégies:

It’s interesting to learn the idioms “être bummer” and “pouvoir scooper” have penetrated the French language. And such a novel conjugation for an ‘er’ verbs!

The various translations for the first paragraph include:

1. Spanish: “He estado bummed que nunca primicia Roy Spencer “.
2. Italian: “Sono stato bummed che non può mai scoop”.
3. German: “Ich habe bummed, dass ich nie SCOOP”.
4. Russian: “Я был bummed что я никогда не совок Рой Спенсер на грн температуры”

Those of you who want to have fun learning how far the verbs “scoop” and “bummed” have penetrated foreign languages can click the other buttons.

(If you are the first to click, this warming will appear briefly, “This page has not been translated yet. The translation process could take a while: in the meantime a semi-automatic translation will be provided in a few seconds.”)

If any native speakers are disturbed by odd translations, there is a multi-click method permitting them to suggest a better translation. Find “Back to Translate” link at the top of a translated page; click. Then enter the mistranslated text. Google will translate it again, and also provide a link to “Contribute a better translation”. Click that. Then provide the improved translation.

With some luck, Google will eventually know how to translate all slang and idiom, including words like “shank-a-potamus”. Better yet, the Google translator will help us discover the correct realclimatespeak to ‘common english venacular’ translation of “a few years” particularly when surrounded by text that specifically mentions the year 2007 twice, including an introduction that emphasizes that the new report is “the most important update of climate science since the 2007 IPCC report” and, moreover, is posted on a group blog established in 2004.

Oddly, when I first read RC’s introduction, I noticed the 2007 AR4 was specifically mentioned by the RC authors, so I assumed “Some aspects of climate change are progressing faster than was expected a few years ago” would use

1. “are” to mean “are (present tense) currently observed to be progressing” rather than “can be shown to have progressing more rapidly than the 2001 TAR model hindcasts post-dicted for historic periods predating the TAR” (but oops. Our writing now suggests we did’t bother to compare this aspect of the hind cast to the existing measurements back when we wrote the TAR and disseminated our projections. )
2. “faster than was expected a few years ago”, to mean either a) faster than was actually predicted/projected in the 2007 AR4, and/or b) faster than what was readily available data disclosed in the 2007 AR4.

However, it appears the phrase means something . . . else. If anyone can suggest the correct translation for the google “realclimatespeak” to “common English venacular” robo-translator, please do.

1. Hiring a secret agent to sneak into Roy’s office, slip Roy sleeping powder right after he computes then numbers, then email me the values before Roy can post the values,
2. Forcing Roy to go on vacation the first week of every month and
3. Developing psychic powers.

None seemed likely to succeed.

So, I decided the only thing to do was: Post a dubious value no one believes. To make it slightly less dubious, I decided I would a) compute it from the value posted for the Channel5 AMSU values posted daily at The AMSU temperature page and b) explain that this is not the real UAH temperature. The real UAH temperature anomaly uses the AQUA satellite which doesn’t drift around so much for technical reasons that involve something called “fuel”.

So, today, I will announce that the average temperature from the Channel5 AMSU was 0.023C. This is down from the May-average Channel 5 AMSU temperature of 0.119C no one reported last month.

Here’s how it’s computed. I subtracted the 20-year average values for each day (indicated by the yellow curve below) from the 2009 values for each day (indicated in green).

Figure 1: Channel 5 AMSU

This created a “Daily Channel 5 Anomaly”; I’ve plotted anomalies for a number of years below:

Figure 1: Channel 5 Average for June.

The black horizontal bar above represents the 30-day average for June, which despite VG’s fondest wishes that UAH goes negative this month, the related average of Channel 5 remained positive.

Time to guess the other anomalies!

Do remember: channel 5 is an imperfect approximation for this month’s UAH lower tropospheric value. However, it does correlate fairly well with that value.

I computed the difference between the UAH and monthly average Channel 5 values for all months back to 1999 and found that has a standard deviation of 0.08C. I also found during that period, the Channel 5 anomaly is 0.017C lower than the UAH anomaly.

On this basis, and a few other numbers, I’m going to be bold and predict that Roy will post a value of 0.02C this month. I will place absolutely no confidence interval on this.

I invite anyone who wants to take a shot at a guess to post one. You need to guess fast though, because unless Roy is on vacation, you know he’ll post the official values shortly!

Wow–what a July open! It feels more like May! Wednesday continues on track to become the coolest July 1 since the 65-degree high in 1930. The official high at O’Hare since midnight Wednesday as of this posting has been 65-degrees–the “normal” July 1 high here is 82. All this means we’re experiencing May-level temperatures in early July–it’s quite unusual.

I love warm weather. It did finally warm up last week and I wanted it to stay that way. I’m wearing a sweat shirt and winter pants. Shoot! Tom does promise warm weather lovers an end to the cooth.

Also, a set of medium range computer forecast models–including the National Weather Service’s GFS model (which is run out to 384 hours four times a day) and the European Center’s ECM Ensemble model, continue indicating a big dome of hot air is to puff up across the nation’s mid-section next week–probably mid and late week. The jet stream retreats north in this scenario. But the scenario does open the possibility of re-establishing a “ring of fire” type pattern in which clusters of thunderstorms flare at the periphery of the hot air, running from the Rockies into the northern Plains and possibly affecting the central and northern Midwest. If true, cooler outflow boundaries could have an effect on the northward extent of the hot air. But the building depth of the hot air mass is significant and should comfort those who prefer hotter weather than today’s that summer’s heat is far from history.

After a cool start to June, we did briefly experience the “ring of fire” and did hit the 90s. The cool patches dominated and we had the coolest June in 23 years. We also experienced the lots of thunderstorms and had the cloudiest June since the ‘69.

I’ll look at the bright side: El Nino is coming. At least fall and winter should be warm. But I hope we avoid drought in August. I hate watering the lawn. Though, I guess with the current soil moisture crops should skate through even if the rain levels fall.

In comments, on articles in the “Fishy” series, people have been asking about the whole “end point condition” issue and how it relates to Rahmstorf’s foolish “method” of testing projections. (The Rahmstorf’s chosen method seems to involve a) guessing future data and b) smoothing using a filter with m=”number of years Rahmstorf currently likes”. I will illustrate a slightly different method, that still involves these steps, bu uses a different end point criterion.)

Today, I am going to mostly discuss a method of guessing the future data used with some smoothing methods: Mannian Minimum Roughness Criteria (MRC) for short. ( I determined this method was used by reading the caption , “All trends are nonlinear trend lines and are computed with an embedding period of 11 years and a minimum roughness criterion at the end.”)

What is the Minimum Roughness Criteria

The “Minimum Roughness Criteria” (or Constraint) was described by Mann in a paper entitled “On smoothing potentially non-stationary climate time series “, Mann describes three possible methods for dealing with endpoints when smoothing data. He describes the ‘minimum roughness’ constraint thusly:

Finally, to approximate the ‘minimum roughness’ constraint, one pads the series with the values within one filter width of the boundary reflected about the time boundary, and reflected vertically (i.e., about the ‘‘y’’ axis) relative to the final value. This tends to impose a point of inflection at the boundary, and leads the smooth towards the boundary with constant slope.

That is to say, to “guess” the future value an analytst imposes the mathematical equivalent of using two mirrors at the end point. The first mirror would take a 3 point series ending with “y3″ and place a vertical “mirror” at the “y3″. This would and create a series that looked like this (y1, y2, y3, y2, y1). (If the analyst stopped there, he would be using the “minimum slope” constraint. When any smoothing filter is applied, this tend to cause the trend at the end of the data to flatten out. )

If the analyst is convinced that existing trends (whether positive or negative) persist over time, they might decide to use another mirror. So, how they put second virtual mirror horizontally at point “y3″. Now the data look like this: (y1, y2, y3, y3+ (y3-y2), y3+ (y3-y1). This is the “minimum roughness” constraint. When any smoothing filter is applied, this will tend to maintain the trend established in within the smoothing window.

If we use recent observations of global surface temperature from HadCrut or GISS, data up to year end 2008 are shown with solid symbols below; the MRC guessed data are shown with open symbols (11-year smoothed data through 2008 are shown with orange and red curves):

Figure 1: Data smoothed using mannian mrc and m=11 years.

As you can see, because the temperature in 2008 is low relative to data near 2007, the Mannian MRC method causes the analyst to guess that 2009 data will be lower than 2008 data; in fact, the method guesses the temperature drop from 2007 to 2008 will recur. Because 2008 is lower than 2006, the guess for 2010 is also lower than 2008; the analyst assumes the temperature will drop the amount it dropped from 2006 to 2008. This continues, but eventually, if there actually was a positive trend in the past, we will reach a point where the past is colder than 2008, and the Mannian MRC method will “guess” the future temperature is warmer than 2008. For annual average surface temperatures ending in 2008, MRC ‘guesses’ a future temperature greater than seen in 2008 will occur in 2016.

Are you laughing at those guesses? The best word I can think of for using this method to guess future data is “bogus”. The only way to avoid diagonsing the bog-o-city of this method of guessing future data is to never look at the guesses. But whether the analyst looks at the sorts of guesses the method makes, the “quality” of the types of guesses made for future data will propagate into downstream analytical steps.

How are bogus guessed temperatures used to create a smoothed temperatures used ?

To obtain a smoothed data curvve, annual averaged data are used smoothing over data over 11 years centered on the year to be smoothed. Back in 2007, the smoothing computation meant that the ’smoothed’ temperature for 2006 was computed using real temperature from 2001-2006, and guessed temperature from 2007-2011.

Of course, back in 2007, the guesses for annual average temperaure from 2007 - 2011 were different from those shown above. These temperatures were guessed based on temperatures ending in 2006 and looked like this:

Figure 2: How Mannian Smooth would look in 2006.

Smoothed values were duly computed, with those after 2003 computed based on guessed data. A paper was written. The fidelity of the TAR projections was diagnosed based on smoothed data computed based on guessed values.

Time passed, temperature for 2007 and 2008 were observed; the guessed values for those temperature were wrong.

Because the guessed values for 2007 and 2008 were wrong, the guesses for future data also changed dramatically. This should not have been unexpected: Guesses for the future data that change dramatically when a new data arrives is a feature of the MRC method.

Using m=11,including the new correct data from 2001-2008 and the new guesses for 2009-2001, one could now compute the temperature for 2006. Since the temperature for 2008 was much lower than one would have guessed using MRC in 2007, and the guessed temperature for 2009 and 2001 are even lower, the value for the 11-year smooth temperature 2006 computed in 2009 was lower than the value computed using his guesses from 2006. In the end, all temperatures after 2001 were revised by the addition of observed temperatures in 2008 and the new lower guesses.

That pretty much explains the method for computing the 11-year trend with MRC end points and shows why the end points oscillate wildly.

Would you like me to explain any method for selecting m=11 in the first place? Or the method for deciding m=11 is no longer any good, and should be replaced by m=15? I can’t.

Update

I originally thought Rahmstorf used “Mannian minimum roughness”. I edited to remove the Rahmstorf parts.

Update July 1

Paul M made an important observation in comments:

Another important revelation from Jean S on the previous thread:
The caption on Fig 3 should not be changed from “smoothed over 11 years” to “smoothed over 15 years”. It should be “smoothed over 29 years”!
Rahmstorf et al 2007’s ‘Embedding period’ is the extension required, half the size of the smoothing window!
Perhaps this is another error that Stefan Rahmstorf “hadn’t noticed”?

This means that the only part of the smooth red curve in Fig 3 that does not involved guessed temperatures is the short section from 1985-1993!

Britain is turning into a pressure cooker as hot, steamy air and intense sunshine roasts the country and kicks off the first serious heatwave of summer.

My reaction: July is two days off. We’re in the northern hemisphere. It’s getting hot here too. What do you expect?

Then, I just about bust a gut when I read the temperatures corresponding to “roasts” or “pressure cooker”:

The Met Office expects a high of 30C (86F) today, but warns that later in the week it could climb to 32C (90F) or even higher. And there could be many more uncomfortable nights ahead as thick clouds help to trap the day’s heat and humidity, with night temperatures of 18C (64F).

Believe it or not, the average June high for my zip code is 84F; the average low is 57F. Many local weather readers may well prefer 75F to 87F, but they probably wound not described 86F as a “pressure cooker” or “roasting”. More likely, they’d suggest it’s a nice day to take a swim.

According to Wikipedia, the all time high temperature for Chicago is 109F. (Yikes!) The timesonline report an all time high for Britain of 101.3F.

Admittedly, right now, we’re sitting pretty relative to the UK. Our current temperature is 74F. I walked to the grocery store and bought veggies for dinner. I do feel sorry for the Europeans sweating in 86F temperatures. If you want to avoid running the air conditioning, I’d suggest wearing a damp towel on your head and sipping a cold mojito. That’s what I plan to do in July.

Figure 1: Comparison of change due to smoothing.

A little history

Back in 2007, Rahmstorf and 6 co-authors published a paper in Science in which they smoothed temperature data using m =11 years, and told the world:

“The global mean surface temperature increase (land and ocean combined) in both the NASA GISS data set and the Hadley Centre/Climatic Research Unit data set is 0.33°C for the 16 years since 1990, which is in the upper part of the range projected by the IPCC. Given the relatively short 16-year time period considered, it will be difficult to establish the reasons for this relatively rapid warming, although there are only a few likely possibilities. The first candidate reason is intrinsic variability within the climate system. A second candidate is climate forcings other than CO2: Although the concentration of other greenhouse gases has risen more slowly than assumed in the IPCC scenarios, an aerosol cooling smaller than expected is a possible cause of the extra warming. A third candidate is an underestimation of the climate sensitivity to CO2 (i.e., model error).”

In comments here today, JeanS suggested the obvious fourth candidate (which did not occur to Rahmstorf at the time), “Of course, the fourth candidate might have been simply the “trend” calculation method they used.”

Flash forward to 2009

Ramstorf published an updated figure which indicated he added data, but still used m=11. Jean S posted a question asking whether m=11 was correct, if he had changed to m=14.

Stefan now says:

[Response: Almost correct: we chose M=15. In hindsight, the averaging period of 11 years that we used in the 2007 Science paper was too short to determine a robust climate trend. The 2-sigma error of an 11-year trend is about +/- 0.2 ºC, i.e. as large as the trend itself. Therefore, an 11-year trend is still strongly affected by interannual variability (i.e. weather). You can tell from the fact that adding just one cool year - 2008 - significantly changes the trend line, even though 2008 is entirely within the normal range of natural variability around the trend line and thus should not affect any statistically robust trend estimate. -stefan]

Hindsight? Hindsight?!

The discovery that Stefan’s method was flawed may have occurred to Stefan Rahmstorf, his many illustrius coauthors and peer reviewers for Science in hindsight, but “The Blackboard” readers will recall that the method Rahmstorf used was criticized in the blogosphere immediately after publication. The main reason: Outside the climate community, it is widely known that testing theories by smoothing data and eyeballing the curve is an abysmally poor method to test any scientific theory or model.

That is: the method Rahmstorf used has been known to be a very poor method for testing projections.

Period.

This has been known for eons. For this reason, few would bother to submit a peer reviewed paper on the subject because the discovery of how bad the method is not new. (Peer reviewed articles are supposed to present something new information. So, the the paper would, justifiably be rejected on the basis of simply rehashing widely known information.)

It is well also known that “Rahmstorf” method is particularly bad near endpoints, and this is true for any and all choices of smoothing, including both m=11 years and Stefan’s current choice of m=15 years. For those wondering which data are near endpoints: when using Rahmstorf’s method, the “end” region is the region within m/2 years of the final data point. So, using m=15 years, the temperature in the region within 7 1/2 years of the end point is sensitive to the addition of a new data point. So, if “robust” means we will conclude the same thing next year we do this year, using m=15, “robust” comparison of any sort can only be made between projections and smoothed data prior to 2001, which is the year the projections were made. (In 2007, Rahmstorf used m=11, but data end in 2006. That meant that the “robust” portion of his comparison as, once again, periods before 2001!)

So, the short criticism of Rahmstorf’s method: Stefan used a bad method (smoothing) and then applied it in the region where it is especially bad. He continues to use the method, evidently believing he can “fix things up” by changing m=11 to m=15. He “discovered” this problem when new data trickled in and no longer gave the answer some believe Stefan is inclined to believe to correspond to “truth”.

As a method to test projections, the method Rahmstorf selected remains a foolish choice.

As there is no existing statistical principle dictating any correct or best choice of “m”, there will be some who suspect Rahmstorf’s choices have been unintentionally influenced is by confirmation bias. It is plausible to suspect his choice of “m=11″ in 2007 was unintentionally influenced by his tendency to believe the higher trends associated with the model projections in the AR4 are more likely correct than those in the TAR. His improved “hindsight” dictating his current choice of m=14 might be similarly influenced by a belief the current flat trend cannot be correct, combined with a wish not to admit that the his of “slide and eyeball” methodology was obviously foolish in 2007 . Rahmstorf’s choice of method remains foolish to this day.

To show that the problems with the method were not discovered in “hindsight”, it’s worth nothing that David Stockwell , Ian Castles who was particularly vocal in my comments (and I believe he wrote letters to the Australian government) and I all criticized the method for various different flaws.

Some readers will be interested to read the David Stockwell not only commented back in 2008, but took the step of writing a formal article showing how the results and conclusions of in Rahmstorfs 2007 paper changed with the addition of new data. David has now published a note in Energy and Environment (ee-20-4_7-stockwell2) . David’s note discusses the peer reviewed article appear in Science IPCC by lead authors Rahmstorf, S., Cazenave A., Church J.A., Hansen J.E., Keeling R.F., Parker D.E., and R.C.J. Somerville. David’s abstract explains:

The non-linear trend in Rahmstorf et al. [2007] is updated with recent global temperature data. The evidence does not support the basis for their claim that the sensitivity of the climate system has been underestimated.

Mind you: one might suggest that, to some extent the method Rahmstorf selected and conclusions are now shown to be wrong only because the temperature in 2008 fell. He’d be looking ok if weather noise had one a different way and temperature had remained level or risen. But, in that case, his method would happen to have given the ‘correct’ answer by accident. That is the way with statistical analysis of stochastic systems. Sometimes, results based on bad methodologies turn out to be right by accident. In fact, because all statistical tests reject hypotheses only at a particular confidence level sometimes, results obtained using correct methods are shown to be outliers later. (That is: things that happen only 5% of the time when a hypothesis is true, will happen 5% of the time!)

But the fact is, Rahmstorf method of testing models was obviously flawed back in 2007 when first published.

Consequently, it was little surprise the results did not hold up when data from 2008 were added to the “analysis”. Not surprisingly, to prevent his original method from showing a flattening of the temperature trends, Rahmstorf needed to “tweak” his arbitrarily chosen parameter of m=11 when later data arrived. Even tweaking, the Rahmstorf method still overturns his main conclusion in Rahmstorf 2007: with both m=11 and m=14, the observed temperature anomalies since 1990 do not fall in the upper range of the TAR projections.

So, using Rahmstorf method (which in my opinion is a silly error prone method, the observed temperature anomalies fall pretty well in the center of the TAR projections.

What of the more recent AR4 which projects more warming than the TAR? A similar analysis applied to the AR4 would indicate that temperature anomalies fall below the multi-model mean trend in the AR4. However, making conclusions based using this method based on the current comparison to the AR4 would be just as foolish as Stefan making conclusion based on the results he showed in his 2007 paper.

Whatever Stefan may currently believe, his premature conclusions in 2007 were not based on using m=11 rather than m=15. Stefan was led astray by trying to “analyze” data using smoothing, which is a very bad method of testing hypotheses. The fact that the change in smoothing was inadvertently not disclosed to readers, and was also not explained in “The (Copenhagen) Synthesis Report” is unsettling.

Readers who are now aware of the change will anticipate that the trend at the end point will continue to oscillate. Since 2009 will likely be warmer than 2008 and we are due for an El Nino, I don’t anticipate Stefan discovering the value of “m” needs to increase to 17 next year or even the year after. But who knows what will happen when the next La Nina hits!