 | J Martin on Home made physics for rea… |
 | Richard111 on Home made physics for rea… |
 | mitigatedsceptic on Home made physics for rea… |
 | Stephen Wilde on Greenhouse Panic Over: New mod… |
| Stephen Wilde on Greenhouse Panic Over: New mod… |
 | tchannon on Twinkling clouds, insolation h… |
 | Brian H on Twinkling clouds, insolation h… |
 | Ian Wilson (aka Nind… on Twinkling clouds, insolation h… |
 | Frank Davis on Andrew Cooper: Were the recent… |
| Frank Davis on Andrew Cooper: Were the recent… |
 | Scute on Andrew Cooper: Were the recent… |
| Scute on Andrew Cooper: Were the recent… |
 | Roger Clague on Greenhouse Panic Over: New mod… |
| Stephen Wilde on Greenhouse Panic Over: New mod… |
| Frank Davis on Andrew Cooper: Were the recent… |
Steven Mosher says:
If you believe in scaffetta curve fitting ( which is dimensionally incorrect)
then I’ve got a better curve fit for you.. or rather Muller does….
For now understand that the SHAPE of the C02 curve is what matters. ONLY THE SHAPE. it is scaled..
you could think of it as the smoothed sum of all GHG forcings..
Hi Mosh. Isn’t “scaled” just a nicer way of saying “dimensionally incorrect”? 
“I’ve got a better curve fit for you…the SHAPE of the…curve is what matters. ONLY THE SHAPE.”
Now gentlemen, you know that beauty is only skin deep, and curves can be adjusted. What you need to look for the kind of thermometer siting that you can take home to meet mom, and who will always be respected by everyone in the end.
@ zeke
I agree with this scrutiny on Mosher. He gets away with a fair bit via his use of authoritiatve-sounding techno-bafflegab. He made a brief appearance on BH with this:
http://www.bishop-hill.net/blog/2012/7/28/best-guess-is-15.html?currentPage=2#comments
I challenged him to clarify his overall position:
He did not reply.
Steve McIntyre commented:
After carefully plotting all the vectors, I have concluded that ego is the forcing variable.
[ grin --Tim]
One of the many ASSUMPTIONS that have become Consensus is that rural stations would have less UHI than Urban stations. I know of no studies that have shown this. Is anyone aware of this type of study that can be linked??
Dr. Spencer has done some work that indicates that this is NOT an appropriate assumption.
In other words, Steven Mc’s suggestion that he would like to see verification of whether the stations are really rural still misses the mark. Has work been done to verify what is happening at each station. This would probably be best done by NASA with a comparison of temps between the location of the stations and outlying areas.
My main complaint of the MoshPup type studies is that they compare so-called rural and urban and find little difference. What it does NOT show is comparisons of urban/rural and relatively unchanged areas!! It also does no detailed analysis of urban/urban rural/rural to see the ranges in the 2 categories!! We may be seeing the same effect as the handle of the hockeystick. Large and small UHI’s at all stations averaging out!!!
I sat into the small hours with Mosh on a couple of occasions discussing all this stuff. We both know the pitfalls of the studies. However, in this case, Scafetta’s study, regardless of the dimensionless quantities and the fitted amplitude, gets the timing of excursions in the replicated proxy study pretty much right over thousands of years. Muller’s study is not only “scaled” for amplitude, but the questionable attribution of negative excursions to volcanic forcing isn’t very convincing when the cooling events begin sometimes years before the volcano in question erupted. It’s also much shorter in terms of the timescale covered – a couple of hundred years vs thousands.
Judith Curry said that sometimes volcanoes can become active quite a while before the main eruption sequence. I’m not convinced by that either, because we are told that volcanoes don’t have global effects unless the ejecta reach the stratosphere. This only happens on major eruptions.
I like Mosh, he’s a fun guy to talk with and he’s highly intelligent. However he needs to get a more even handed perspective on this stuff IMO.
Mosher really is a disappointment. It’s like he’s competing with Nick Stokes for a lifetime AGW apologist award. Sad.
I’ve been thinking about TOBS adjustments and how calculate an upper bound for them using the ISH hourly data here ftp://ftp.ncdc.noaa.gov/pub/data/noaa/.
The Karl TOBS paper here ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/karl-etal1986.pdf calculated TOBS bias for a hand picked subset of site(s) with data from mid-century. A huge job in the 1980′s.
IMO, it should be easier with the computing power available today to do the same thing for all sites and get a true upper bound for TOBS adjustments.
It would also be necessary to compare min/max for those days with hourly min/max for that day.
Algorithm.
Use Hourly data to get min/max (reset at midnight).
Calculate max, min and mean values +trend.
Use hourly data but reset at 2pm (it would be nice to run and compare with different values).
Calculate max, min and mean values +trend.
compare steps 2 and 4.
hey Rog. Live by the sword and die by the sword. People who believe in curve fitting ( supporters of scaffetta) should have no problem accepting Berkeley’s curve fit.
I dont think the berkeley curve fit supports the attribution argument that richard tried to make with it. two things were evident however. the volcanic stuff is very striking. ( willis gets it horribly wrong) and TSI explains nothing.
Eric.
the code to do the test to prove it to yourself is on CA.
go find it and run it.
Steven Mosher says:
August 9, 2012 at 5:40 pm
To determine an upper bound for TOBS from the Hourly data? I don’t remember seeing that, but if you can point me to it I’d appreciate it.
Hi Mosh, what about the volcanoes not included. What are the selection criteria?
Re UHI, Hans von Storch had this post http://klimazwiebel.blogspot.com.au/2012/07/studying-urban-heat-island-effect-on.html
I made these comments
cementafriend said… 1
It appears that you do not look much at literature. Torok et al (Aust. Met Mag 2001) did surveys of a few Australian small towns and came up with the formula UHI (delta T) = 1.42*log(population)-2.09. Prof Ole Humlum has done surveys across a number of towns including Oslo Norway and St Andrews Scotland.
I note the UHI effect just about every day with the outside temperature thermometer in my Japanese SUV. The difference between suburban village centre (ie shopping centre, and post office which once was an official BOM weather station) and the urban rural outskirts is 1 to 3C day and night of which roughly a low difference of 1C occurs on cloudy days and 3C occurs on cloudless clear days and nights. The Village centre is always higher.
cementafriend
July 29, 2012 7:08 AM
Anonymous said… 2
Not impressed with the manuscript. It does not compare the urban areas with true rural or native vegetation areas.
The Prof Humlum reference is here http://www.climate4you.com/UrbanHeatIsland.htm#Urban%20heat%20island%20effects%20in%20Oslo
cementafriend
July 29, 2012 7:54 AM
This article http://www.wur.nl/NR/rdonlyres/5F932193-414C-4A91-B1F9-119362FC63FE/110556/Heusinkveld_Biomet_extended_abstract_26feb2010s.pdf. mentioned in the last comment on the von Storch post (see above) is interesting.
In the article it is mentioned that city parks are cooler than surrounding areas and that treed [wooded] areas are cooler. I have also measured (but not scientifically recorded) that forested areas are cooler than open grasslands. It is likely that the class 1 stations in Watts study (ie having 100m clear of very low vegetation) may still give temperatures above forested areas or the bush (as it is called in Australia).
Watts et al draft line 293 on:
This instantly nailed for me exactly why Muller had arrived at results I was absolutely certain were not properly excluding UHI. I was already very sure Muller was wrong, from my own collection of “amateur” as well as professional studies, that started with my detailed study of John Daly’s global temperature records – knowing that in real science, it only takes one piece of data that does not fit a hypothesis, to upset its credibility.
Watts merely showed the way to a statistically-irrefutable backup of what I already knew. Sure, we know Watts’ work needs to be cleared of TOBS biassing – but again, all the evidence suggests to me that the TOBS correction will be minor in comparison with the new Leroy recategorising.
http://rankexploits.com/musings/2012/on-volcanoes-and-their-climate-response/
tallbloke the selection criteria was simple. We used the data from Gao.
Gao, however, was missing one volcano so we contacted him and he pointed us to the
source for the missing data.
If one accepts the scaffetta approach, then one is quite stuck with it. That is, if you stood on the sidelines and did not criticize it, you are stuck when it comes to criticizing the same approach when applied by Muller.
The temperature record used is essentially the same as that used by Scaffeta. Except that we extend it back in time using methods that have been suggested many times by skeptical writers.
We then do a regression to see how much of the record can be explained by two factors
A) volcanic particulate matter
B) GHG ppm.
Turns out, that these two explain most of the record with a residual that looks like AMO.
Now, unlike scaffetta this regress actually has a physical theory that can turn ppm into Watts!
And then watts into temperature. so that the underlying model is dimensionally correct, unlike scaffeta which has no physical theory to turn positions into temperature.
Does this mean, as Muller suggests, that all the warming is due to man. I think that goes too far for such a simple analysis. but the simple analysis also showed that adding TSI did not improve the fit.
That is, it explained none of the residual variance.
Now, people who objected to scaffeta because he used a temperature record that they think is bogus are at least consistent. People who objected to his curve fitting are also on consistent grounds.
But if you bought his approach then, you cant object to it now when somebody uses it to show something contrary to your position.
MoshPup just can’t get away from his pop meaningless pronouncements:
“But if you bought his approach then, you cant object to it now when somebody uses it to show something contrary to your position.”
Uh, no Moshpup. The method has to be shown to be effective for the actual task involved. Curve fitting can be quite appropriate, or not. Broad rationalizations add nothing to the discussion.
The arguments advanced by Mosher are as usually non-senses. He simply does not understand the physical/mathematical issues of what I did in my papers.
Just to clarify the issue.
Regression models, as those that I use and those used to predict tides, for examples, are useful and physically appropriate only if they are proven to reproduce physical information outside the calibration interval used for getting the regression parameters.
For example, in tide predictions, what is done it to accurately calibrate a model made of numerous astronomical harmonics (up to 35-40 harmonics) for a certain period against tides observations, and then the model is used to forecast the tides for a certain period. And the method is found to work well because the harmonic components are evidently the right ones.
If, for example, the harmonics or some of them were wrong, the regression model would be still able to fit the period used in the calibration quite well. However, the model will have no predictive power, and will immediately diverge from the observations outside the regression interval.
Thus, it is the forecast/hindcast test that supports the physical accuracy of a regression model. It is this crucial test that distinguishes a proper way to use a regression model from simply “fitting the elephant “.
What Mosher does not understand is that in my papers I am supporting my arguments by carefully testing the models that I propose on their hindcast/forecast capability. For example, I calibrate the model for the period 1850-1950 and check whether the model reproduces the oscillations observed during the period 1950-2012, and viceversa. I calibrate the model during the period 1750-2010 and check whether the model captures major data patterns for thousands year before 1750. I calibrate the model for the period 1970-2000, and I check whether the model reproduces the period 2000-2012. And I demonstrate that the models that I have proposed are able to do a good job.
This is how the regression models have been always used in science since ancient times.
On the contrary, the curve fitting exercise made by Muller does not attempt any hindcast/forecast test. He simply takes a period and fit it with a set of forcing functions. It would be possible to get a good fitting of the same temperature record also by excluding the CO2 forcing from the model and letting the regression algorithm to find the other parameters to well fit the record, one can find that the warming would be almost 90% explained by CH4.
Only a hindcast/forecast test can demonstrates that a regression model is made of real physical constructors.
With the regression model that Muller proposed there would be no way to get the MWP and, for example, in UK today is as warm as it was during the MWP. This pattern cannot be get without a strong solar effect on the climate.
A far better regression model has been published by me in this work
http://scienceandpublicpolicy.org/originals/climate_change_and_its_causes_a_discussion_about_some_key_issues.html?Itemid=0
See the discussion in section 4.
I do not comment on the Mosher’s “dimensionally/scale” problem because it is clearly wrong and does not make any sense. He simply does not understand physics and its interrelation with geometry and mathematics.
Nicola: Many thanks for clarifying why it is that Moshes comparison of the planetary theory methodology and Muller’s is false.
I shall enjoy Fairport Convention’s Cropredy Festival all the more today.
Ahh, the difference between arm waving, mine and Moshpup’s, and actual knowledge and ability!!
I also thank you Nicola.
Nicola Scafetta says, August 11, 2012 at 7.23 am: …the curve fitting exercise made by Muller does not attempt any hindcast/forecast test. He simply takes a period and fit it with a set of forcing functions. It would be possible to get a good fitting of the same temperature record also by excluding the CO2 forcing from the model and letting the regression algorithm find the other parameters to well fit the record, one can find that the warming would be almost 90% explained by CH4…Only a hindcast/forecast test can demonstrate that a regression model is made of real physical constructors.
Muller and Mosher and others of the same ilk all make the same mistake so elegantly exposed by Scafetta. They assume, without being able to test their assumption, that because they have found a mathematical model that fits the data reasonably well over a particular period, it proves that their model is true over all other periods.
Thus Muller’s BEST crew have a mathematical model for their reconstructed temperature record from 1753 to 2011 that is largely a function of just two varibles: volcanism (“1.5 degC plus or minus 0.5degC per 100 Tg of atmospheric sulfate”) and anthropogenic effects (“logarithm of CO2 concentration”). But they have no temperature data prior to 1750 and none, of course after 2011. So they cannot do a hindcast or a forecast on their model, which is thereby rendered useless.
Mosher’s burblings about curve fitting simply reflect his adherence to the BEST style of untestable mathematical model.
No wonder BEST have failed peer review. And no wonder Mosher has become a laughing stock.
Since the BEST temperature series is land only, attempting to use it for attribution is a doomed project anyway. It ignores the ocean bulk, which contains most of the thermal energy stored in the Earth’s climate system.
If the IPCC does a Lazarus job on Muller’s paper and promotes it as state of the art climate science they will become (even more than they are already) the laughing stock of the climate community.
TB:
“Since the BEST temperature series is land only, attempting to use it for attribution is a doomed project anyway. It ignores the ocean bulk, which contains most of the thermal energy stored in the Earth’s climate system.”
Right. I’ve been wondering what all the fuss was about.
I can see that one of these days I’m going to have to get round to writing that boring post about which temperature records you can use for global comparisons, which ones you can’t use for global comparisons and which ones you can’t use for anything.
Posted at WUWT:
tallbloke says:
August 15, 2012 at 11:20 pm
Leif Svalgaard says:
August 15, 2012 at 1:59 pm
tallbloke says:
August 15, 2012 at 1:26 pm
“Bad papers are soon and deservedly blissfully forgotten.”
You have to get a proof that it is bad published
No need, a bad paper stinks to high heaven on its own and any scientist worth his salt can spot it.
This is a content free appeal to ‘consensus’. Along with your bad behaviour in rejecting Scafetta ‘s paper on spurious grounds you are evidently behaving more and more like Phil Jones.
“Kevin and I will keep them out of the literature somehow – even if we have to redefine what peer review is”
Well, you’ve failed. The W&P paper stands, and Scafetta has built on it as well as formalising and publishing the work we’ve been doing together on the Talkshop in the excellent and impactful JASTP. He has also performed proper statistical tests on out of sample projections of the model which demonstrate its validity.
Note however, that we are not claiming the Leighton-Babcock dynamo doesn’t exist, but that its behaviour can be modulated by the interaction of cyclic tidal/magnetic/momentum effects caused by planetary motion. We believe this is the best current explanation for the obvious and irrefutable correlations which we have discovered between planetary motion, solar activity and proxy records of changes in Earth’s climate going back thousands of years as well as with the modern temperature record which the co2 driven theory can’t even successfully hindcast past 1960.