“It turns out that none of our models were totally correct,” says Dean Pesnell of the Goddard Space Flight Center, NASA’s lead representative on the panel. “The sun is behaving in an unexpected and very interesting way.”

http://www.redorbit.com/news/space/1697053/new_solar_cycle_prediction/

Dude, the sun warms the earth. Go outside on a sunny day to test this theory. When the earth gets hotter and cooler, solar phenomena should be the first place to look, according to Occam’s razor.

There are oscillations that appear to be tied to the sunspot cycle.

http://www.john-daly.com/theodor/DecadalEnso.htm
http://www.john-daly.com/sun-enso/revisit.htm

And weather is cyclic:

In the 1890′s there was global cooling, but then in the twenties there was obvious warming with 1934 being the hottest year in the twentieth century. Cooling again from the 40′s through the mid-70′s followed by warming in the latter part of the 20th century.

1922: http://docs.lib.noaa.gov/rescue/mwr/050/mwr-050-11-0589a.pdf
1933: http://docs.lib.noaa.gov/rescue/mwr/061/mwr-061-09-0251.pdf

If the blog will let me, here’s more:
http://wattsupwiththat.wordpress.com/2008/03/16/you-ask-i-provide-november-2nd-1922-arctic-ocean-getting-warm-seals-vanish-and-icebergs-melt/

Suggest you read something from a climatologist. This is from R. Spencer, research scientist at the University of Alabama in Huntsville: http://www.energytribune.com/articles.cfm?aid=828

And our evidence against a “sensitive” climate system does not end there. In another study (conditionally accepted for publication in the Journal of Climate) we show that previously published evidence for a sensitive climate system is partly due to a misinterpretation of our observations of climate variability. For example, when low cloud cover is observed to decrease with warming, this has been interpreted as the clouds responding to the warming in such a way that then amplifies it. This is called “positive feedback,” which translates into high climate sensitivity.

But what if the decrease in low clouds were the cause, rather than the effect, of the warming?

Let me define “climate” for you: Climate is the average weather over a large region, or the entire world, over a period of thirty years or more. In any case, the trend in the past few years is up, not down …

As I asked TheBlackCat, when would entertain the notion that you’ve crested a hill and started down the other side? Would you wait to hit bottom.

About the only thing that might cause global warming hysteria to end will be a prolonged period of cooling…or at least, very little warming. We have now had at least six years without warming, …

Of course, that’s not a trend.

Thank you for the comment.

If the Sun were a ball of hydrogen, heated by H-fusion in the solar core (the standard solar model), then your statement would be valid.

In fact, Nobel Laureate W. A. Fowler noted several years ago that the Sun is so massive that its annual energy output would drop by a negligible fraction over many years if H-fusion completely stopped in the solar core.

Such an object would not exhibit cycles of sunspots nor squirt out massive eruptions of energy like that Richard Carrington observed at 11:18 am on Thursday morning, September 1, 1859.

Although hydrogen covers the solar surface (like the red peel on an apple), many measurements since 1960 suggest that our Sun is the unstable remains of a supernova that exploded 5 billion years ago, ejected the material that now orbits it as planets, comets, moons and asteroids, and then re-formed on this unstable energy source:

http://www.omatumr.com/Photographs/Suns_core.htm

Within the past couple of weeks, NASA reported that such objects are explosively violent on very short time scales.

http://www.spacedaily.com/reports/Powerful_Explosions_Suggest_Neutron_Star_Missing_Link_999.html
With kind regards,
Oliver K. Manuel

No doubt there have been anomalous momentary brightenings and dimmings of the sun. But what counts is the average over a long enough time to make a difference. I find that the average solar constant for 1859 was 1365.3 watts per square meter (Lean 2000). The average for 1951-2000 was 1366.1 w m^-2. Not much of a difference there.

See

Lean, Judith 2000. “Evolution of the Sun’s Spectral Irradiance Since the Maunder Minimum.” Geophysical Research Letters, 27, 2425-2428.

[[If, indeed, the trend since 2001 is downward]]

The trend of seven years?

Let me define “climate” for you:

Climate is the average weather over a large region, or the entire world, over a period of thirty years or more.

In any case, the trend in the past few years is up, not down:

http://members.aol.com/bpl1960/Ball.html

[[You didn’t provide any reference for this but I found an AAAS paper. The CO2 lagged the antarctic by 800 (+/-200 years) but led the NA deglaciation.
How do you explain the discrepancy?
Why isn’t that a contraindication the CO2 forcing hypothesis?]]

Why would it be? We know from radiation physics that putting more CO2 in the air, all else being equal, will warm the ground. In a natural deglaciation the tiny warming from small orbital changes is amplified by the CO2 thereby released. It’s a feedback.

And the present warming is not a natural deglaciation. CO2 is leading temperature, and we know the CO2 is not from natural sources because of its radioisotope signature. Want the numbers?

[[***Small variations in the distribution of sunlight due to cycles in the Earth’s orbital eccentricity, axial tilt and precession cause a slight increase in the Earth’s mean global annual temperature. ***

Are those variables in the current models?
Are there any other variables?]]

No, they’re not in current models. The variations due to Milankovic cycles take tens of thousands of years to be noticeable. Modern global climate models cover a few hundred years at most. In some kind of Earth history simulation you might put in code to cover Milankovic cycles, but there would be no reason to do so in a GCM.

Your statement, “In general the solar output varies very little over the course of a year, less than 1%”, does not negate the fact that solar energy INPUT to Earth has been observed to change by a much larger fraction in a shorter time period.

[Remember, INPUT to Earth <<1% of solar OUTPUT.]

Thus, Richard Carrington observed blinding white light from a monstrous solar eruption at 11:18 am on Thursday morning, September 1, 1859. Following that event, “the entire Earth was engulfed in a gigantic cloud of seething gas, and a blood-red aurora erupted across the planet from the poles to the tropics.” [See “The Sun Kings: The unexpected tragedy of Richard Carrington and the tale of how modern astronomy began” by Stuart Clark

http://press.princeton.edu/titles/8370.html

With kind regards,
Oliver K. Manuel

One of the biggest problems in data analysis is noise elimination. Very often high frequency changes are the result of noise. To eliminate them, some form of low pass filtering is employed. A moving average is one example. To anyone familiar with electronics, a parallel capacitor works in a similar fashion.

One of the most severe filtering methods is a linear regression. All frequency components are removed and only the “DC” signal is retained. Despite this, a linear regression is highly desirable because it is one of the simplest mathematical models and offers more robustness than most higher order polynomial forms.

But like anything else, a linear regression can be an abuse. The primary problem is determining if the data really support a straight line and have a non-zero DC component.

One test is a statistical test of fitness such as Pearson’s “r”, the so-called correlation coefficient. Except for the extreme values (-1 and +1), it is difficult to interpret except that values close to either extreme can be considered significant.

Because of the interpretation difficulty, another test used is the Normal Q-Q plot, which is a plot of the residuals (actual differences from the model). It’s purpose is to visually depict how close the residual values are to being normally distributed. A regression is not valid unless the residuals are normally distributed regardless of the r2 value.

So what would one do if a regression of all data isn’t valid or if one is interested in other components beside the pure DC component? One method (outside of using a different regression type) is to use shorter line segments placed (hopefully) where the slopes are more or less linear. The problem here is deciding which frequencies to allow and which to discard but this is not much different than the decision to totally discard ALL non-DC components by regressing the complete dataset.

The endpoints selected in a segmented regression should correspond to changes in slope in the component of interest. Selecting which is a judgement call.

One of the trickiest areas in analysis is what to do with seeming outliers. It’s a problem as tough as distinguishing between flowers and weeds. Outliers can be caused by measurement error or they might be real.

The first problem is identifying them. One method is a boxplot of the residuals and a rejection all values lying outside of the “whiskers.” Most judgement calls of outliers are usually unconscious application of boxplots. The second and most important problem is what to do about them. Often, the best approach is to NOT ignore them — especially when they can’t be explained away.

In summary, the important things in a regression are: 1) what was included (and what was not); 2) knowing if the regression is valid: straight line QQnorm plot and high r2; and 3) identifying the purpose. How it’s done depends upon what is being sought. The only incorrect method is one that doesn’t accomplish the goal.

So, what if the question is, “OK sure there have been past increases but is that continuing?” The first step is to see if recent trends support the idea. If not, the answer is then obvious. But one thing is clear: you can’t answer the question if you insist on using points on both sides of a major zero derivative. Any insistence effectively is demanding that the question not be answered.

—

There are other types of regression. An important one is auto-regression but going there would require too much more space. This post is too long as it is.

BTW and FWIW: the plots that I have linked in previous posts aren’t mine.

If you don’t know what a Normal Q-Q or box plots are I suggest the many explanations which can be found around the net. Wikipedia has some good explanations.

Trend lines can be altered simply by choosing different endpoints. The trend line plot I’ve linked has differing endpoints; the most obvious is the 2001-2008 line.

Fair enough. I’m not sure how I missed that but I did.

At the very least, theses:

If, indeed, the trend since 2001 is downward <

and

Trend lines can be altered simply by choosing different endpoints. The trend line plot I’ve linked has differing endpoints; the most obvious is the 2001-2008 line.

TBC: “Which rhetorical trick of theirs did I use?”

How about cherry-picking by ignoring what was said?

TBC: “Unfortunately analyzing time-series data is not my area of expertise. ”

Yes, obviously, which makes your self-assuredness (“complete and total garbage without any usefulness whatsoever”) all that more jackass-like.

Despite your rudeness I will answer some of your points in my next post. It will take some time. try to be patient.

TBC, as usual, you completely have ignored what I have said to puff your own image of self-importance.

What did I ignore?

You apparently despise “deniers” but have no compunction against using the rhetorical tricks attributed to them. Shame on you.

Which rhetorical trick of theirs did I use?

In case you aren’t aware, 1998 was an El Nino year. It’s hardly an extraneous outlier.

El nino is, by definition, an unusual event. To see that it is an outlier all you have to do is look at how it compares to the rest of the data.

Also, in case you aren’t aware (and it appears you aren’t from your statement), the trend is regression line so it’s irrelevant what the starting value is. What’s important is what has been included.

You have obviously not hear of optional starting and stopping. With regression lines it is quite possible to pick the region over which you measure the regression in order to get the plot you want. If you pick a starting area with some anomalously high values and an ending area with some anomalously low values (which you can see is the case just looking at the data) then you will get a regression that slopes down. If you have a valid reason for them to pick those particular starting and stopping values, then I will reconsider my stance. I, however, can see none other than to get a certain outcome. Arbitrary selection of what range over which you wish to analyze the data is a classic method by which to fudge results to get the conclusion you want. That is a standard warning sign.

If you think that plot is garbage, then why isn’t this one or this as well? The former also includes the El Nino 1998. Both regress the entire dataset which is only valid if the residuals are normally distributed.

The first link is to this page, the second is totally illegible. I can draw no conclusions about the second plot because I cannot read it.

Who gets to decide what data are important and what are irrelevant outliers? You?

Including an outlier is of course important. That is not what they did in that plot. The selected several periods (approximately, they are not listed so I have to eyeball it): 1979-1993, 1993-1997, 1997-1999, 1999-2002, 2002-2007. So we have: 14 years, 4 years, 2 years, 3 years, 5 years. Why? Why these particular time period? Why did they choose to select the period around one spike, then separate out the low value then followed, then another that goes from a spike to a couple of low values? I am supposed to believe it is just a coincidence the last line includes the peak but not the slope or low area immediately before? Why have a 15 year period followed by a bunch of period of 5 years or less? The issue isn’t me deciding what is an anomaly, the issue is them picking what periods to do their analysis over and selecting those periods in such a way that they include anomalies in what appears to be an intentional manner.

If you’ve crested a hill and started down the other side, how long would it take you to realize it? Would you continue to claim you are still climbing simply because you are now higher than where you started regardless of recent trend?

I don’t know, but I do know that a a couple of low winters does not a trend make. I can also see lots of small ups and downs over 4 or 5 year periods in that plot, so I know that 5 years is not enough.

Unfortunately analyzing time-series data is not my area of expertise. I do know a few standard techniques for manipulating statistics, however, and this is one of them.

In case you aren’t aware, 1998 was an El Nino year. It’s hardly an extraneous outlier. Also, in case you aren’t aware (and it appears you aren’t from your statement), the trend is regression line so it’s irrelevant what the starting value is. What’s important is what has been included.

If you think that plot is garbage, then why isn’t this one or this as well? The former also includes the El Nino 1998. Both regress the entire dataset which is only valid if the residuals are normally distributed. Who gets to decide what data are important and what are irrelevant outliers? You?

If you’ve crested a hill and started down the other side, how long would it take you to realize it? Would you continue to claim you are still climbing simply because you are now higher than where you started regardless of recent trend?

I would be very interested in knowing what you think the proper baseline is and why it wouldn’t also be construed as cherry-picking. In fact, I challenge you to take the data and produce what you think is the “correct” methodology; show it to us and explain why you think it’s the best. I’m sure you will forgive me if I don’t hold my breath in the meantime though.

Andy, you may be interested in looking at this. It’s a preliminary plot of the decadal trends of all four datasets. Here’s a composite.

It’s too early to make any meaningful comment as yet. I am not saying (or ever have said) that GW isn’t happening — after all there was that Little Ice Age and things are clearly warmer now. I am questioning the methods used to establish the rate and the conclusions being drawn in re AGW.

Trend lines can be altered simply by choosing different endpoints. The trend line plot I’ve linked has differing endpoints; the most obvious is the 2001-2008 line. The most commonly used is a 30-year trend line. I don’t know if this is just convention or if it has a purpose.

Games have been played in the past. One of the most glaring is the plaque at the Badwater station which not only uses a 70+ year baseline but the station itself was positioned in a place guaranteed to be warmer than the surrounding terrain as if someone was trying for a world temp record. BTW: 30-year trend lines would show a different picture.

Here’s a 30-year plot of the RSS data (at RSS). The IPCC’s 2007 report said: “Six additional years of observations since the TAR … show that temperatures are continuing to warm near the surface of the planet.” The graph I linked is satellite data but obviously the IPCC is using a different baseline.

Given the variability in the record, I think it’s very difficult to say that the record proves AGW or to even say that such a conclusion is reasonable. If, indeed, the trend since 2001 is downward then I think it obvious that any CO2 contribution is swamped by far more powerful factors — the equivalent of spitting into the ocean.

Someone also plotted the difference between the satellite (UAH, RSS) and the surface measurements (GISS, HadCRU). I’ve lost the link. It’s buried in the comments. I’ll try to find it again. The satellite and surface measurements do appear to be converging.

Also, scroll down to the histograms and read the discussion.

Enjoy!

DAV, are you seriously doing a direct visual comparison between those charts? Look at the time axis on Hansen’s data; it’s starts before 1900, HadCRUT starts in 1988, UAH and RSS cover about 4 years, hence why the trend looks somewhat different to the much shorter time frames in the other charts. When you consider the longer term picture (i.e. Hansen’s 100+ year data), it is clear that Jan 2007 was not an oddity, but rather consistent with the ever increasing trend (the fact that, on its own, it was the highest is not evidence for global warming [in the same way that Jan 2008 in no way disproves global warming], but the fact that 2007 is consistent with the increasing trend adds further confirmation to the existing evidence).

Yes, it’s true Hansen claims 2007 was the hottest year on record but compare his data (GISS) to the other three. Note that Hansen’s has a more or less consistent positive slope that isn’t in the others. Of course, they ain’t NASA. Compare it to the UAH plot — oh, wait! — that’s that damnable denier John Christy again! — How the heck did he ever get his degree? But gosh! The RSS plot is very similar to Christy’s.

Anyway, all four showed that it was JAN 2007 that was the oddity in the previous year and all four showed a respectable drop from JAN 07 to JAN 08.

Mike, in all fairness, Phil probably doesn’t read here very often. This is a pretty old thread. GW/AGW is one of his blindspots when it comes to being skeptical. I don’t think his integrity is slipping.

Have you sent him an e-mail?

I see. So you believe it was AGW deniers who named the IPCC?

You also write “For some reason, people want to blame the Sun for global warming.

This, despite there being no evidence for it, and plenty of evidence against it.”

No evidence at all?

http://researchnews.osu.edu/archive/5200event.htm

I think you are dangerously approaching intellectual dishonesty here. There’s a great deal more undertainty than you let on.

I have to jump in here because this is just plain ignorant.

Ignorant is it? Stupid question, too, no doubt.

Over the past three decades, the number of datasets has exploded, ranging from proxy data from all over the world to satellite measurements. These datasets have been investigated many times from many independent research teams all over the world.

I can make a lot of claims too — but then I’m ignorant. You say many but I’d like to know: How many is many? What were the datasets used? Can you name five of them that are truly independent and aren’t just “adjusted” versions of the same one?

Strangely, the paleoclimatologists aren’t really an independent group — and they don’t constitute a very large group either. How many climatologists in general do you think there are? Are they collaborating together? If so, why would you still consider them independent? Just a gut feel?

Even assuming they got it “wrong” in the 1970s, modern climate science was in its infancy back then, only beginning to obtain the tools needed to understand and detect climate change.”

Yet it was the GC scare that gave the impetus to publicly fund climatology beyond its infancy. Today, it’s not a small industry in terms of funding. Maybe climatologists really have learned from the past.

BTW: the tools for detecting climate change aren’t much different than they were 100 years ago. Maybe you could provide a short list of new ones? Ill help you out: Satellites,…? Any new tool would have a relatively short track record. Very hard to get accurate trends from a timeline that only goes back a few decades.

Would you like to see some of the gyrations being done to the base thermometer record? The U.S. is the gold standard in weather stations. Go to http://www.surfacestations.org which is auditing all of them.