Satellite measurements of total solar irradiance (TSI) come from a number of different satellites, each of which produces its own data set, each with its own calibration issues. In order to form a single continuous estimate of TSI since satellite measurements began, the various data sets must be “sewn” together into a composite; one hopes that the resultant tapestry will be seamless.
In another post, a discussion arose about the use of the PMOD composite by Lockwood & Frohlich to suggest that changes in TSI cannot be responsible for modern global warming. It’s no surprise that Lockwood & Frohlich chose to use the PMOD composite, as Frohlich is one of its authors. But there are other choices; the chief “rival” is the ACRIM composite. On another blog it was suggested that Lockwood & Frohlich’s entire results depended upon trusting the PMOD composite, while the ACRIM composite was superior. The suggestion came from none other than Dr. Richard C. Willson, one of the authors of the ACRIM composite:
Lockwood & Frohlich’s thesis depends entirely on the PMOD TSI composite time series and its lack of a significant TSI trend during solar cycles 21-24.
This is not just a false, but a disingenuous claim. Lockwood & Frohlich use multiple data sets, including sunspot counts, the interplanetary magnetic field, and neutron counts, to explore multiple possible physical mechanisms by which the sun may influence earth’s climate. More to the point, they make this crystal clear in their paper. Dr. Willson claimed:
… the ‘other’ solar activity features you mention that LF examine are only proxies for TSI, not potential climate forcings themselves.
Lockwood & Frohlich make it clear that they use magnetic field strength and neutron counts to look for trends in cosmic rays; the production of cloud condensation nuclei by cosmic rays is the chief mechanism which is proposed to regulate climate, other than direct forcing by solar irradiance changes.
Regardless, a reader here wonders
Was it appropriate for Lockwood to use an 11 year solar cycle instead of a 22 year cycle?
Can you please elaborate? It is well known that the sun has 11 AND 22 year cycles. Why were the numbers run for one and not the other
And a commenter suggests,
Same with ACRIM/PMOD versions of TSI. Run for both and let the chips fall.
It’s worthwhile to compare the two chief rivals, PMOD and ACRIM, to address the question, which is likely to be more correct, and to consider whether they give different results when tested for possible trends which might be responsible for climate change. In this post I shall compare the results obtained with each composite; in the next I shall compare the two composites, and consider the reasons for preferring one over another.
Both composites are available for download; the PMOD data can be found here, while the ACRIM data are downloadable here. The two composites are graphed here:
The clearest feature in each is the roughly 11-year solar cycle; TSI fluctuates along with the solar cycle, changing by 1 W/m^2 or more between maximum and minimum. First, let’s estimate the magnitude of the influence of solar changes on climate. Because earth is a sphere, and the surface area of a sphere is one fourth its cross-sectional area, a change of 1 W/m^2 in solar irradiance causes a change of only 0.25 W/m^2 in climate forcing.
A good estimate of the effectiveness of climate forcing is that every 1 W/m^2 of climate forcing produces, eventually, 0.75 deg.C change in global average temperature. The 0.25 W/m^2 or more change in climate forcing throughout the solar cycle would therefore change global temperature by roughly 0.1875 deg.C, if it persists long enough for its effect to be complete. But due to the immense thermal inertia of the oceans, it takes time for the full impact of climate forcing to take effect. The influence on TSI of the solar cycle goes both up and down, so unless there is a persistent trend, over the long haul the ups and downs cancel with net effect zero. In fact, the 11-year solar cycle is barely, if at all, discernable in earth’s global temperature signal.
Hence we must remove the influence of cyclic changes in order to identify the presence of trends, to identify changes in the sun which can influence global temperature. There are indeed cycles other than the 11-year cycle in solar behavior, most prominently a 22-year cycle which involves reversal of the polarity of the solar magnetic field. But do these cycles apply to TSI? To test for the existence of longer cycles in TSI, we’ll need TSI estimates over a longer time span than the mere 28 years covered by satellite data. The best available proxy reconstruction of TSI before the introduction of satellite measurements (and the one which agrees best with satellite measurements during their period of overlap) is the reconstruction of Lean (2000), graphed here:
If we look for the presence of cycles using Fourier analysis, this is the result:
The very-low-frequency response has the character of red noise; it represents the response to the long-term changes. There is no evidence that these changes are in any way periodic. In order to look for genuine cycles (especially the 22-year cycle), we should remove the long-term changes, which can be done using a low-pass filter. This gives the following long-term evolution:
We can now scan the residuals from this fit, to look for cycles, giving this result:
There is still a hint of red noise (although now it is below the level of statistical significance), and there is a strong broad-band response to the 11-year solar cycle, but there are no other periods present with anywhere near statistical significance. And there isn’t even a trace of a 22-year cycle. In case that signal may be transient, or may change period sufficiently to suppress its response to Fourier analysis, we can look for it in a wavelet transform:
Again, the 11-year cycle is evident, although its presence is time-dependent (it’s absent from about 1640 to 1720, during the “Maunder minimum”). But there is no hint at all of a 22-year cycle (or any other cycle). Therefore, although the sun does indeed have a 22-year cycle, that cycle appears not to manifest itself in changes of TSI. Hence when studying the trends in TSI, it is indeed safe to ignore those cycles which do not affect it, in particular the 22-year cycle. When looking for trends in TSI, it is necessary to remove the impact of the 11-year cycle, but no other.
How then shall we remove the influence of the 11-year cycle in order to discern persistent rather than cyclic TSI changes? Willson & Mordvinov choose to characterize it by comparing values at the minimum of the solar cycle, as illustrated in this graph from the ACRIM website:
Lockwood & Frohlich use moving averages with a number of different averaging intervals, thus:
Hence not only do the two research teams use different TSI composites, they use different methods to characterize their changes apart from the 11-year solar cycle.
Characterizing the secular evolution by using only the values at minimum seems to me to be astoundingly naive. The climate forcing due to TSI throughout the solar cycle is certainly not simply the result of its minimum value, and it’s folly to ignore the difference between the maxima of various cycles (or the rest of the cycle, for that matter). It’s noteworthy that according to the ACRIM composite, the maximum of cycle 21 is considerably higher than that of any of the succeeding cycles. Hence considering values other than only the minimum will weaken, if not remove, the upward trend posited by Willson & Mordvinov.
I’ll adopt a more basic, but time-tested method. I’ll remove the best-fit sinusoid from each data series, leaving residuals which will not represent a perfect removal of the cyclic influence, but will certainly remove most of its effect. This leaves the following residuals, which are plotted together with the trend in each according to a low-pass filter:
We can directly compare the secular evolution from ACRIM and PMOD:
The ACRIM residuals show far greater variability than the PMOD residuals. But even accepting the ACRIM data, there’s no evidence for anything close to the trend claimed by Willson & Mordvinov. What they have done is to characterize the secular evolution by using only the values at the times indicated by the arrows in this graph:
In fact the much-touted increase of 0.04%/decade is based on using only the values at the first two arrows! But it’s evident that this does not tell the whole story. If one starts at the beginning and ends at the ending, then the trend identified would be distinctly decreasing rather than increasing! That’s due to the very high maximum of cycle 21 in the ACRIM composite. A better idea would be to fit a trend line to the residuals, which yields a slope of 0.01 W/m^2/yr, or 0.1 W/m^2/decade. This does indeed represent an upward slope, but the rate is nowhere near the 0.04 %/decade touted by Willson & Mordvinov; it’s a mere 0.008 %/decade. If that trend were sustained throughout the 28-year duration of the data, it would lead to a net secular increase in TSI of 0.31 W/m^2, which would cause an increase in climate forcing of 0.076 W/m^2. That’s only marginally bigger than the climate forcing due to anthropogenic power generation. At a climate sensitivity of 0.75 deg.C/(W/m^2), it would lead to a net global temperature rise of 0.06 deg.C, far smaller than what is observed. Even using the ACRIM composite, satellite estimates of TSI will not support the idea that TSI changes are responsible for modern global warming.
The PMOD composite, however, shows a slight overall decline in TSI. The biggest differences are from 1979 to 1982, at the very beginning of the satellite observations, and from 1989 to 1991, the time of the “ACRIM gap” during which ACRIM satellites were not available to measure TSI directly (an unfortunate result of the Challenger disaster). Other satellite measurements are used to bridge that gap, and the difference between the ACRIM and PMOD composites depends on how that process is carried out. It’s interesting that the biggest increase of ACRIM over PMOD, and in fact the biggest increase in the ACRIM residuals, occurs at exactly the time of the ACRIM gap. But that is the subject of an imminent post…
UPDATE UPDATE UPDATE
A commenter states:
You are wrong here I think. There has been no acceleration in recent years. In fact, if anything, there has been a levelling off.
He refers to a graph from NASA GISS, showing the monthly mean temperature anomaly, both based on land+ocean data, and on meteorological stations, since 1997.
Let’s take a look at the actual time period in question (1978 to the present). Fitting a trend line to the data will not reveal whether the rate of warming is accelerating or decelerating, it will only give its average value. But if we fit a 2nd-order polynomial (a quadratic), we will get an indication of whether the rate of warming is getting faster or slower, and we can also compute a test statistic to determine whether or not the result is statistically significant. For the land+ocean data, we get this quadratic fit:
Also, the quadratic fit is statistically significant, so in fact during the time period in question, global warming has accelerated, not decelerated. For the meteorological-station data, we get this quadratic fit:
Again, the quadratic fit is statistically significant, so in fact during the time period in question, global warming has accelerated, not decelerated.
The commenter also states:
look at for example the GISS data at http://data.giss.nasa.gov/gistemp/graphs/Fig.C_lrg.gif - this shows no apparent increase over the last 7 - 10 years.
Let’s consider the smaller range suggested (the last 7 years, or more conveniently, since 2000), since smaller data sets are a greater statistical challenge. To determine whether or not a trend exists, we can simply fit a trend line. This gives:
Both series show an increasing trend, and in both cases the trend is statistically significant.
Using only the last 7 years (starting with July 2000 rather than January) gives the same result. Using the last 10 years (starting with 1997 rather than 2000) gives the same result.
UPDATE #2 UPDATE #2 UPDATE #2
A commenter asks a number of interesting questions:
Tamino, does the positive trend in your linear fit post 2000 T’s agree with that predicted or calculated for the same period by your quadratic fit of 1998-2007 T’s?
Yes (note: the quadratic trend was for 1978 to 2007, not 1998 to 2007). Here’s a graph of the two computed trends (data in black, quadratic trend in blue, linear trend in red):
The slopes during the last 7 years are not identical — the linear trend for the last 7 years is a little steeper than the quadratic trend over the last 29 years — but they are within each other’s error limits, so statistically speaking, they are in agreement with each other.
Also, can you duplicate this result with HadCRUT3 data?
Yes. Both the linear trend for the last 7 years, and the quadratic term of the quadratic trend over the last 29 years, are statistically significant in the HadCRUT3 data set.
And, what is the justification for a quadratic fit of the data rather than a linear trend.
The motivation for a quadratic fit is to investigate acceleration of the trend over the time period covered by satellite data; a linear trend cannot identify this. The justification is that the quadratic term passes statistical significance tests.
The reason I ask is that it seems you can choose many ways to slice and dice and give statistical significance, yet get different results. Is there a rule that directs us to the correct method for analysis?
When statistical significance is properly applied, and results are compared within their error limits, the results actually are not different, they’re in agreement. I’ll mention that it’s pretty much impossible, statistically, for them to be in perfect agreement; the random fluctuations present in the data prevent that. But if the results are correct, they will agree within their error bounds.
There’s no single hard-and-fast rule to determine which analysis method should be applied. The method should always be statistically testable, in order to avoid being fooled by purely random fluctuations. It’s generally preferable to choose the simplest test possible, because its statistical behavior is usually well understood and there is more experience using it. It’s usually best not to include too many degrees of freedom in the model — no more than are necessary to answer the question under consideration — because that will dilute the statistical significance of the results. It’s also generally a good idea to closely examine the residuals (what’s left over after subtracting the identified pattern from the data), and to at least consider alternate explanations of the behavior of the signal in order to avoid being fooled by rare but “bizarre” phenomena.
I too tend to be highly skeptical when people (especially bloggers, but research scientists too!) resort to a variety of different statistical methods to establish results. I often suspect that they haven’t applied statistical significance tests properly, that they haven’t computed error estimates for determined parameters, that they may not be aware of the correct “null hypothesis,” and that they may have simply tried one analysis method after another until they found one which returned the desired result. The latter is a subtle form of “cherry-picking” analysis methods; if multiple methods are applied in order to answer the same question, the statistical significance level must be adjusted to compensate for the increased probability that one of them will “hit” purely by accident, but this, alas, is almost never done.
In fact, I would suggest that possibly the greatest single source of mistaken results in peer-reviewed scientific literature is the improper application of statistical analysis. This is in large part because astronomers, physicists, chemists, etc. are not statisticians; there are many subtleties involved of which scientists may not be aware. However, I’m a specialist in the statistical analysis of time series (and have made some valuable contributions to this mathematical discipline), so I trust myself to choose analysis methods appropriately and return reliable results. But I’m certainly not infallible.
68 responses so far ↓
George // July 24, 2007 at 6:00 pm
I find it very surprising that Willson would have used just a couple points to determine change in solar irradiance from one cycle to the next.
Given the large amount of therml inertia involved, from the standpoint of its impact on the earth’s temperature over the long term (climate), isn’t the time average of the irradiance over a cycle really the thing that is most important?
In other words, if you want to see whether the irrdiance is increasing (on average) from one cycle to the next, wouldn’t you want to compare the time average of the irradiance for subsequent cycles? — ie, take the difference between the two.
SomeBeans // July 24, 2007 at 7:34 pm
Nice work - it did seem odd that the claimed increasing trend on the ACRIM series was based on the values of the two minima only.
John Willit // July 24, 2007 at 10:30 pm
Please chart the solar cycle length from minimum to minimum based on the 11 year cycle.
Ian Rae // July 24, 2007 at 10:51 pm
Fascinating. Thanks for working through that. Solar forcing seems to line up with the 1998 “warmest” year, but as you point out there is no long-term upward trend. If only we’ld have invented satellites in 1880!
nanny_govt_sucks // July 24, 2007 at 11:14 pm
The ACRIM high values in solar cycle 21 appear to come from “Nimbus 7/ ERB”. My understanding is that this is actual data.
The PMOD composite does not have the same high values in solar cycle 21. Since the only source of data for this time period appears to be Nimbus 7/ERB, it is apparent that PMOD has adjusted this data considerably, or used some other source.
What is their justification for doing so?
isaac held // July 25, 2007 at 12:34 am
Tamino, when estimating the forcing due to TSI variations, multiply by 0.7 after dividing by 4, to account for the Earth’s albedo — no point in making the solar forcing look bigger than it really is.
[Response: Oops! Right you are.]
EliRabett // July 25, 2007 at 12:35 am
Let me turn that around. If I by eye, compare the major excursions in the TSI plots (the short period negative going ones) they are VERY MUCH alike. The cross-correlation would probably be unity except before 1982. I therefore would trust neither reconstruction before 1982
paminator // July 25, 2007 at 1:14 am
Your ACRIM smoothed graph of residuals looks a lot like the lower troposphere satellite and radiosonde temperature record that peaked in 1998 and has remained flat since then. Why not plot the MSU or RSS lower trop data with the ACRIM smoothed residuals and see how they compare? As we all know, the lower trop data should have an even larger temperature signal (30% higher trend) from greenhouse effects than the surface temperature, and should therefore provide an even larger spike in the coffin of the solar hypothesis. Or not.
Geoff Sherrington // July 25, 2007 at 2:46 am
In structured examinations, one starts with the fundamentals and then goes to more detail.
I assume that the satellites measure the solar irradiance heading towards the earth and that the variation as measured to date is too small to exert much direct effect on surface global temperatures.
The question that follows is, does the solar irradiance (which shows some correspondence with suspot cycles) have other more secondary effects, such as as differential thinning of the atmosphere of the earth, that allows the sulight flowing past the satellites to be changed in intensity between the satellite and the earth (for clarity of discussion, assuming a measurement where the satellite is between sun and earth).
Can the wise men and women who examine these things please comment quantitatively for a non-specialist scientist who is trying to get order from chaos through scientific data management?
george // July 25, 2007 at 2:55 am
Given that any two points define a line, by Willson’s standard any two points also define a trend.
To call that naive is being overly kind.
Chris C // July 25, 2007 at 3:00 am
Good work, have you considered writing this up and publishing it?
A question. A particularly high TSI value appears during the early part of the ACRIM time series (~1977). This high value is not present in the PMOD time series. This high value drops away quickly at a time when, according to the GISS surface temperature data, the Earth was warming rapidly after the slight cooling/static period during the mid-century.
At first glance, it seems the two data sets correlate quite well, however the ACRIM data set appears to over accentuate the features that are present on the PMOD data set, almost as though ACRIM was produced by reconstructing PMOD signal with a positive gain factor (or vice-versa with a negative gain).
Is there anyway to account for the differences in the data sets? Is it the satellite instrumentation, post processing or something else?
Cheers
Hank Roberts // July 25, 2007 at 5:13 am
Read quickly, this reminded me to ask whether this is related. Of course I have only the few lines of the brief abstract:
“By projecting surface temperature data (1959–2004) onto the spatial structure obtained objectively from the composite mean difference between solar max and solar min years, we obtain a global warming signal of almost 0.2°K attributable to the 11-year solar cycle. The statistical significance of such a globally coherent solar response at the surface is established for the first time.”
http://www.agu.org/pubs/crossref/2007/2007GL030207.shtml
Paul Graham // July 25, 2007 at 5:29 am
It true then we don’t have enough satellite date so we must use the best proxies available.
Also I feel its necessary to point out that EUV increases by a factor of up-to 10 during any solar cycle. This alone would could explain temperatures increases along with certain weather patterns.
IE Rain falling mostly after a cycle has completed; as the atmosphere expands and contracts, changing its dew point.
Simon // July 25, 2007 at 8:24 am
The Earth’s magnetic field strength has steadily declined about 5% since 1850, is this taken into account by the theories that invoke variation in solar activity as a significant climate forcing factor?
PaulM // July 25, 2007 at 1:28 pm
Tamino, you have missed a simple but important point. The trend claimed by Lockwood and Frohlich is 0.1 W m^(-2) over 20 years. This is less than one part in 10,000! It is also much less than the accuracy of the machine, and less than the difference between PMOD and ACRIM. The only conclusion one can draw is that there is no significant trend. L&F make the absurd claim that the trend is ‘highly significant’. They deserve all the criticism they have received.
[Response: According to Frohlich (2004, Astronomy & Astrophysic Rev., 12, 273–320), "A detailed error analysis shows that the PMOD composite has a long-term uncertainty of less than about 90 ppm per decade."]
Earle Williams // July 25, 2007 at 3:01 pm
[Response: According to Frohlich (2004, Astronomy & Astrophysic Rev., 12, 273–320), “A detailed error analysis shows that the PMOD composite has a long-term uncertainty of less than about 90 ppm per decade.”]
In other words, the PMOD composite has a long-term uncertainty of less than (approximately 0.12 W/m^2) per decade. That does make it difficult to argue that the trend is statistically significant when it is on the scale of or less than the stated uncertainty.
Dr. Richard C. Willson // July 25, 2007 at 6:31 pm
Responding to several issues raised:
The trend of 0.04 %/decade based on values at succesive solar minima during cycles 21 - 23 was based on the TSI averaged over the 6 month period centered as closely as possible on the minima. The rationale for using the minima TSI to rekon long term TSI variation is that it is the ‘background’ TSI level during the relevant solar cycles. The interest in tracking these minima values is to determing the long term (many solar cycles and longer) TSI solar output. It is only this long term behavior that is of significance for climate change.
As for our declining to integrate over the cycles instead of using minima TSI to look for trending, an analogy comes to mind: your physican doesn’t take your temperature to determine general health after you’ve run the ‘4 minute mile’, but when you’re in a state of rest.
The satellite TSI database began only in late 1978. The < 29 years of results to date are woefully inadequate to make any kind of quantitative statement about their climatological significance. It is absurdly premature for anyone to speculate quantitatively on the presence or absence of TSI forcing in climate change with the current TSI database. This was my point in raising objections to the Lockwood/Frohlich conclusions and the press release that appear to pander to the Anthropogenic CO2 global warming scenario promulgated by the IPCC’s Fourth Assessment.
J Edwards // July 25, 2007 at 7:03 pm
[Quote: Because earth is a sphere, and the surface area of a sphere is one fourth its cross-sectional area, a change of 1 W/m^2 in solar irradiance causes a change of only 0.25 W/m^2 in climate forcing.]
I think you have this backwards. The cross-sectional area is 1/4 the total surface area, such that the TSI of 1 w/m^2 illuminating the cross-sectional “circle” of the Earth, yields only .25 W/m^2 over the entire spherical surface.
[Response: Oops again! Right you are.]
Dr. Richard C. Willson // July 25, 2007 at 7:52 pm
Re: the comment of nanny_govt_sucks // Jul 24th 2007 at 11:14 pm :
“…….The PMOD composite does not have the same high values in solar cycle 21. Since the only source of data for this time period appears to be Nimbus 7/ERB, it is apparent that PMOD has adjusted this data considerably, or used some other source.
What is their justification for doing so?”
—————-
The facts of this issue are:
The PMOD composite was constructed to agree with the linear regression solar proxy model of Judith Lean and took considerable liberties with the satellite TSI database to accomplish it. Frohlich and lean modified the results of the Nimbus7/ERB and ACRIM1 experiments published by their science teams to agree better with Lean’s model. In the case of ACRIM1 this was in direct contradiction with the (published) satellite performance issues and observations and without any consultation with the science team.
To construct a multi-decadal composite it is necessary to relate the ACRIM1 and ACRIM2 results across the two year gap between them. There are two choices to do this and they give quite different results. The highest quality ‘ACRIM gap’ comparative database is the Nimbus7/ERB which produces a TSI composite demonstrating significant upward trending during solar cycles 21 - 23, then a return to cycle 21 levels approaching cycle 24. The other ‘gap’ database, the ERBS/ERBE, clearly inferior to the Nimbus7/ERB in calibration, precision and sample rate, produces no significant TSI composite trend when used to bridge the ‘gap’.
The difference between Nimbus7/ERB and ERBS/ERBE results during the ‘gap’ is caused by uncorrected degradation of the ERBS/ERBE sensors. Nevertheless Frohlich and Lean chose the ERBS/ERBE connection to relate the ACRIM experiments. The resulting PMOD composite shows no significant trend and agrees better with the predictions of Lean’s proxy model than if they had used the Nimbus7/ERB comparisons. This facilitates their conclusions about solar trending and climate change but does not represent the most objective use of the extant TSI satellite database.
Lockwood & Frohlich’s paper would have had credibility if they’d compared the effects of using the PMOD and ACRIM composites and discussed the different results. Their paper would have yielded different conclusions and would have less of the flavor of a public relation effort in support of the IPCC’s anthropogenic global warming scenario.
Dr. Richard C. Willson // July 25, 2007 at 8:20 pm
Re: the comments by Chris C // Jul 25th 2007 at 3:00 am
“……Is there anyway to account for the differences in the data sets? Is it the satellite instrumentation, post processing or something else? “
The data used in the ACRIM composite database are the observations published by the science teams responsible for each experiment’s results. The data used in the PMOD composite were modified by Frohlich to provide better agreement with Judith Lean’s TSI proxy model. In particular, the peaking Nimbus7/ERB TSI results during the maximum of solar cycle 21 were decreased substantially by Frohlich based on a tenuous comparison with early results from a similar sensor on the VIRGO experiment and an opinion by Lean that the sun couldn’t generate such TSI peaks. (It should be noted the sun has produced several similar peaks during cycle maxima since.)
tamino // July 25, 2007 at 8:23 pm
Dr. Richard C. Willson said:
Although I have good reason to believe the PMOD composite is far superior to the ACRIM composite, let’s (for the sake of argument) accept the ACRIM composite as correct. Trend analysis of the residuals of the ACRIM data indicate that the trend rate over the ~28 years of data is 0.011 W/m^2/yr. The 2-sigma error of this estimate, including the effect of autocorrelation, is +/- 0.0059 W/m^2/yr. So we have enough data to make the quantitative statement that based on the ACRIM composite, the trend in TSI for the 28-year observation window is between 0.0052 and 0.017 W/m^2/yr (95% confidence interval).
Hence we can say, quantitatively, that the rate over the last 28 years is not more than 0.017 W/m^2/yr. Over 28 years that’s a total increase of not more than 0.476 W/m^2. Dividing by 4 to account for the distribution over the spherical earth, and multiplying by 0.7 to account for the earth’s albedo, we have a net increase in climate forcing of not more than 0.0833 W/m^2. At a climate sensitivity of 0.75 deg.C/(W/m^2), this would lead to a net warming not more than 0.062 deg.C. And that is the most generous estimate possible, using the ACRIM composite, the 2-sigma upper bound on the trend rate, and allowing for all the warming to be already in effect with no delay due to thermal inertia. The actual global warming over the time interval in question is, in fact, in excess of 0.5 deg.C.
Furthermore, it is patently false to claim that the warming observed in the last 28 years can be due to TSI increase prior to 1978 which was delayed due to the thermal inertia of the climate system. If that were the case, the rate of global warming over the last 28 years would be slowing, when in fact warming over the last 28 years shows acceleration, not deceleration. And of course, proxy reconstruction of TSI over the last century and longer indicate no real increase in TSI since about 1950.
Although the present database of satellite-measured TSI does not tell us everything about solar influences in the past, it is long enough and accurate enought to tell us, quantitatively, that TSI increase cannot be responsible for the global warming observed over the last three decades.
tamino // July 25, 2007 at 8:26 pm
Dr. Richard C. Willson said:
Interested readers should stay tuned, as the next post on this blog will address exactly these issues.
Dr. Richard C. Willson // July 25, 2007 at 9:56 pm
Tamino: If you have good reasons (”Although I have good reason to believe the PMOD composite is far superior to the ACRIM composite”) then do share them with us.
Oddly enough I don’t remember anyone that I can associate with your pseudonym having been involved in the details of these experiments over the years . Absent actual knowledge of these experiments and their inner workings, perhaps you can enlighten us based on your divine inspiration or gifted intuition. But then, this is the method of choice for the Anthropogenic CO2 global warming sycophants, n’est-ce pas?
[Response: As I stated in my last comment, in my next blog post I will share my reasons for having far greater confidence in the PMOD composite than the ACRIM composite. Stay tuned.]
george // July 26, 2007 at 2:20 am
Richar Willson said:
“It is absurdly premature for anyone to speculate quantitatively on the presence or absence of TSI forcing in climate change with the current TSI database”
So, why doesn’t the same statement apply for looking for trends in TSI based on minima from just a few successive cycles?
I can see from the ACRIM graph that there is an increase between the minimum at the end of solar cycle 21 (in about 1986) and end of solar cycle 22 (1997) and then a decrease between the minimum at the end of solar cycle 22 and the end of solar cycle 23 (which had not yet been reached when that graph was produced)
It looks like the minimum at the end of 23 is about even with that at the end of cycle 21. The text on the graph says the change is +0.01%, even though the minimum has not yet been reached!
So it would appear that the “trend” (if there is any) depends on which minima (of which cycles) you look at.
So, why should I take your +0.04% trend (between cycle 21 and 22) any more seriously than the subsequent negative trend (of about 0.03%) between the end of cycle 22 and the end of 23?
Also, the use of the change from the minimum from one cycle to the next as a gauge of how much TSI is changing would seem to be based on the assumption that the cycle is riding on some “baseline” (”background” as you called it above) and that when minimum is reached the sun is inactive (”resting”, if we are take your runner analogy seriously).
I don’t know a lot about the sun, but I do know from what i have read that the solar minimum is not a time of complete inactivity, as this indicates
“…not absolutely quiet,” adds Hathaway. “During solar minimum we can have occasional sunspots and solar flares.” Indeed there was at least one monster spot and one X-class solar flare (the most powerful kind) during each of the last three minima in 1976, 1986 and 1996.
If solar minimum can have monster flares and sunspots, presumably that can also change the TSI averaged 6 months to either side of the minimum, right?
If we assume for argument’s sake that the “background” has zero change from one minimum to the next, and also assume that the minimum of the first cycle had such a flare and the second did not,
wouldn’t that nonetheless affect your average — and therefore the change from one cycle to the next?
I’d think it would.
Also, why 6 months to either side? Why not a year? The time affects the result, of course and the fact that no two cycles are precisely alike also affects the result when using such a method.
I must say that Tamino’s method of trending from residuals strikes me as more robust than the method you have described.
Dr. Richard C. Willson // July 26, 2007 at 4:45 am
I’m involved here to provide insight into the TSI observational science of the past nearly three decades, George. Not to introduce you to coherent thought processes or solar physics. You’ll have to come up on the curve with these yourself.
Geoff Sherrington // July 26, 2007 at 6:07 am
On the evidence produced to date, one would have to adopt a position that these satellites measured a change in outgoing solar radiation that would cause a pretty small change in global temperature, even making lots of assumptions about lags and inertia and so on.
My interest has switched to what happens to radiation in the region between the satellite and the surface of the earth, be it land or sea. I cannot get my mind quite around this because of the inconstancy of the position, rotation, wobble etc of the earth re the sun, but in a 29 year satellite period would there be enough locations on the earth where the temperature was measured when the sun, the satellite and the centre of the earth were aligned and the temp was known where the shadow of the satellite would be at high noon? If there were, would this not provide a control for a number of other possible variables and allow better comparison of irradiance at the satellite with temp on the ground directly below? Would this be a good data set to examine a trend over 30 years of sat work? Sorry these are all questions and not answers, but at least they are not synthetic forcing estimates or arbitrary corrections to past data.
george // July 26, 2007 at 12:45 pm
I’m involved here to provide insight into the TSI observational science of the past nearly three decades, George. Not to introduce you to coherent thought processes or solar physics. You’ll have to come up on the curve with these yourself.”
In other words, you have no answer for why it is that you can claim an upward “trend” of 0.04% TSI increase based on only a couple cycles.
AS you say, I don’t know much solar physics, but I do know when someone is avoiding a direct question.
PaulM // July 26, 2007 at 1:02 pm
Tamino,
“Furthermore, it is patently false to claim that the warming observed in the last 28 years can be due to TSI increase prior to 1978 which was delayed due to the thermal inertia of the climate system. If that were the case, the rate of global warming over the last 28 years would be slowing, when in fact warming over the last 28 years shows acceleration, not deceleration.”
You are wrong here I think. There has been no acceleration in recent years. In fact, if anything, there has been a levelling off. This is discussed on numerous sceptic websites which I suppose you wont believe, but look at for example the GISS data at http://data.giss.nasa.gov/gistemp/graphs/Fig.C_lrg.gif - this shows no apparent increase over the last 7 - 10 years.
And please take more note of what Richard Willson says - he knows what he is talking about.
[Response: Regarding the trends in global temperature, you are wrong on ALL counts. See the update to this post (at the end, just before the comments).
Regarding Dr. Willson, I agree he knows a great deal about satellite measurements of TSI. But he is not infallible, and numerous equally qualified individuals disagree with him; that's why the ACRIM composite is only one of three contenders. My opinion, based on examination of the evidence, is forthcoming in an imminent post.]
guthrie // July 26, 2007 at 1:52 pm
Paul, I see the problem- you’ve used the monthly chart instead of the yearly one, which does indeed show a fair bit of warming.
ChrisC // July 26, 2007 at 2:18 pm
PaulM
The graph you have chosen to display is the monthly mean, not the more usual annual mean. There is a much larger variability in the monthly data sets than in the annual and inter-annual data set. This increases the noise in the signal somewhat, which can obscure the trend.
Check out:
http://data.giss.nasa.gov/gistemp/graphs/
for a full complement of GISS graphs, several of which make it quite clear that warming of the Earth is continuing, and does not appear to be “leveling off”.
Also, by fiting a linear line to the data in the graph you provided, I found an rough increase of 0.2 deg per decade, which is roughly in line with the IPCC estimates.
[Response: Actually, using the monthly rather than annual means makes the trend *easier* to identify mathematically, although it generally makes it less apparent to the eye. Annual averages tend to remove the very short-timescale fluctuations, making longer-term trends more visually apparent, but they also remove a (slight) bit of the information content of the time series, slightly obscuring the trend mathematically.]
Eli Rabett // July 26, 2007 at 3:15 pm
I think it is a very good thing that Dr. Willson is participating. Given that, he and we need to understand three things.
The first is for Dr. Wilson. Given the nature of the internet, tamino could be Judith Lean. People use an alias for their own reasons. When they do so, they are what they write, and Tamino has shown a pretty high standard.
The second (for both sides) is that when you try and blow people off you get crapped on.
The third (for us) is that when someone involved in the issue joins, they get a bit of extra room if only for the extra information they bring.
I think the last is a lesson that Climate Audit and some other blogs have not learned to their own cost.
[Response: Although I intend to remain anonymous, I will reveal that I am not Judith Lean, nor am I directly associated with the satellite measurements of TSI, or constructing any of the composites thereof.]
Eli Rabett // July 26, 2007 at 7:11 pm
Dr. Richard Willson wrote:
“The trend of 0.04 %/decade based on values at succesive solar minima during cycles 21 - 23 was based on the TSI averaged over the 6 month period centered as closely as possible on the minima.”
Looking at the graphs with the perspective of a number of decades fitting noisy spectra I would say that the value you get by eye is going to be very close to the value you get by taking the average over six months, so that still does not really answer the question as to why the quiet sun is most representative of TSI. After all, we know that ocean heat content which integrates over the entire period is a strong driver (and is ~proportional to sea surface temperature so let’s not start that).
As to resting and excited blood pressure, well it depends what you are interested in, you have heard about stress tests, no? (see comment on blowing off)
Dr. Richard C. Willson // July 26, 2007 at 7:21 pm
Good points Eli.
My primary reason for participating is to provide information on the inner workings of TSI satellite observations and TSI composite time series over the past ~ 29 years to those who may not have read the voluminous literature on it.
The climate implications of TSI variability are the province of others. But I am dedicated to keeping an accurate picture of TSI variability before the scientific community and public as a candidate for natural climate forcing while the research on all forcings proceeds.
I don’t promote natural climate change forcings to suppress consideration of anthropogenic ones. But I believe quantitative climate science is not mature enough to provide the level of understanding required for sensible conservation programs and the political movements required to effect them.
george // July 26, 2007 at 8:03 pm
Richard Willson above says: “It is absurdly premature for anyone to speculate quantitatively on the presence or absence of TSI forcing in climate change with the current TSI database”
That’s interesting because 10 years ago — in October, 1997, (after he had seen barely one cycle of satellite data from minimum to minimum), Willson said this:
“Solar forcing would provide only about one-fourth as much warming [as Greenhouse warming], if the solar trend [0.036% higher in 1996 than 1986, according to Willson] persists over the same period,” Willson said. “Solar forcing could be significant, but not dominant.” — from Willson: The Sun May Be Factor in Global Warmup, by Bob Nelson
Bob Nelson (who wrote the above article in Columbia university record”, Oct 1997)
“according to Willson, solar forcing—the sun’s effect on long-term climate—might account for between 0.7 and 1.4 degrees of warming over the next 100 years, if sustained at the pace his observations suggest.”
So, in other words, it appears that Willson was speculat[ing] quantitatively on the presence or absence of TSI forcing in climate change with the current TSI database” back in 1997 — with even less data than he has today.
So, back to my original question: how can one claim there is an “upward trend” in TSI of 0.04% per decade based on only a couple cycles of satellite data? (or on just a single cycle(!) minimum to minimum as Wilson appears to have done back in 1997)
One does not have to have a PhD in solar physics to understand the issue with that.
Dr. Richard C. Willson // July 26, 2007 at 10:47 pm
The points of PaulM and Earle Williams are well taken. The 90 ppm/decade uncertainty is optimistic.
Other significant problems with the VIRGO database that contribute to its uncertainty budget:
The uncorrected degradation of the ERBE sensors during the ‘ACRIM Gap’ would alone be a source of ~ +/- 100 ppm/decade.
The failure of Frohlich’s VIRGO/PMO6 instrument’s shutter mechanism at launch prevented his instrument from self-calibrating degradation of its TSI monitoring sensor of 3000 ppm during the first 6 years of the SOHO mission. Although Frohlich minimized this through comparisons with the Belgian DIARAD sensor (the only one to work properly on VIRGO) considerable residual uncertainty would accrue to the VIRGO database which comprises a large fraction of the PMOD composite.
Another problem for the VIRGO data is the ‘summer vacation’ period in 1998 when control of the SOHO spacecraft was lost. The VIRGO sensors cooled well beyond their ‘storage temperature limits’ and suffered permanent sensitivity changes. Continuity of the VIRGO database relied on cross comparisons with the ACRIM2 TSI results. This would add significant uncertainty to the VIRGO results and the PMOD composite.
Timothy Chase // July 27, 2007 at 4:08 am
guthrie wrote:
Yes, but don’t throw it out - I have little doubt Tamino will find some use for it.
Tamino - nicely done.
PS You don’t have to publish this if you don’t want to.
Anon123 // July 27, 2007 at 5:07 am
Hello tamino and readers/posters:
i’m not a climate expert, but have some quantitative background. Your blog and the discussion here have been very engaging and educational on the subject of global warming.
I found it surprising that not many seem to have explored the possibility that the two major dips in the land+ocean annual temperature anomalies profile (precipitously in the mid 40s, and around 1953, and fuzzily in the 60s and 70s) may have been influenced/caused by major wars, namely, WWII/Korean war/Vietnam war.
It seems to me that large scale wars could influence climate in many ways: WARMING (CO2+heat released in various ways), COOLING (Sulphur/sulphates from gun powder, particulates from rising dust, cement production and other factors from reconstruction after the war).
Does the impact on climate from Hiroshima/Nagasaki (perhaps a mini “nuclear winter”) explain part of the mid 40s dip?
Thank you.
John Willit // July 27, 2007 at 12:33 pm
I don’t know why you would use satellite measurements for the Sun’s irradiance but you won’t use satellite measurments for the temperature of the earth.
If you are going to use GISS for the global temperature trend, of course it is going to be increasing. That is what GISS does to the temperature record.
The RSS temperature trendline, however, is only 0.001C per decade over the past 10 years (July 1997 to June 1997.)
In fact, the June 1980 RSS global average temperature in the troposphere (0.130C above average) is almost exactly the same number as June 2007 figure (0.139C above average.)
Ie. no global warming over the past 10 years and none over the past 27 years.
Sure it went up, predominantly during the record 1997-98 El Nino, and the record solar cycle 22 ending in May 1996, but now it is down.
[Response: You should run the numbers before making such claims.
Using the RSS MSU TLT (lower troposphere temperature) data series, the trend from 2000 to the present is 1.9 +/- 1.2 deg.C/century. Using RSS MSU TLT from 1997 to the present, the trend is 0.85 +/- 1 deg.C/century. Using RSS MSU TLT from 1978 to the present, the trend is 1.8 +/- 0.2 deg.C/century. All these results are in excellent agreement with the GISS surface temperature data.
Interested readers should be aware that the MSU TLT data are *not* observations; they are combinations of observations from various layers of the atmosphere, in an attempt to reconstruct lower troposphere temperature (in effect, to remove the impact of the stratosphere from the troposphere temperature changes). As such, they are inherently less reliable than direct measurements of surfact temperature. Readers should also be aware that MSU TLT are not surface temperature data; they're for the lower troposphere.
Your snarky comment about the motives of NASA GISS, together with your failure to know what the MSU data trends really are, say a great deal about your attitude, but nothing at all about the reliability of GISS data.]
guthrie // July 27, 2007 at 1:35 pm
John, satellites are used to take the Earths temperature.
Heres a random link citing 12 year old information:
http://www.nsc.org/EHC/climate/ccucla8.htm
More up to date we have wikipedia, where the RSS data shows a definite warming trend:
http://en.wikipedia.org/wiki/Satellite_temperature_measurements
It seems to me that you are cherry picking months to make your case better. However mathematical analysis shows a definite trend.
John Willit // July 27, 2007 at 2:22 pm
The chart you linked to is 2 years old.
Here are the up-to-date charts from RSS
http://www.remss.com/msu/msu_data_description.html#msu_amsu_time_series
Here are the figures for the lower troposphere.
http://www.remss.com/pub/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Land_and_Ocean_v03_0.txt
Dr. Richard C. Willson // July 27, 2007 at 4:44 pm
Tamino: re your commentary on John Willit’s input:
The surface temperature measurements are aflicted with many problems including urban heat islands, poor calibration, outdated measurement networks, ad infinitem….
What this blog could do well without are your immature personal attacks on some posters. They really detract from the quality ideas expressed here.
[Response: When I stated that John Willit should run the numbers before making claims about the trend in the satellite record, it was a simple statement of fact -- and a correct one.
Referring to his comment about NASA GISS
as snarky, and as more a reflection of his attitude than the reliability of GISS data, I stand by that appraisal. I invite readers to judge for themselves whether they consider my evaluation correct.
It's puzzling that you defend him, despite his clear implication that GISS was perpetrating a deliberate fraud, but you object to my response as an "immature personal attack." Again, readers are invited to draw their own conclusions.]
John Willit // July 28, 2007 at 12:45 am
Well, I don’t know. Tamino starts with a complete “data selection” paper and a data selection thread and I just quote other data selected because it shows something else and I get called snarky.
[Response: You were called "snarky" because you directly and clearly implied that the GISS data were a deliberate fraud. This is a blatant insult to the scientists at NASA GISS, and under the circumstances "snarky" is a rather mild reproach.]
Even the latest IPCC report did not use the GISS data. They used the Hadley Centre’s data instead.
But in terms of solar irradiance and the average global temperature, there are other measures and other data sources available.
If the lower troposphere satellite data shows no trend over the past 10 years and no increase in the past 27 years, shouldn’t that be presented and thus known to general public as well.
[Response: The lower troposphere satellite data *do* show a *sizeable* increase in the past 27 years, which is comparable to the increase shown in the NASA GISS surface temperature data.]
If solar cycle 22 was one of the shortest ever and thus, represents of the most active periods of solar activity, shouldn’t that be presented and thus known to the public as well.
[Response: What is your evidence that the length of the solar cycle represents the level of solar activity? Isn't that better estimated by measuring solar activity -- a.k.a. TSI?]
I don’t know. I like to see all the data.
Alan Woods // July 28, 2007 at 4:41 am
Tamino, does the positive trend in your linear fit post 2000 T’s agree with that predicted or calculated for the same period by your quadratic fit of 1998-2007 T’s?
Also, can you duplicate this result with HadCRUT3 data?
And, what is the justification for a quadratic fit of the data rather than a linear trend. The reason I ask is that it seems you can choose many ways to slice and dice and give statistical significance, yet get different results. Is there a rule that directs us to the correct method for analysis?
[Response: Basically: yes, yes, and no. See UPDATE #2 at the end of the post (just before the comments).]
John Willit // July 28, 2007 at 1:25 pm
Beyond TSI, you have solar wind and solar storms, geomagnetic effects, the fact that the UV spectrum of the TSI varies more than the TSI, C-14 production and Be10 production varies more than TSI. The Maunder Minimum and variations in solar cyle length all indicate that more is going on with the Sun’s impact on the earth’s climate than the TSI measure indicates.
guthrie // July 28, 2007 at 7:21 pm
Sure John, but then you have to posit a mechanism, and show how they all tie together. Nobody seems to be able to do so in a properly scientific fashion.
george // July 28, 2007 at 9:53 pm
Alan Woods says “it seems you can choose many ways to slice and dice and give statistical significance, yet get different results. Is there a rule that directs us to the correct method for analysis?”
I’d say that caveat applies pretty well to the basic claim of an “increase of 0.04% / decade in TSI.”
If analyzing the data in different ways (eg, using different satellite data and piecing it together in various way to bridge the “gap”) gives different answers, “watch out!” is a pretty good rule of thumb when it comes to making extraordinary claims (in this case, about solar output).
John Cook // July 30, 2007 at 9:19 am
Following on from John Willit’s comment re the Maunder Minimum, it seems to me this whole debate about PMOD vs ACRIM is distracting from the main point about the sun’s role in climate change. The most instructive paper I’ve read on solar influence is Usoskin et al 2005. Basically, they find a close correlation between solar activity and global temperatures over the past 1150 years, going a long way in explaining the Maunder Minimum and Medieval Warm Period. But the crucial part of the study is the correlation ends in 1975 when temperatures started rising. Sun and climate have been in lockstep through most of the last millenium (barring the odd volcanic influence) and then drastically diverged 30 years ago. The fact that we’re debating whether the sun is showing a very slight upward trend or a very slight downward trend - in fact, if there’s a trend at all - underscores the stark divergence between steeply rising temperatures and solar activity.
Gunnar // July 31, 2007 at 8:37 pm
Tamino,
I’m dealing with the premises behind all your arguments, which are completely wrong.
>> Because earth is a sphere, and the surface area of a sphere is one fourth its cross-sectional area, a change of 1 W/m^2 in solar irradiance causes a change of only 0.25 W/m^2 in climate forcing.
I don’t see any basis for this. The earth is one entity. The sun is heating it all the time. A change of 1 W/m2 is a change of 1 W/m2.
>> A good estimate of the effectiveness of climate forcing is that every 1 W/m^2 of climate forcing produces, eventually, 0.75 deg.C change in global average temperature. The 0.25 W/m^2 or more change in climate forcing throughout the solar cycle would therefore change global temperature by roughly 0.1875 deg.C,
This maybe some kind of statistical hand waving, but that is not an ideal approach. This effect can be better isolated by looking at the summer vs winter temperature.
Summer SI - Winter SI = ~200 W/m2
Summer Temp - Winter Temp = 33 - 7 = 26
dT per dSI = 26 / 200 = .13 T/W/m2
@ dSI = 4 W/m2 dT/dSI = .52
Yes, the data clearly indicates TSI max - min =~ 4 W/m2.
>> if it persists long enough for its effect to be complete. But due to the immense thermal inertia of the oceans, it takes time for the full impact of climate forcing to take effect.
You have a fundamental misunderstanding of physical systems, and you are guilty of an argument switch here. The question is, does the atmosphere, not ocean, have a large thermal inertia? Based on observation of night to day temperature rise, there is no large atmospheric thermal inertia. As such, a change of 4 W/m2 does certainly affect the atmospheric temperatures. Looking at the satellite data, it basically shows that the temperatures stays mostly in the range +/- .5 deg C. This is completely expected normal variation, bases solely on TSI variation.
>> The influence on TSI of the solar cycle goes both up and down, so unless there is a persistent trend, over the long haul the ups and downs cancel with net effect zero.
Absolutely completely wrong. Physical systems like the earth are not doing computer smoothing. Every change in input is absorbed.
>> In fact, the 11-year solar cycle is barely, if at all, discernable in earth’s global temperature signal.
Absolutely wrong. I have plotted the satellite temperature data along with solar activity, and it’s quite discernible.
>> But the crucial part of the study is the correlation ends in 1975 when temperatures started rising. Sun and climate have been in lockstep through most of the last millenium (barring the odd volcanic influence) and then drastically diverged 30 years ago
John Cook, you must be looking at highly massaged data. The satellite data shows no drastic divergence starting in 1975. It’s all clearly solar activity, enso events, volcanoes, inertia. In reality, there was a peak in 1998 at almost .8, which has cooled to .22 since then. If you’re looking at data that doesn’t show a peak in 1998, and cooling since then, you are being deceived.
tamino // August 1, 2007 at 3:49 am
Re: Gunnar
Whew! Where do I begin? How about at the beginning?
This comment alone reveals that you haven’t even scratched the surface of understanding the climate system, or the forces that drive it.
Climate forcing is defined as the average incoming energy per square meter of earth’s surface; it’s usually expressed as watts per square meter (W/m^2). Solar irradiance is defined as the incoming solar energy in W/m^2, for an area which is above the atmosphere and directly facing the sun.
The average solar irradiance is about 1366 W/m^2. Some of this is intercepted by earth; the amount intercepted is proportional to the cross-sectional area of the earth, which is pi x r^2, approximately 1.275 x 10^14 m^2. So, earth intercepts solar energy at a rate of about 1.74 x 10^17 watts.
This energy is responsible for climate over the entire surface of the earth. Earth’s surface area is 4 x pi x r^2, about 5.1 x 10^14 m^2. Hence the average input to the climate system from solar energy is 341.5 W/m^2.
This is about as basic as it gets. And you don’t even have to take my word for it. You don’t even have to go to a reliable source of climate information like RealClimate. Go to a denialist website like climateaudit or Roger Pielke Sr.’s blog, and ask them. Even denialists will tell you that you’re as wrong about this as is possible.
And that’s just your first point. In fact every aspect of your comment is as wrong as this one. You don’t need a refutation; you desperately need an education.
Alan Woods // August 1, 2007 at 4:07 am
Tamino, considering you take offence whenever you imagine that somoene is questioning the motives of GISS or whoever, I’m literally stunned that you use highly emotive language like ‘denialist’ to describe Roger Pielke Sr.’s blog. Let’s not beat about the bush - the terms ‘denialist’ and ‘denier’ are used to imply that those who are skeptical about anthropogenic climate change or the extent of it are no better than those who deny the holocaust. Its a disgraceful tactic and, given your obvious intellect, one that I’m amazed you indulge in.
[Response: OK let's not beat around the bush. For the record: I do not put climate denialists in the same despicable league with holocaust deniers.
The terms "denialist" and "denier" are meant to denote those who deny the truth about global warming, even *after* they've examined the evidence. The suggestion that it's meant to imply that they are in the same despicable league with holocaust deniers, is a very clever propoganda trick meant to discredit the motives of those who affirm the truth about global warming.]
Alan Woods // August 1, 2007 at 4:33 am
So there is only one ‘true’ interpretation of climate data. What kind of scientific method is that?
Gunnar // August 1, 2007 at 3:33 pm
>> Climate forcing is defined as the average incoming energy per square meter of earth’s surface
That is useful for calculations of how much land and sea are being directly heated by the sun. BUT, we are talking about the atmosphere, which is subjected to SI directly.
This is all made irrelevant, since by calculating dt/dSI, I cut through all that.
However, if you wanted to calculate power delivered to earth, you would still be wrong. The average SI is 1366 W/m2, which already implies the amount intercepted by earth, and has area in it. Over one day, like a pig on a spit, the entire earth is subjected to this level. The whole earth then is receiving 1366 W/m2 * 5.1 x 10^14 m^2 = 6.97 x 10^17 Watts.
>> This is about as basic as it gets. … you desperately need an education.
It is basic, and you seem to be just regurgitating descriptive text from RealClimate, instead of actually thinking about it. I do have an education, a BSEE, which gives me a solid understanding of math and science. I can only surmise that you are confused when you take the amount of SI that reaches the land/sea, and incorrectly assume that to be the amount heating the atmosphere.
[Response: Not only are you completely mistaken, you're obstinate in your error. When it comes to climate science, you *do* need an education -- desperately. Again, I urge you to go to climateaudit or Roger Pielke's blog -- sites which cannot possibly be considered advocates of AGW -- and ask them. They'll set you straight. With any luck, you'll listen.]
dhogaza // August 1, 2007 at 4:20 pm
Think about what you’re saying very, very deeply for a moment.
Taken literally, you’re saying that we should accept flat-earth interpretations of the physical evidence of the shape of the earth seriously.
On and on …
Sometimes the scientific consensus is right, you know? Sometimes there really IS only one reasonable interpretation of the data, and when evidence is overwhelming, there’s nothing “anti-scientific method” about recognizing the fact.
Pete DeSanto // August 1, 2007 at 7:30 pm
Gunnar - so the pig on the spit cooks equally on all sides whether you rotate it or not?
Gunnar // August 1, 2007 at 8:07 pm
>> Gunnar - so the pig on the spit cooks equally on all sides whether you rotate it or not?
It’s rotating, therefore, all sides of the pig are exposed to the heat source. It’s totally incorrect to consider the heat source as 1/4 of it’s actual strength.
Analogy:
Wife: Honey, the roast is charred black, did you cook it at 350 for one hour?
AGW husband: Why, yes, the temperature did AVERAGE 350 for one hour. For 10 minutes, it was at 1500, so I compensated by setting it to 120 for 50 minutes. That’s the same, right?
>> Not only are you completely mistaken, you’re obstinate in your error.
If it’s so basic, why can’t you explain why we should reduce the sun strength to 1/4 of it’s value? If you understand why, you should be able to explain it, unless you are just regurgitating descriptive text without thinking.
[Response: I have explained it. More than once. Everybody else gets it.]
Gunnar // August 1, 2007 at 8:15 pm
>> Sometimes the scientific consensus is right, you know?
1) Consensus means nothing in the scientific method.
2) Argument by authority is a logic fallacy
3) Besides, there is no consensus. It has now been shown that Oreskes & Science magazine’s claim of “consensus” was fraudulent. Judging from papers written since 2004, only 7% explicitly endorse AGW, and fewer than half implicity/explicitly endorse AGW. The trend is definitely away from AGW.
>> Sometimes there really IS only one reasonable interpretation of the data, and when evidence is overwhelming,
But if you’re talking about temperature data, it clearly shows a max in 1998, with .6 deg of cooling since then.
>> there’s nothing “anti-scientific method” about recognizing the fact.
If it’s not following the scientific method, it’s anti-scientific
Petro // August 2, 2007 at 5:21 am
Now we got this Gunnar character polluting the discussion with his inane grasp of geometry. What is due next from the denialist camp? 2+2 is no more four, mathematics is optional too?
Barton Paul Levenson // August 2, 2007 at 11:58 am
[[Because earth is a sphere, and the surface area of a sphere is one fourth its cross-sectional area, a change of 1 W/m^2 in solar irradiance causes a change of only 0.25 W/m^2 in climate forcing.]]
Less than that. Due to Earth’s 30% or so albedo, the Earth system only absorbs 70% of the sunlight that falls on it. You get a 1.75 W m^-2 forcing and a 1.31° K. change in temperature (if the trend is persistent — and as you point out, the 11- and22-year cycles don’t have a net trend).
Gunnar // August 2, 2007 at 12:01 pm
>> Response: I have explained it. More than once. Everybody else gets it.]
Just repeating the same words is not an explanation. Why don’t you explain WHY? If you don’t, it will mean you really don’t know. This “Everybody else gets it” is quite childish.
I could be quite wrong about this, so please explain why we should reduce the solar strength to 1/4 of it’s actual value.
Pete DeSanto // August 2, 2007 at 2:05 pm
Gunnar - if you perform a proper radiative energy balance on the Earth, you can see where the factor of 1/4 comes in. Incoming energy from solar radiation = (1-a)*TSI*pi*r^2, energy radiated by Earth = 4*pi*r^2*sigma*T^4. TSI is only averaged over the projected area (a circular disk) of the spherical object upon which it is incident, whereas the radiated energy is from the entire surface of the spherical object. This is equivalent to saying that the average TSI incident upon the projected disk is distributed over the entire surface area of the Earth and must then be reduced by a factor of 1/4. This is really textbook stuff. Now if you rearrange the energy balance equation to solve for T and take dT/dTSI and plug in the numbers (using TSI = 1367 W/m^2), you get .047 K/W/m^2. For a change in TSI of 4 W/m^2, you would get a temperature change of 0.188 K . No statistical handwaving, just plain old physics and math. BTW - if you don’t do the energy balance properly and neglect the factor of 1/4, then you get an average T = 361 K, which is 88degC and would leave the Earth uninhabitable for us! I think Tamino has explained it clearly, you just haven’t understood it.
Pete DeSanto // August 2, 2007 at 5:01 pm
All of my calculations neglect feeback effects, I used solar forcing (not climate forcing) alone. My bad. It’s not as simple as dT/dTSI for overall climate sensitivity!
Gunnar // August 2, 2007 at 7:44 pm
>> TSI is only averaged over the projected area (a circular disk) of the spherical object upon which it is incident, whereas the radiated energy is from the entire surface of the spherical object. This is equivalent to saying that the average TSI incident upon the projected disk is distributed over the entire surface area of the Earth and must then be reduced by a factor of 1/4.
Yes, but this is incorrect, since it pretends that the earth is not rotating. Over the course of the day, the entire earth is subjected directly to the sun. Just like with the pig on the spit, the whole pig is cooked. With physical systems, it is incorrect to average daily, yearly or 11 year cycles and consider that to be the input to the system. Read the charred roast analogy again. My daily exposure to the sun, averaged over the whole day is very low, and yet I will get a sunburn. Real systems react to current input value.
However, as I said, this is all made irrelevant, since by calculating dt/dSI, I cut through all that. I’ll repeat this reality check calculation:
Summer SI - Winter SI = ~200 W/m2
Summer Temp - Winter Temp = 33 - 7 = 26
dT per dSI = 26 / 200 = .13 T/W/m2
@ dSI = 4 W/m2 dT/dSI = .52
There is no doubt that the change in TSI is reponsible for the change in summer/winter temperature.
Therefore, a 4 W/m2 does result in a delta T of .52 deg C. And this is what the data shows. Hence, solar activity explains just about all the variation we’ve seen.
>> if you don’t do the energy balance properly and neglect the factor of 1/4
No, you’re mixing up what we’re doing here. By definition, when I plug in 200 W/m2, I get 26 deg C of DELTA temperature.
Gunnar // August 2, 2007 at 7:55 pm
>> you get .047 K/W/m^2. For a change in TSI of 4 W/m^2, you would get a temperature change of 0.188 K .
If that were true, the measured TSI delta between summer and winter of 200 W/m2 would only result in delta T of 9.4 degrees. You’re wrong, since it’s 26.
tamino // August 2, 2007 at 8:08 pm
Gunnar,
I’ll make one more try.
Imagine two squares, one meter on each side. Square “A” is always kept pointing at the sun. It therefore intercepts solar energy at a rate of about 1366 W/m^2.
Square “B” is fixed to the surface of the rotating earth. At noon let’s say the sun is directly overhead. Then at noon it’s getting energy at a rate of 1366 W/m^2. But at midnight (in fact, from sunset to sunrise), it receives energy at a rate of ZERO.
Square “A” gets 1366 W/m^2, 24 hours a day. Square “B” never gets more than 1366, but half the time it gets zero.
Which square is getting more solar energy? Which will have a higher average temperature?
Gunnar // August 2, 2007 at 9:07 pm
>> Square “A” gets 1366 W/m^2, 24 hours a day. Square “B” never gets more than 1366, but half the time it gets zero.
You’re not getting either of my two points. 1) by calculating this dt/dSI, I’m cutting through all that. I’m not trying to determine the Joules of energy delivered to the earth.
2) With regard to the Energy delivered problem, you are still incorrect. The earth is one entity, which is exposed to the sun all the time, like a pig on a spit. You’re obviously not getting that analogy, so let me try another:
Imagine a circular pool. The pool is being filled up with a hose at a certain fixed position. Now imagine that the pool is rotating, so that the hose ends up filling up the pool from all sides. No matter which side the pool is being filled from, water is being added.
[Response: I give up. You're too stubborn to get it; just don't expect anybody else to share your delusion.
Perhaps some day you'll realize how wrong you are, and it'll dawn on you that you're equally wrong about just about everything you've said.]
Gunnar // August 2, 2007 at 10:09 pm
>> I give up. You’re too stubborn to get it; just don’t expect anybody else to share your delusion.
So, the circular pool analogy did it. You obviously realize that I’m right.
[Response: You are wrong. Completely wrong. About everything.
Not only in the history of this blog, but as far as I can tell in the history of blogs everywhere, you hold the record for making yourself look like a fool.]
Alan Woods // August 2, 2007 at 11:56 pm
dhogaza, you wrote:
“Taken literally, you’re saying that we should accept flat-earth interpretations of the physical evidence of the shape of the earth seriously.”
If you can show me some scientific data that supports the flat-earth theory, then I’ll take your suggestion to heart.
There is a lot of climate data (and I’m not suggesting most) that either contradicts or doesn’t positively support the hypothesis that anthropegenic CO2 is having a major effect on world climate.
Now I’m not suggesting for a moment that these data disprove the hypothesis, but Tamino is saying that we should only interpret such data in a way that affirms the ‘truth’ of greenhouse warming. To do otherwise is to deny the truth.
To me, that is a violation of the scientific method.
Hank Roberts // April 20, 2008 at 3:04 pm
Well, this was good for a while, with involvement of Dr. Wilson on the TSI record. It got derailed completely by the flat-earth stuff. Any more real work to do, with a better filtering policy now?
Leave a Comment