The Sun Blog

We’ve got some interesting new data regarding the Sun’s brightness variations in a paper just accepted for publication in The Astronomical Journal. I’ll discuss a few of our conclusions in this post, starting with some background.

Back in the 1980s, my colleagues Wes Lockwood, Richard Radick, and Brian Skiff set out to determine how stars’ brightnesses varied during the their activity cycles. At that time, it was becoming known from space-based observations that the Sun’s brightness changed as its activity did, so it was natural to wonder if the brightness of the stars did as well. So, for over a decade Wes, Rich, and Brian patiently observed the brightnesses of a set of 34 stars also on the Mount Wilson Observatory stellar cycles program, which was continuing at that time under the direction of Sallie Baliunas. They eventually published a beautiful paper (Radick, Lockwood, Skiff, & Baliunas 1998) titled Patterns of Variation Among Sun-Like Stars, which included the following conclusions.

First, they found that the less active (i.e., Sun-like) stars varied as the Sun does, with brightnesses increasing as activity increased. More active stars, however, varied in the opposite way: they got dimmer as they got more active. This can be seen at left, which is from Figure 8 of their paper. This shows mean activity level on the y-axis versus color (a measure of temperature) on the x-axis. Open circles show stars with direct (Sun-like) activity-brightness correlations, while filled circles show inverse correlations. While there are a few stars that “break the rule,” in general, active stars show inverse correlations and inactive stars direct correlations.

Second, they found that for stars of its activity level, the Sun’s brightness variations were small. The figure at left (Figure 7 from their paper) shows this. The activity level is plotted along the x-axis, with more active stars at right. The amplitude of the brightness variations is shown on the y-axis. The Sun lies well below the regression line. A follow-on study (Lockwood et al. 2007) with several more years of observations confirmed this result.

This has led to a lot of wondering whether the present 0.1% variation of the Sun over its cycle is typical of Sun-like stars, or if the Sun is somehow “special” (e.g., Gustafsson 2008 and Böhm-Vitense 2007). There have been well-acknowledged questions about the Radick et al. sample, principal among them being that the Mount Wilson stars were not selected to be proxies of the Sun. When Olin Wilson started the program in 1966, he was principally interested in detecting the occurrence of activity cycles in a broad sample of cool stars, so he made no effort to target the so-called solar analogs (indeed, the term did not then even exist). So there has long been a suspicion that the Sun’s apparent quiescence may just be a selection effect.

We address this problem, among others, in our new paper. We observe the stars in a manner very similar to Wilson and Baliunas, but we have narrowed our targets to the bright stars most similar to the Sun. At left is Figure 11 from our paper, directly comparable to the figure just above. We retained a few very active and variable stars (upper right) as controls, to ensure we were getting the same answer for the same stars as Lockwood et al. 2007 (we did). But look at lower left. The Sun is the square, and all the nearest solar twins (including the well-known twin 18 Scorpii) are distributed around it. Among this sample of the closest Sun-like stars we can find, solar total brightness variations are not unusually low.

What does this have to do with the price of tea? Well, there’s fodder there for a number of posts, but I’ll point out two things that jump to mind.

First, if one is trying to understand a natural phenomenon, it helps to know whether that phenomenon is in an abnormal state. For things that evolve rapidly, that can be straightforward to determine: just wait a bit until the range of properties of the system becomes apparent. But for stars, which vary on timescales of decades, centuries, millennia, or more, our modern window of observations is distressingly myopic. The stars provide a much faster way to get a snapshot of the ensemble properties of brightness variations, and allow us to conclude for now that we can approach modern total solar variability as representative of its typical behavior.

Second, one of the other results of Radick et al. that we do confirm is that the Sun (and 18 Scorpii, by the way) have pretty vigorous activity cycles, at least as manifested by the Ca II H&K activity proxy. Their chromospheric cycles are near the top of the distribution for the most Sun-like stars, and even for stars of similar temperature in the broader Mount Wilson sample. Thus, while the Sun’s total brightness variations appear typical of its nearest stellar analogs, its activity cycle is fairly pronounced. This idea appears in recent other papers as well that suggest the Sun has been recently in a grand maximum of activity, from which it may soon exit.

We therefore come full circle to what I’ve blogged about in earlier posts: to whatever extent solar activity affects climate, we are likely to find the smoking gun not in its apparently typical overall brightness variations, but in the spectral distribution of those variations. This suggests some important observational programs for the future, and I hope to have some good discussion about that at our upcoming Solar Analogs workshop here at Lowell in September. As always, stay tuned!

Whew, one could spend nearly full time these days keeping track of all the efforts to estimate the strength of the upcoming solar cycle. There’s a new paper out by Brajsa et al., who find the maximum sunspot number for the upcoming cycle should be about 90. This joins the growing set of results arguing in favor of a weak upcoming cycle, and is based on application of independent empirical prediction methods.

(As an aside: there are two main ways to try to deduce future solar activity: the “empirical” methods, based on principally on analysis of the statistical properties of (principally) the sunspot record, and the “dynamo” methods, which attempt to predict activity by modeling the physical processes driving the cycle. There appear to be severe problems with the latter; more on that in another post.)

The prediction itself aside, there are two other interesting points to note.

The first is that the authors predict Cycle 24 to conclude in 2017. Since there is some suggestion that we may have reached minimum in December 2008 (see last post), this would make Cycle 24 possibly as little as 8 years long, among the shortest cycles on record. However, this statement is probably dated by now due to (a) the lengthy minimum and (b) the several months that elapsed between receipt and publication of the manuscript. Cycle 24 as shown in the paper appears to start rising in 2008, which is not clearly not the case. But it does mean that the apparent length of the predicted cycle 24 in the paper is closer to 10 years, not 8, which would put the next minimum closer to 2019. (I checked on this timing slippage idea with one of the co-authors [Svalgaard] and he confirms this reading of the result.)

Second is this quotation: “On the other hand, we should mention the possibility that the commonly used sunspot number series for the last 165 years is not calibrated correctly (Cliver & Svalgaard 2007; Svalgaard & Cliver 2007). In the case that this will prove true, the secular increase of the solar activity in the 20th century would be called into question and our prediction for the next solar maximum would be affected too..We plan to include the analysis of the corrected sunspot number series in our future research.” This is based on the referenced 2007 work (e.g., this AGU meeting abstract) on discrepancies between the Wolf and Hoyt/Schatten “group” sunspot numbers. Interesting implications there; I look forward to seeing this in print.

Some articles have appeared recently in the news about this statement from the Space Weather Prediction Center. This is the latest NOAA / NASA / ISES statement, and it’s pretty straightforward: most of the panel now agrees on a fairly weak cycle 24, with a maximum sunspot number (SSN) of 90 in May, 2013.

The background here goes back several years, and I have a longer blog article in prep about all that. Suffice to say for now that the 2007 panel was unable to reach a consensus, so they published two, um, consensi: a strong cycle case with a maximum SSN of 140 in October 2011 and a weak cycle case with SSN of 90 in August 2012. However, the protracted solar minimum since then favors the weak cycle case, and opinions seem to have converged somewhat in that direction.

The press release also states that solar minimum occurred in December 2008. True? Perhaps. In our solar activity observations, the mean HK index for the end of 2008 was 190 ± 0.81 mÅ, while since April 1, 2009 it is 192 ± 1.6 mÅ. (The HK index is a measure of the Sun’s activity level obtained from near-UV spectra of two absorption lines of calcium that closely track the solar cycle.) Cycle 24 might be starting to rise, but at least in the HK measure it is not yet significant. The trend, if real, should be clearer in ~3 months or so.

The SORCE (Solar Radiation and Climate Experiment) folks have just published a very interesting paper with some surprising results about the solar irradiance over the end of Cycle 23.

As is well known, the TSI (Total Solar Irradiance) varies directly with the Sun’s activity level, with an amplitude of about 0.1%. The Sun is about 0.1% brighter at activity maximum than at minimum. SORCE carries an instrument called TIM (Total Irradiance Monitor) that measures just this, but it also includes another intrument called SIM, the Spectral Irradiance Monitor. This instrument measures solar variability in six different wavelength bands, and SIM has turned up something very interesting.

At left is part of Figure 1 from the Harder et al. paper. This shows two of the six wavelength regions the SIM observes, from 2004 to 2008. During this time, solar activity has declined to its current minimum between cycles 23 and 24. The key observational point is that while the UV irradiance has decreased (purple line, 201-300 nm), there has been an increase in the visible (green line, 400-691 nm). Other increases in the infrared are offset by declines in the red and near IR (691-972 nm) and near UV (300-400 nm).

Harder et al. make two basic points.

First, all the various SSI trends, added together, must reproduce the observed TSI. Since the trends in the visible and IR are out of phase with the solar cycle, the trends that are in phase, in particular the UV, are larger than has been previously assumed in models of solar variability. In fact, they find that UV variability between 200 and 400 nm is almost a factor of 10 larger than was estimated from earlier satellite data.

Second, they note that most general circulation models and climate models assume that the Sun’s spectral variability tracks its total variability in all wavelengths. But as the figure above clearly shows, this is not the case. The interaction and effect of varying UV radiation in the stratosphere is well documented, even if the coupling of these effects to the troposphere is poorly understood. As I’ve noted in some other posts, although the globally averaged contribution of solar variations to climate change appears to be small, regional changes due to solar variability can be much larger (e.g., Europe during the Maunder Minimum). A nice 2008 paper by Judith Lean and David Rind makes the same point. Harder et al. conclude: “It is critical that these new solar spectral variability results from SIM be included in climate models to understand the potential direct and indirect consequences to climate change.”

I wonder if Sun-like stars vary similarly in the UV and IR as their cycles wax and wane? I just might have to try to find out…

The solar cycle prediction industry continues to roll along with a new paper arguing that Cycle 24 will reach a maximum sunspot number of 87 in December 2012.

By cycle standards this is pretty weak, and suggests a maximum slightly delayed from previous estimates (which range anywhere from now (!) to August 2012). A number of observations, as well as the statistical behavior of past solar cycles, do support the weak cycle 24 hypothesis, so I think there’s a reasonable chance this paper will turn out to be correct. Whether that’s because of the validity of their method (which is purely numerical and not based on the underlying physics) or because of coincidence is another matter. Discussion at the conference I attended recently provided some good reasons to view solar cycle predictions based on dynamo models with some skepticism and those based on numerical methods with a lot of skepticism, due to an inherent degree of randomness in the phenomenon. As Solanki pointed out, the recent consensus predictions would yield a sunspot number somewhere between 17 and 40 right now; in fact, the average sunspot number so far for 2009 is 1.

We covered quite a bit of ground in the second day of the Space Climate Symposium here in Saariselkä, Finland. I’ll focus this report on two topics: the evolution of solar irradiance over the past 400 years and the occurrence of grand maxima and minima of solar activity.

How much has the Sun brightened since the end of the Maunder Minimum (MM) in 1715? Many irradiance reconstructions have been developed, and Jürg Beer presented a new effort based on Greenland ice core analysis and a comparison of beryllium-10 and C14 production rates over the past ten millennia. He and his colleagues find evidence for numerous grand maxima and minima of solar activity in that time, and their overall reconstruction agrees well with the recent work of Krivova et al. (2007) — and both of these agree with the general recent thinking that solar irradiance has increased only modestly since the MM. Our stellar data support this viewpoint.

This provided an opportunity for a couple of speakers to discuss the physical nature of the MM, and I’d like to remind everyone of one major point. It is quite clear from the isotope records that the 11-year solar cycle did not stop during the MM. Although the MM is frequently described as a “period of exceptionally low activity” or some such, the correct observational description is that it was a 70-year period when almost no sunspots were visible. Sunspots are manifestations of solar activity: one must have magnetic activity to have sunspots, but the reverse is not necessarily the case. One cannot infer that the Sun was magnetically dead and unusually faint during the MM from the available data.

What about the next Maunder Minimum — which certainly will happen someday? Some interesting results presented by Ilya Usoskin showed that grand minima like the MM, as well as grand maxima like the period 1920-present, tend to occur in clusters. In the past 12,000 years, he and his collaborators identify 27 grand minima and 19 grand maxima. The grand minima seem to come in groups, with individual minima separated by about a few centuries (e.g., the most recent Spörer, Maunder, and Dalton Minima). No periodicity of the maxima is apparent. The Sun appears to spend about 1/6 of the time in a grand minimum state, and 1/10 in a grand maximum. What causes such states is unknown, and does not appear to happen in any predictable way. As noted in yesterday’s post, the solar cycle is at some level unpredictable, so this is not surprising.

Results by Jose Abreu (whose paper I have discussed elsewhere) suggest that the present grand maximum is going to end soon, within about 15 years. Whether the Sun will go into a grand minimum or merely a period of weaker cycles is not known. However, the extended nature of the present solar minimum points toward a weak Cycle 24, and there is at least one prediction of a very weak Cycle 25. Thus, significant evidence points toward a period of weaker solar activity or possibly an upcoming grand minimum. Could the Sun fool us? Of course, but we have to go with what the preponderance of the data says, and for now, weak activity is it.

What are the implications of different levels of solar activity for Earth’s atmosphere and climate? That’s coming up in the next set of talks.

I arrived in Saariselkä, Finland Wednesday evening.   It is about 150 miles north of the arctic circle, a stark and beautiful land of conifers and rolling fells.  The snow is obviously not going anywhere soon, and the Sun traces a long, graceful arc around the southern sky.  It is also 9 hours ahead of Arizona time and +6 from EDT, so my blog post timing will be a little weird.

Despite having the first talk in the morning yesterday after 29 hours of travel, I managed to avoid drooling or other horrors and was reasonably coherent.  (I think.)  I’m one of the few stellar astronomers here; most of the attendees are solar folks.  I do think they enjoyed hearing about the stellar take on solar variations.

Yesterday’s sessions covered a number of interesting topics.  Here are a few highlights from my notes.

We had a couple of talks on predicting the strength and timing of future solar cycles.   I’ve blogged about this and have written a summary article on this web site.  The opinions at the conference are in general agreement that long-term forecasting of solar cycle strengths is impossible, for much the same reason that long-term weather forecasting (as distinct from climate forecasting) is impossible: random perturbations to the magnetic dynamo that drives the solar cycle lead to large and unpredictable changes in the future configuration.   Short-term cycle predictions (i.e., of the immediately upcoming cycle) should be possible, however, with sufficiently good observational data.

If short-term predictions are possible, then we must ask why the current Cycle 24 predictions have such a huge range, even when the various predictions use similar models.  A key difference in existing models appears to be the treatment of magnetic diffusion, which, put simply, affects how long the Sun “remembers” the nature of previous cycles as it prepares to generate upcoming ones.  Models predicting a strong cycle 24 employ a fairly long memory of past cycles, while models predicting a weak Cycle 24 assume greater diffusion and less memory.

With all these uncertainties, is attempting to predict solar cycles just useless?  Not at all, since, as one presenter put it, the prediction is the ultimate test of your model.  Reliably predicting solar activity just 4-5 years into the future, rather than 40-50, using a model based on physical processes rather than statistics or numerology, would reveal important information about cycle-generating process.  The key ingredients are the physical basis of the model and the observational data to support it.

We also had a round table dicussion toward the end of the day about the present solar minimum, and there’s no question things on the Sun are pretty weird right now.  The present cycle minimum is unusually long and weak.  The Sun is dimmer right now than it was at the minimum of Cycle 22/23, and the minimum is persisting beyond all predictions.  The latest “consensus” predictions (of which there were actually two, one for a strong Cycle 24 and one for a weak one) suggested that by now (March 2009), the sunspot number should be between 17 (weak case) and 40 (strong case).  In contrast, the average sunspot number so far in 2009 is 1.  Other heliophysical data point to an unusually quiescent Sun, and so far (as is certainly the case for our own observations) there is no sign at all of an uptick in activity.  Additionally, the activity is quite asymmetric, with what few spots there are appearing in the southern hemisphere of the Sun only.  This highly asymmetric pattern was a characteristic of the Maunder Minimum.

So are we headed for a grand minimum?  So far, we cannot say for sure.  There will be more talks today about solar grand maxima and minima, but one presenter (see my blog post of January 31 for a discussion of his paper) argued during the round table that the current solar grand maximum of activity, which began around 1920, will end in 15±8 years and that we are headed either for a period of weak cycles or an outright minimum in the near future. 

In my own talk, I gave a review of stellar variations and their implications for solar cycles, including a sneak preview of some very interesting results we have in a paper I’ll be submitting next week.  I’ll blog about those, but would like to get the paper past peer review before I do.  Stay tuned.

The executive summary of day 1 is that this is a very interesting time to be working in this field.   The Sun is behaving very strangely, and a weak or nonexistent Cycle 24 is a distinct possibility.  Later today and tomorrow we’ll be hearing more about this, as well as the implications for effects on climate.  I’ll be back with further posts as the talks progress.

The figure below is from a recent paper by W. D. Pesnell in Solar Physics that nicely summarizes over 50 predictions of the strength of solar cycle 24. The y-axis shows the predicted maximum sunspot number R, and the various bars are color-coded by category of prediction method. The full paper, which has all the references for all of these studies, is here.

Obviously the predictions are divergent. They are also not all of uniform quality: most are the result of thoughtful work, but a few are downright weird (there’s one paper that forecasts cycles out to #38, of which I am a wee bit skeptical).

A very recent report from NASA examines the upcoming cycle in a number of ways, the mean of which is a maximum sunspot number of about 118±30, quite comparable to Cycle 23. However, this prediction assumes that Cycle 24 is going to be a normal cycle — and other indicators (e.g., the solar wind) hint that it might not. It certainly seems possible, though by no means yet assured, that we are headed for an extremely weak upcoming cycle.

A number of authors, including the careful folks in Sami Solanki’s group, have noted that the present-day Sun is in an unusually high level of activity compared with the past several millennia. Actually, one doesn’t have to look even that far; here is a plot I made of the sunspot record since 1610.

The Maunder Minimum is obvious, as are the weak cycles of the Dalton Minimum. There’s another dip in the early 20th century, followed by a number of very strong cycles. The most recent cycle, #23, was rather modest continues a decline in strength since Cycle 21. Clearly there are patterns of solar variability well beyond the 11-year Schwabe cycle.

One way to push the record of solar variability well back beyond the start of the sunspot record is by examining concentrations of isotopes of specific elements, such as beryllium 10 and carbon 14. These isotopes are created by cosmic rays, and the intensity of cosmic rays hitting Earth’s atmosphere is modulated by the magnetic fields in the local space environment, and those are direct manifestations of solar activity. The higher the isotope concentration, the weaker the heliospheric magnetic fields, and by inference, the less solar activity there was.

A few months ago a group from Switzerland and the UK published a Geophysical Research Letters paper on the beryllium 10 concentrations in Greenland ice cores, deposited there over about the last 10,000 years. The ice cores reveal many solar grand maxima and minima, enough so that the authors could examine the duration of the events with reasonable statistical confidence.

They predict the present solar grand maximum will have a lifetime of 95 years. Since this maximum has been underway for about 80 years, it therefore has ~15 years to go, or roughly 1-2 more cycles. However, the authors also find no correlation that suggests a grand maximum is reliably followed by something drastic like the Maunder Minimum. All the data support is that if Cycle 24 turns out to be weaker than Cycle 23, then it is likely the Sun will continue in the upcoming decades with either a set of weak to average cycles, or a full-blown grand minimum.

These results agree qualitatively with Solanki et al.’s, so the evidence for an increasingly less active Sun over the next several decades appears to be growing. It surely makes it an interesting time to be watching our star!

I’ve updated my article about solar cycle prediction with a few of the latest discussions from the literature. As I note in the article, not only are Cycle 24 predictions rather divergent, but there are now also a number of arguments that cycles are inherently unpredictable.

In any case, solar activity continues to remain extremely low as of January 2009, and our latest Ca II H&K observations continue at levels characteristic of dead minimum. Such a long minimum seems to imply a weak upcoming cycle, and one of the most recent predictions, from a November 2008 paper in Solar Physics, predicts that Cycle 24 will have a sunspot amplitude of 87±7, considerably weaker than Cycle 23, which was itself rather weak.