The Sun Blog

I’ve posted the last of the technical articles I’ll add to the site for a while, detailed the procedure that generates our results for solar observations of Ca II H&K. You can find it here. For now, here is the main plot, which shows solar cycle 23 in terms of three quantities: the HK index (blue), Mount Wilson S (yellow) and excess magnetic flux (red). These quantities are defined in the article about our stellar time series, but I’ll soon be adding some non-technical descriptions of them as well.

One thing to note about this plot: it does look like Cycle 23 has finally bottomed out. Our most recent data, obtained with our new cameras on 55 different days since August 2008, yield a mean activity that appears to be slightly above that of 2007.

If you’re wondering about the little red X’s — our routine looks for bad spectra and flags them to be ignored in analyzing the results. The X’s show which ones they are. Not surprisingly, they primarily turn up for data points far outside the typical activity values for a given time frame.

One of our goals for the calendar year 2008 was to reconcile our new camera data with the old camera data, and we just made it!   I’ve posted some blog articles about our data reduction procedures, and in the past few weeks we’ve been working on the final step, which is to extract stellar activity measurements from the reduced spectra.

I have just uploaded a technical article about the data products, which includes an example of one of our Ca II H&K time series.  In January, I’m planning to switch from all this technical writing to some more layman-level articles about how we measure sunlight and starlight, so stay tuned to the blog for those.

If you are interested in solar and stellar variability, you’ll hear a lot about the so-called H and K lines of singly ionized calcium. They lie in the near ultraviolet part of the spectrum, and my earlier posts about our data reduction procedures are ultimately about extracting these spectra accurately from our CCD frames. Having added articles to the site about all that, now it’s time to explain the analysis procedures.

The first step is to deal with a difficulty in studying solar and stellar spectra in the blue part of the spectrum. Stellar spectra are composed of a bright, generally smoothly varying continuum with superimposed dark lines.  For analysis of the lines, we want to set the continuum to 1.0 everywhere; then we can compare lines from spectrum to spectrum consistently.  But in the blue spectra of cool stars, there are so many lines that continuum is obscured.  The technical term for this obscuration is line blanketing,  and in the HK spectrum, the blanketing is severe.  The figure below shows a sample spectrum of the solar twin 18 Scorpii. The two huge dips in the middle are the K (left) and H (right) lines of Ca II, and the small changes at the very bottoms of these lines are the linchpin for understanding long-term stellar activity.

The dark blue spectrum is what we get from our data reduction routines, but it’s not correct. You can see two points of the spectrum that equal 1.0, but that’s not the true intensity there due to the heavy line blanketing. We explain (albeit in somewhat technical fashion) how we come up with the light blue spectrum, which is what we want, in this article I just added to the site.

The power of this method is that we can now measure the variations of stars from the spectra. We can’t image them like we can the Sun, but the spectrum above contains information we can use as a proxy for variability. And by observing the Sun the same way, we can compare it directly with the stars.

The latest update to the SSS site is a set of articles for researchers and die-hards out there about how we reduce our data from raw frames to usable spectra. I’ve posted a couple of blog entries earlier about the first steps, and I’ve now finished the rest of the documentation in four separate articles on the “tech info” part of the site. These describe the process from the point where we have traced the locations of spectra on our CCDs to extraction and normalization of the solar and stellar spectra. Each article begins with a (hopefully) layman level introduction to the step being discussed, followed by all the details.

None of this would make a particularly exciting screenplay, but we think it’s useful and valuable to have our methods online. The table of contents for all the articles is here, under the “Detailed Procedure” header.

It’s fun to sprinkle a resarch blog with some celestial eye candy now and then. Lowell Observatory Advisory Board member Mike Beckage sent me this really nice photo of last night’s conjunction of the waxing Moon, Venus, and Jupiter, taken from his home near Los Angeles. Thanks Mike for a great shot! I guess I could tie it to the Sun by noting that everything you see is reflected sunlight…or multiply reflected sunlight in the case of the cool Earthshine on the non-illuminated part of the Moon.

Our local paper ran an editorial yesterday in which the author lists several references from “credible scientists who are not convinced” of the existence of a human-caused component to global warming. One of these references is a widely quoted paper by Habibullo Abdussamatov of the Pulkovo Observatory that, according to the editorial’s author, “showed the correlation of solar activity to global temperature.” This is untrue; Abdussamatov’s paper is deeply flawed and does not make a remotely convincing case that the Sun is the principal force behind global warming (or global cooling).

The paper is based upon the argument that the solar radius is larger at activity maximum and smaller at minimum, and that this drives the Sun’s luminosity variations. Long-term, consistent measurements of changes in the solar radius are extremely difficult to do. A number of astronomers have worked on the problem, and the activity-radius correlations they have reported are minimal to nonexistent. (That’s a lengthy discussion in its own right, and I will blog about it in another post). It is not radius variations, but brightening of the Sun by faculae versus darkening by sunspots dominates the cycle-related luminosity changes.

The rest of this paper is equally erroneous. Example: “Increase of the [Sun's] core’s temperature and corresponding expansion of the whole Sun can catalyze the raise of activity…with the amplitude of the temperature variations determining the power of a cycle.” Wrong; the catalysts of the activity cycle are the Sun’s differential rotation and its subsurface turbulent convection, which combine to create a periodic variation in the strength and extent of regions of strong magnetic fields, which manifest themselves in the familiar signatures of “activity” (e.g., sunspots, prominences, flares and coronal mass ejections).

Unfortunately, this paper was picked up widely by the media, and it has been parroted ever since. For credible opinion on solar variations and climate change, one can do much better, and for an excellent survey by a genuinely reputable scientist, I recommend The Sun and Earth’s Climate, by Dr. Joanna Haigh of Imperial College, London.

I’ve posted an article detailing how we determine the location of spectra on our CCDs, following up to part 1 of our data reduction posts.   The design of our spectrograph is such that we observe 20 (for our old cameras) or 16 (for the new cameras) separate chunks of the spectrum, arranged in a rectangular array (shown in the image below).

Each of the strips of spectrum visible in the image is called an “order,” and tracing their locations across the detector gets a bit tricky since most of them are curved. This curvature isn’t careless design or needless complexity; it’s a byproduct of our “red” spectrograph’s need to disperse the sunlight and starlight twice. We use a diffraction grating that disperses light at very high resolution along the x-axis of the detector, but the tradeoff for the good resolution is that we end up with lots of overlapped orders. So then we have to use a prism to “unstack” the orders, spreading them out along the y-direction. The grating is called the disperser, and the prism is the cross-disperser. This double dispersion makes the orders curve. The order at the very bottom of the image comes from our “blue” spectrograph, which only uses a single grating and therefore produces a straight order.

Creating an accurate ordermap is critical to the rest of the data reduction process. Here are all the gory details describing how we go about this.

I’m sometimes asked under what conditions we can observe, and whether we actually observe all night.

The SSS project is assigned what astronomers call bright time, or the moonlit part of the month.  We are observing bright stars and we don’t really care how bright the sky is, since it’s much fainter than any of our targets even during full Moon.  Our targets are scattered all over the sky, so there’s always something to observe and in general we do pull all-nighters during our assigned time.  We typically have a 7-12 night observing run each month, and since there are three of us to split the duty, any one of us usually only observes 2-4 nights per month.

We can’t observe when it’s cloudy, as our optical telescope can’t see through clouds.  Cloudy doesn’t have to mean overcast, either.  It could be a fairly nice night by ordinary standards, but if we have a layer of the thin cirrus common here in the Southwest, we may have to stop observing.  It’s not so much the clouds themselves, but if bright moonlight is coming through them, then we basically get sunlight (which is what moonlight is, after all) in all of our stellar spectra.  Not good for getting the right answer.   We can indeed keep observing if there are thin clouds and no moonlight — we just have to increase our exposure times a bit.

As an example, I was scheduled to observe just this past Wednesday night, and it turned into one of those irritating times when you can’t decide whether to drive out to the telescope site or not.  There were bands of cirrus drifting through all night, with a nearly full Moon.  Astronomers call the brief clear intervals during such nights “sucker holes.”  That is, you see a clear sky and get suckered into packing up a night lunch, driving to the telescope, getting everything powered up and ready to go…just in time to have the next big blob of clouds arrive.  This time, I decided get a good night’s sleep instead.  At dawn, the Sun rose through a pretty thick layer of clouds, so I made the right call.  Sometimes I’ve blown it, though!

This is the first in what will be a series of posts about data reduction, which is the science (and art) of taking raw charge-coupled device (CCD) observations and converting them into solar and stellar spectra suitable for analysis.  Several steps are required to get from point A to point B, and as part of our current research grant, we’re documenting what we do to the data and posting the methods, the software documentation, and ultimately the code itself on our Web site.

Today, I finished the article and software documents for the first step, called debiasing.  It’s a pretty straightforward business.  Every raw CCD frame has a fairly uniform background unrelated to the observation, and we need to remove it before doing anything else to the data.  Look here to read the thrilling article about how we do this, and the record of this background over 16 years of observations.  This article (as will upcoming ones) contains a paragraph in boldface green type that is — is at least is supposed to be — a nontechnical description of what we’re doing to our data.  Hopefully it is!

In January 2008, something — perhaps a power surge during a winter storm — fried the electronics in our old detector system.  Bad news: no observing for a while.  Good news: the new cameras we installed this past summer are much more sensitive, and we can plow through our observing list much faster.  Just a few weeks ago, we finished a huge observing run, during which we collected about 700 stellar spectra over 14 nights.  With the old cameras we might have obtained 250-300 spectra.

We record sunlight and starlight on CCD’s, or Charge Coupled Devices.  They are detectors of choice in modern astronomy.  They are square or rectangular arrays of light-sensitive pixels.  The CCD electronics convert the amount of light hitting a pixel to a number, so our raw data frames are basically a big matrix of numbers that record the pattern of light landing on the CCDs.   The figure below shows what we got from the old cameras (left, falsely colored in red) and the new cameras (right, in blue).

The straight line at the bottom is recorded by one CCD; the several curving lines above it are recorded by another. Each of these lines is a different part of the spectrum of the Sun or a Sun-like star, and the essence of our program is analysis of these spectra to examine the stars’ magnetic activity and variability. To do this, we have to reduce the data (i.e., extract the spectra from the raw data frames above into a form suitable for study), and then measure what’s going on in them. That procedure is fodder for lots of Web pages and blog posts, and our formal documentation is developing here. For now, I’ll just note that we did our best to keep the new data format as similar to the old as possible. The green and pink ellipses above indicate key spectral features, and as you can see, they’re roughly in the same spots. I have expanded our data reduction software so it will know which type or frame is which and handle the reduction of each correctly, and we’re presently working on making analogous expansions to the analysis routines.

Ultimately, though, the figure above shows the nature of the game: we have some 30,000 arrays of numbers that contain some of nature’s beautiful patterns. Converting numbers to understanding is the interesting part. Stay tuned, and we’ll try to document and blog thoroughly and clearly about what we’re doing.