Three Rules

Dominik Gronkiewicz

NOTE. Even if you are an advanced solar or planetary photographer, please read the following page carefully, since the process of collecting scientific data is slightly different than shooting breath-taking images.

Excited by the idea of catching your first solar flare on image?

Before you set up your telescope and start shooting pictures for F-CHROMA, please wait a moment and let me introduce three rules for collecting solar amateur scientific data.

1. Time: we need a sequence of images (so we are able to trace the evolution of a phenomenon), each of them being well localized in time (to make possible the joint analysis with data from other sources, including satellites).

2. Calibration: possibly all instrumental effects should be removed in the data that we receive from you.

3. Objectivity: digital enhancements such as sharpening, denoising, contrast adjustments are not allowed since those are subjective and destroy the raw signal.

Rule 1: Time is important

Someone said that „observing without noting the time is barely looking at the sky”. Being an amateur astronomer I didn't understand this until I started astronomical studies. Understanding the importance of time is essential to perform observations valuable for science.

While information about time is rather an interesting completion of your beautiful prominence photograph, for doing science it's absolutely essential. It's because a single instrument is rarely enough to obtain a full information about the flare or eruption. To synchronize temporarily X-ray images from RHESSI telescope with ultraviolet images from SDO observatory, or X-ray flux from GOES satellites with H-alpha images from ground-based observatory, or a combination of all four, we need to know precisely when each image or measurement was taken. Because your data may be used along or compared to observations from other instruments, your pictures must be precisely localized in time.

Nowadays, the easiest way to obtain a decently precise time is to use a computer. There are numerous clients that allow precise and frequent time synchronization with NTP servers. An example of very good and free application is NetTime (http://www.timesynctool.com/). It periodically synchronizes the system clock to desired accuracy (up to 50 milliseconds). Such precision is far more than enough to consider your observations well timestamped.

Users of DSLRs must pay special attention to clocks in their cameras. A good idea is to synchronize the clock every time the battery is removed. We need an accuracy of a few seconds or better. You can watch the drift of your camera clock by comparing it to a well-synchronized clock (for example, your computer). This is how you can determine the required frequency of clock adjustments.

Timezones are always very confusing for all the people. To avoid confusion, all times of solar observations are given in UT (Universal Time). Of course, we don't expect you to have your computer or camera clock running in UT! However, remember to have the timezone and daylight savings set correctly. Otherwise, we're not able to convert your local time to UT. In such case, your observations are useless.

FireCapture users can force the program to save all the times in UT. I strongly recommend using that setting.

Another thing that might suprise you as a solar photographer is that we professionals are rarely satisfied with a single picture and prefer sequences of images. That allows us to trace the evolution and changes of phenomenon. If you visit SDO site (http://sdo.gsfc.nasa.gov/data/), you can always see our Sun almost live. SDO takes a picture of our Sun in ultraviolet light every 12 seconds in amazing resolution of 4096×4096 pixels. This has two great effects: a huge scientific value for astronomers and fabulous colorful movies of eruptions and flares that SDO team publishes regularly on their blog.

If your observatory works in H-alpha line, you might expect quite frequent (at least one per day) occurrences of big or small flares once a decent active region pops up on the solar disk. However, white-light observers are seeking for extremely rare, elusive and short (up to 3 minutes) brighenings during the strongest flares. Regardless of whether you observe in H-alpha, Ca-II or in white light, we need your images to be taken in cadence of between 10 and 20 seconds. If, for some reasons, you cannot maintain that cadence, less frequent observations will also be valuable, especially in H-alpha or Ca-II lines.

Rule 2: Calibrate your data

It does not matter whether you observe with a 60 millimeter refractor or Hubble Space Telescope. Telescopes are imperfect. All of them. Luckily, we have developed numerous techniques that are used for elimination of these imperfections. You might have heard about these techniques, but as an amateur astronomer you probably never associated them with solar and planetary photography.

The very basic reduction of solar images is based on three kinds of calibration frames: dark frames offsets and flat fields. Let me discuss the role of each of these frames.

Offsets and darkframes are two kinds of calibration frames that are taken with the telescope lid covered. You could ask „why should I waste my time for recording darkness?” Well, that's not entirely true. Apart from darkness, your camera will register its intristic noise and a very low (but nonzero) signal level. In other words, we take a photograph of camera noises. Later we can subtract it from our images in order to reduce the noise! The difference between offsets and dark frames is their exposure time.

Offsets (or bias frames) are taken using shortest exposure time available. Their aim is to register the „offset”, that is, the lowest value that each pixel can indicate. You could expect that signal value in an unilluminated pixel should be zero. That's simply not true, and for many reasons (that I'm not going to describe here) the „zero point” of each pixel is a non-zero value. To calibrate our pictures (in other words, to move the „zero point” to zero), we take a sequence of offset frames, average them (to reduce the random readout noise) and subtract it from our image of the Sun. Easy, isn't it?

Dark frames are taken at slightly longer exposure time – to be precise, identical to exposure time of celestial object (Sun in our case). Dark frames, apart from the offset level and random noise, contain signal associated with dark current, which is a noise generated by thermal movements of electrons. Dark current increases with exposure time and with temperature. By cooling the camera, dark current may be reduced to almost non-existent level. In solar photography, we usually use very short exposure times and low camera gain (or ISO speed). For these reasons, we may often consider dark currents negligible and limit our reduction to offset subtraction only.

Flat field is totally different from two previous frames. It's a photograph of uniformly illuminated surface. That makes all image defects visible, such as dirt and dust on camera sensor, telescope vignetting and even differences between the area of single pixels. A good example of uniformly illuminated surface is blue sky. Laptop screens with blank white web page displayed also work well. Those solar photographers who use instruments with very narrow field of view (capturing little portion of solar disk) can photograph a slightly unfocused image of quiet area on the Sun. Of course, flat field is a photograph just like any other – therefore a proper dark frame (or at least an offset frame) must be subtracted before it is usable. Just like in case of offsets and dark frames, it's advisable to capture a series of flat fields and calculate an average in order to reduce random noise.

Note for advanced amateurs. If you feel surprised right now, that's okay – so I was. After all, why would anyone use reduction characteristic for deep-sky imaging for processing of the brightest object on the sky – our Sun? That might be understandable while trying to extract faint galaxies from a noisy photo, but is it really necessary in solar imaging?

My answer is: it depends. Effects associated with dark currents as well as telescope vignetting are indeed negligible during collecting data for a pretty picture and can be compensated with level adjustments, stacking and other post-processing. However, in F-Chroma we do not collect nice pictures. What we collect is scientific data captured by amateurs. Hidden signal defects that are not detectable to human eye may have influence to the analysis and interpreted incorrectly. Therefore, if anything can be corrected, it should be. Period.

Last but not least, let me say a word about a powerful technique called stacking. The principle is rather simple: average a sequence of frames in order to reduce the noise and cancel out atmospheric distortions. However, anyone who has ever used a telescope, probably know that guiding telescope at high accuracy is very often an impossible task. Luckily, stacking software is able to stabilize the images before averaging in order to obtain the best sharpness and resolution possible. More advanced software (such as AutoStakkert, autostakkert.com) is even capable of unwarping the atmospheric distortions on large image areas! Images obtained through stacking are usually very soft in their nature. Averaging many frames stabilizes the geometry and boosts the S/N ratio but causes a slight blur of images (or, scientifically speaking, convolution with a Gaussian kernel). However, nowadays there are many good deconvolution algorithms that allow to extract subtle details from the image. In this way, one can get resolution much better than normally allowed by atmospheric seeing. However, all limitations caused by resolution of optics still hold.

To summarize this point, I would like to show you two versions of an image taken with my camera and 20cm Newtonian telescope working at focal length 2500mm, monochrome camera, UV/IR cut filter and 540nm pass filter. Both are sharpened stacks of about 300 best frames picked out of sequence of 2000 frames. Right one was corrected for dark frame and flat field, while the left one was not (which is, unfortunately, a very common practice among solar and planetary observers). I leave the comment on the images below to you.

Rule 3: Don't enhance

The purpose of F-HUNTERS is create a database of scientific observations performed by amateurs. For that reason, we appreciate that you send us the data in lossless format (such as FITS or TIFF) at full bit depth (at least 16 bit). The data should be only corrected for darkframe/offset and flatfield. High-cadence movies should be stacked to TIFF format but not sharpened (please attach FireCapture log text files).

Any further enhancement are dependent on interpretation and should be left to end user of the data. Below I present a comparison of raw stack (of course, corrected for dark frame and flat field) and two attempts to processing, one for enhancement of sunspot details, and other to boost the contrast in surrounding photosphere and make traces of supergranulation visible. As you can see, the final processing is strongly dependent on aim of the observations. Providing data as raw as possible and allowing the user to decide is the best way.