The CBA Belgium Automated Supernova-Search Program

Amateur astronomers play an important role in the discovery of supernovae : the vast majority of supernovae nowadays are found by amateurs, using CCD cameras attached to small- or medium size telescopes (20-cm to 60-cm). There are many different ways to systematically search for supernovae, although most methods share some fundamental characteristics. This article provides a detailed description of the CBA Belgium Automated Supernova Search Program, and the way it has been conceived. We hope it will be an inspiration to other amateur astronomers, who are considering to start a similar search program.

Discovering supernovae in a successful way, requires are basic understanding of what supernovae are - especially the distinction between type Ia and type II supernovae is essential to compile an intelligent list of target galaxies. That's why this article starts with a short explanation on the anatomy of supernovae (chapter 1). The most critical step in any supernova-search program is the selection of the target galaxies. We list the parameters, one has to consider for selecting galaxies and we describe how to compile a Search Program (chapter 2).

The rest of any Supernova-Search Program is a combination of hardware (chapter 3), software (chapter 4), persistence and luck. Depending on your telescope, mount, CCD camera and programming skills, you might put together a search program using commercial, off-the-shelf software building blocks, or you might want to sophisticate and tune the program to your individual needs, by writing your own scripts or software. Ours is a combination of commercial tools and own programming logic (chapter 4).


You're now ready to kick off the actual supernova hunting, and to collect large amounts of CCD images. The final bit of your Supernova-Search Program will consist of methods and tools to analyse images, and to compare them with images acquired on previous occasions. Again, there's a multitude of ways to do this post-processing. We present our method as an example (chapter 5).

You should of course ask yourself what the scientific value is of (discovering) supernovae (chapter 6). For some amateurs, this will probably be the main motivator to start a search program, for others it merely is "good-to-know" stuff. We conclude (chapter 7) with a list of Internet references, that are "musts" for any supernova (re)searcher. 


1. Anatomy of Supernovae

Supernovae (SN) are massive exploding giant stars, existing either as a single star or a binary system. They are so bright that they can collectively outshine all the stars in their host galaxy.

SN are divided into 2 basic physical types :

- Type Ia SN are binary systems, in which a white dwarf orbits a large, but less dense companion. Both stars are that close to each other, that mass flows from the companion to the white dwarf. After some time, the dwarf star takes on more matter than its supporting core can handle, causing it to go into thermonuclear instability and it collapses. Those are amongst the largest explosions in the universe.

- Type II SN occur at the end of a massive star's lifetime, when its nuclear fuel is exhausted. If the star's iron core is massive enough, it will collapse, producing tremendous shock waves, and become a supernova.

What happens after the explosion of a SN, depends on the type and mass of the progenitor star. Mostly, they produce a gas cloud called a supernova remnant. A good example of this is the Crab Nebula or M1.

Type II SN release more energy than Type Ia, but in Type Ia SN the released energy is more in the form of visible light, and therefore they are - generally speaking - brighter than Type II SN. The diagram at right shows a comparison between the light intensity (luminosity) of type I and type II SN, as a function of the number of days since peak activity. Type I SN are much brighter but decay much faster than type II SN. Type Ia SN all have very similar lightcurves, whereas type II SN show many differences in their lightcurves. A lot of type II SN have a "plateau" in the 2 months following the explosion.

Adapted (and corrected) from Chaisson & McMillan, Astronomy Today, 2nd Ed., Prentice-Hall, (1997).


2. Selecting target galaxies

Beginning SN hunters typically select target galaxies according to their brightness only. They build a list of bright galaxies and start hunting. This is a quite inefficient selection procedure, seriously reducing the chances on success. Compiling an intelligent target list requires a number of selection criteria to consider. The criteria we use for our SN-search program are : galaxy distance, morphology or type, luminosity, size, inclination, etc. 

The best source to start from is the Third Reference Catalogue of Bright Galaxies from G. De Vaucouleurs (1991), mostly called the RC3 catalogue. It provides extensive data for 23022 galaxies. 

a) Distance

The RC3 catalogue provides radial velocities for many of its galaxies. Starting from the radial velocity, one can derive the galaxy distance, using an (approximate) value of the Hubble constant. Once you know the distance of a galaxy, you can determine the maximum possible brightness of a type Ia and type II SN. Depending on the size of your telescope, the sensitivity of your CCD and your local observing conditions, you will notice that there's a cut-off point in distances. Simply said : there's a whole series of galaxies that you can exclude from your target list, because any supernova produced by such a galaxy will be too faint to be discovered with your equipment.

b) Morphology

Type Ia SN require a dwarf star with a companion star. In general, dwarf stars are very old stars, and therefore type Ia SN typically occur in older star populations in spiral galaxies or elliptical galaxies. However, type II SN are the result of quickly evolving young stars, and therefore happen in spiral galaxies where star formation is still ongoing. They never appear in elliptical galaxies.

To put it differently : spiral galaxies produce both type Ia and type II SN, whereas elliptical galaxies only produce type Ia SN. To maximize discovery chances, one might concentrate on spiral galaxies only. We nevertheless recommend to include a fair amount of elliptical galaxies in your target lists too, as type Ia SN have more scientific value to astronomers (see chapter 6). 

c) Luminosity

The frequency of SN explosions is directly correlated to the luminosity of the host galaxy. This is quite logical : the more stars in a galaxy, the higher the chances on having a SN amongst them. 

d) Size

Galaxies that are too small (major axis) should be excluded from your target list as well. Again, use a cut-off point to decide on this criterion, or divide galaxy sizes in classes (bands) giving a weight to each class.

e) Inclination

This is a fairly obvious criterion, and for most galaxies, you will find its inclination in the RC3 catalogue using the so called "Log R25" value. 

A formula for sifting galaxies 

It is obvious that the creation of an intelligent target list is not straightforward, but the more energy you spend in making it as good as possible, the better you maximize your discovery chances afterwards. 

The idea of building a "formula" to intelligently sift galaxies from a catalogue came from Berto Monard, who told me he was using this approach to quite some success. Such a formula combines all of the above parameters, and for any given galaxy will return a "probability number" indicating the chance on seeing a SN in that particular galaxy.

I spent weeks building a suitable set of formulae in Excel, exchanging many emails with Berto Monard. I finally came up with a formula that I tested on many of the past SN discoveries, and very much to my satisfaction, the resulting probability numbers were reliable and consistent. The biggest problem in setting up the formula, was to decide on what to do if one or more parameters are missing. In that respect, we recommend to complete the RC3 catalog as much as possible with data coming from other sources, before starting your selection process.

Try and build your own formula to do the sifting. You can make it as complex as you want. For instance : for each possible morphology type (we distinguish 40 different ones in our formula), we are applying two "weight" factors : one for type Ia SN and another for type II SN. 

Once the sifting criteria were determined, we applied them to the entire RC3 catalogue and eliminated all galaxies with a probability number below a certain threshold. We then sorted all target galaxies by minimizing the slew time between objects, hence increasing the efficiency of our robotic telescope mount. There is various ways to accomplish this sorting job. We used Software Bisque's TheSky.

The resulting sorted Target Galaxies List then was split in Right Ascension (RA) chunks of "2 hours" or sometimes 3 hours. For each chunk, I produced an A Program and a B Program. The A Program consists of target galaxies with a probability number above a certain threshold (highest SN chances), and therefore is the more important one. The B Program contains the remaining galaxies, and is to be used on evenings when all A Program galaxies have been exhausted. 

Some numbers : on average, a 2 hour chunk in my Supernova Search Program consists of 100 to 150 galaxies. Evidently, not all galaxies of the A Program can be visited in the course of a month (e.g., some will be too close to the Sun). I try to visit the A Program galaxies twice a month, although I usually don't succeed, due to the limited time I spend on SN hunting (to not compromise my Cataclysmic Variables research), and - more importantly - due to the very limited amount of clear skies we have in Belgium. 

All my Target Galaxies chunks are stored as Microsoft Excel files. At left is an excerpt from the A Program chunk for RA 0h to 2h. 

The columns in green color are the ones that contain input data for my SNHunt program (see further). The yellow columns consist of output produced by SNHunt or comments.


Extract of my A Program target galaxies list for RA 0h to 2h. Notice the Points column, that contains the "Probability Number" of each target galaxy. 

3. Telescope, mount and CCD camera

In principle, any telescope and CCD camera can be used to do supernova hunting, as long as your mount points and tracks well enough. Most amateurs are using a relatively small SCT-type telescope with a sensitive CCD (B&W). 

However, the story becomes a lot more complicated when you consider to automate your SN-search program. First of all, you need an accurate permanent mount with "GOTO" capabilities. Precise tracking is of paramount importance, as typical CCD exposure times will be ranging from 40 till 100 seconds or more (depending on the limiting magnitude you desire). Next challenge is to bring your telescope, mount and CCD camera in "perfect harmony", i.e. you will need a software program that can sweep your telescope to a target galaxy, make a CCD exposure, store it away, and move on to the next galaxy in a long list of targets to explore. A truly automated SN-search program should be capable of working autonomously for several hours on a row, while you catch some sleep (especially if you must get up and go to work the next morning).

All of the above might sound fairly easy to accomplish and there's indeed some good commercial mounts and software on the market, that can do the job. I definitely would recommend Software Bisque's mount/software suite combination for any amateur considering automated SN-searches.

My observatory setup, however, was far more demanding, and I had to overcome several hurdles, taking me many months to complete a software program, that finally automated my equipment to the desired level of autonomy. Here's a quick overview of the most important hurdles I encountered :

- my permanent mount has limited tracking accuracy, and becomes unreliable for unguided exposures above 30 seconds. Since all of my SN CCD exposures are 60 seconds or more, I had to find a way to automatically guide every image. I'm using MaxIm DL/CCD for camera control and imaging, and had to program it in such a way that it would automatically identify a suitable guide star for every image. But what to do, if there's no guide star available ? 

- my mount is not perfectly polar aligned. As long as the sweep distance between two successive targets is relatively small, the target will end up in the CCD frame.  In all other cases, it won't. I have written software to sync every CCD image with the actual telescope position, and to correct if necessary. 

- my mount controller sometimes "gets lost", that is : every now and then, but unfortunately not predictable, a sweep operation is terminated in RA, a few RA minutes before or after the destination is reached.  Detecting such cases is hard, and it took me lots of effort to enable my software to finally recover from such situations.


4. Software for automated SN searching

4.1  SNHunt - the heart of the CBA Belgium SN Search Program

I am using a combination of commercial software packages and self-written programs to implement a fully automated SN search program. These are the ingredients :

- CCD camera control, imaging and guiding are all done using MaxIm DL/CCD. MaxIm provides an ActiveX Automation interface for externally controlling the CCD camera operations. ActiveX is a platform-independent method for linking different applications together. I'm using Visual Basic to completely control and automate MaxIm DL/CCD. 

- Telescope control (sweeping to target galaxies) is done using the ASCOM (Astronomy Common Object Model) platform. This is a set of technologies and drivers that form the basis for interoperability between ASCOM-based tools. The ASCOM Platform supports a variety of telescope types via standard drivers. 

- Telescope syncing (correlating a CCD image with the actual position of the telescope) is done using TheSky. TheSky has no direct automation interface for external control, so I had to use other means to tell TheSky what to do.

- Target galaxies are loaded in my program as Microsoft Excel files (see example in chapter 2).

I have written an extensive Visual Basic program, called SNHunt, to glue all of the above components together, and to build in the logic to overcome all of the (hardware) hurdles described in the previous chapter

SNHunt executes on the laptop computer in my observatory. It takes as input a simple Excel file with target galaxies, and will slew the telescope from one galaxy to the other (provided its altitude is within a controlled range), instruct the CCD camera to take a guided exposure, consult TheSky to ensure that the CCD image contains the target galaxy, and reposition the telescope if needed.  

On average, once the telescope has been setup, it takes no more than 5 minutes to slew to the first target galaxy, using SNHunt, and kick-off a night's searching.


4.2  SNDashboard - to catch some safe sleep

In addition to the above functions, SNHunt also produces a log file with relevant information. That log file is used by another software package, that I wrote, called SNDashBoard. It follows-up and controls the proper execution of the search in real-time mode, using my home computer (connected to the laptop computer in the observatory).

Amongst other things, SNDashBoard monitors sky conditions and a computer-controlled voice will wake me up if clouds are entering the sky. In fact, my wife and I have 2 baby phones in our sleeping room : one to monitor our youngest son, the other is connected to SNDashBoard.  

The image at left is a typical sample of SNDashBoard in life operation. As long as the leds are all green, the SN-search program is executing well and no human interaction is required.

A "Remote Abort" button allows to interrupt the imaging sequence in the observatory, directly from the home computer. 



5. Post Processing of Images

A successful observing night will produce dozens of CCD images, that need to be inspected for SN candidates. Although this part of the search process can be automated too, I prefer to visually inspect my CCD frames. Nevertheless, I implemented a software tool that brings a lot of relief to the inspection job.

The most important step, evidently, is to compare your CCD images with reference images. The internet provides instant access to many professional sky surveys (e.g., the Digital Sky Survey), that are tremendous tools for every SN searcher. Yet, it is my experience that the best reference images still are your own ones. I therefore always first compare my CCD images with own reference images, before moving on to professional exposures. 

SNViewer is the software tool, that I wrote in Visual Basic, to assist me with the post processing of my SN-search images. It works in close conjunction with MaxIm DL. I first select the folder, that contains the images to be explored (called the "Tonight folder") and the folder containing the reference images. I then follow a sequence of steps, which is made explicit in SNViewer by separate push buttons. Short :

- "Get Image and calibrate" loads a new image from the "Tonight folder" and opens it in MaxIm DL, after which it is automatically corrected for bias, dark frame and flat fielding. Simultaneously, SNViewer loads a corresponding DSS (Digital Sky Survey) image from a library of images on my hard disk, showing the same galaxy.

- "Get reference" loads a reference image from my references folder and opens that image in MaxIm DL for a first visual comparison.

- "Blink images" is an optional button that I can use to aid the inspection. It will overlay both images and "blink compare" them. 

- Sliders in SNViewer allow to adjust the brightness, contrast and image scale of the DSS reference image. This is important as it allows to modify images to examine both the core and the faint outer regions. 

- Once an image has been inspected, I use the "Save as suspect", "Save as reference" or "Archive" buttons to move the image resp. to a folder with suspect images (possibly containing a SN), to the reference folder (from then on, the current image will act as reference) or to an archive folder (for backup purposes).

- One additional button, worth mentioning, is "Create Priority File". When pressing this button, my home computer notifies SNHunt (chapter 4) executing on the laptop computer in the observatory, that a priority target has been identified, and that the telescope has to move to the target, upon completion of the current imaging sequence. This is especially worthwhile to double check suspect images.

Using SNViewer, I can process approximately 60 - 100 images per hour, sometimes more.  

The SNViewer tool will load newly acquired CCD images one after the other. Every image is presented in MaxIm DL, together with a so called 'reference image', acquired on previous occasions. In addition, SNViewer will also load a Digital Sky Survey image of the galaxy under investigation. I have included a "blink images" button in SNViewer, that allows to blink-compare CCD images (by overlaying them). 


6. Scientific value of SN hunting

Starting around 1933, astronomer Fritz Zwicky was the first one to extensively research SN. He not only coined the term "supernova", but today he still is the individual who holds the record of the most SN discovered. Since the pioneering work of Fritz Zwicky, the know-how on SN has vastly extended. It is beyond the scope of this article to elaborate on the physics of SN.

Yet, it is important to realise that many SN questions are waiting for an explanation. Hence, the discovery of SNe is still attracting a lot of attention from the professional astronomers community. The importance of an automated SN-search program therefore is to find SN shortly after they explode, so that professionals can track the exploding star over its entire cycle, and learn as much as possible about the physics of supernovae, providing clues to the fate of the cosmos.

There are many good examples to illustrate the importance of SN discoveries. Here is one, pointing at a recent and very significant finding, made by two independent teams of astronomers, studying distant SN. Type Ia SN for quite some time are used as a "standard candle" to estimate galactic distances. Astronomer Alexei Filippenko (University of California) and an international crew of astronomers - the High-Z Supernova Search Team - drew upon the distances of Type Ia supernovae to conclude that the expansion of the universe is accelerating, an observation that implies the existence of a mysterious, self-repelling property of space first proposed by Albert Einstein, which he called the cosmological constant. Independently, the same conclusion was reached by another team of astronomers, the Supernova Cosmology Project.


7. Internet references

Some useful references for SN (re)searchers :

International Supernovae Network : the place to be for SN hunters !

The "College de France" has an excellent site on SN (in French) discussing theoretical as well as practical aspects, research programs and professional & amateur SN-search initiatives.

Marcos J. Montes is maintaining an extensive list of WWW pages related to SN and SN remnants.

Tenagra Observatories is an example of a successful pro-am collaboration in SN research, and amongst the leading SN-search teams in the world.

The Pucket Observatory is conducting the best-known amateur SN-search program.

Tom Boles of Coddenham Astronomical Observatory is one of the leading UK SN-searchers

... and see the list of Internet references on my links page (e.g., to access the Digital Sky Survey DSS).





Copyright © 2002 - Tonny Vanmunster.