TASS – The Amateur Sky Survey

Science History In The Making

Perhaps one of the most ambitious and sophisticated amateur undertakings in the history of amateur astronomy, TASS is the brain-child of Tom Droege a former staff member of the Fermi Laboratory. He has assembled an extremely able team of fellow astronomers that have been able to write all of the software required, analyze and catalog the data and co-author sophisticated technical papers.

The team consists of Michael Richmond (RIT), Tom Droege (FNAL), Chris Albertson, Arne Henden (USRA/USNO), Glenn Gombert, Herb Johnson, Ron Wickersham (TASS), Nick Beser, Marty Pittinger, Bernie Kluga (JHU/APL), and “a supporting cast of tens”.

Tom Droege originally built 7 “triplets”, called the Mark III, each of which contained 3 Kodak KAF-0400 CCD cameras on a single mount. These were distributed to members around the country, who placed them in their backyards and pointed them at the celestial equator. Each camera made a continuous drift-scan image of the sky. Analysis of the data was handled by software written to identify stellar sources and measure their properties.

One of the early goals of this most ambitious project was to map the night sky and discover variables and other night objects heretofore unknown. Specifically, to map bright stars, of sixth to fourteenth magnitude, in a region around the celestial equator. The position and brightness of each star in several different passbands (Johnson V and Johnson-Cousins R and I) was to be measured. All their measurements were to ultimately reside in a single database which anyone could access via the World-Wide Web.

They began scanning in October, 1996 and accomplished the following:

  • total number of detections: 9,562,450
  • total number of sources: 2,560,731
  • number of “good” sources (seen >= 10 times): 178,570
  • detections of “good” sources: 6,901,776
  • accuracy of a single detection’s position: 3 arcsec
  • accuracy of mean position of a good source: <1 arcsec
  • precision of a single photometric measurement used for differential photometry:
  • 0.02 mag for V <= 10, 0.10 mag for V <= 14

Tom Droege is currently constructing the first of a series of twenty (20) new CCD cameras that are called the Mark IV. The first of these is running and taking data. #s 2-7 are going to various locations in the US including Mt. Wilson and the Naval Observatory. #8 goes to Pretoria South Africa. Built around a Loral 2Kx2K chip, the Mark IV has a field of view of view is 4 x 4 degrees. Unlike the Mark III, it is operated in point-and-shoot mode, so a mount and control system was designed for it. The Mark IV will be able to view all areas of the sky overhead, not just the strip around the celestial equator. It will reach about fourteenth magnitude from backyard sites.

The project is entirely self-funded, with Tom Drodge doing the bulk of the heavy financing. The plan is to place the latest version of their instruments, the Mark IV, around the world and pass data back and forth via CD and/or the Internet. Linux was chosen as the operating environment of choice and nights are spent among water filled tubing, electronics and an array of optical devices.

The group is currently writing the definitive scientific paper on line. The first (very) rough draft is included below. The group is presently writing the definitive scientific paper on their project on line. Watch this space to see how scientific papers are pulled together and edited. The very rough words of this version will be smoothed as everyone works out exactly what the group wants to say.

The TASS web site URL is :https://www.tass-survey.org.

The Amateur Sky Survey

Last edit: August 5, 1999


This document is the result of many years of painstaking research and development, laborious data collection and collation, and system refinement by Tom Droege and the contributors listed in the credits. However, as time went by, and with ever increasing membership in the TASS group, others have added significantly to the success of this venture.


(in alphabetical order by author) Data Base: Chris Albertson Field Operation: Nick Beser Editor: Richard Clingempeel Original Idea, Introduction: Tom Droege Error Analysis, Analysis Pipeline: Michael Gutzwiller Error Analysis: Arne Henden Error Analysis, Scientific Motivation, Sample Data Results: Michael Richmond


Tom Droege

When the comet Shoemaker-Levi 9 collided with Jupiter in 1994, I started following the news on the sci.astro news group and the world wide web. As an instrumentation specialist[1] with no previous experience in astronomy, I was attracted to the idea of building a machine that would perform an “all the sky all the time” search for such objects. A few posts to sci.astro determined the status of searches and it seemed that it might be both worthwhile and fun.

From the beginning, it was obvious that the apparatus was only a small fraction of the effort for such a search. The problem was not how to build hardware, but how to develop a scheme that would assemble the right sort of team to attack this problem without spending a fortune. Meanwhile, the “establishment” efforts for such searches [2] seemed to be fading for lack of government support.

My thinking was to try to design a scenario where it is possible to perform such a search with no more resources than might have been available to a nineteenth century gentleman amateur. Further motivation was to design a project that would provide stimulation during retirement.

While a lot of hardware can be built with modest resources, hiring full time programmers is expensive. Faced with the problem of finding programmers, I was reminded of Willie Sutton who, when asked why he robbed banks, replied “That’s where the money is.” So, when faced with the problem of creating software, it was natural to turn to the internet because that is where the programmers are.

The effort was started by making information requests to the sci.astro news group. After a little knowledge was gained, a note was put on sci.astro that asked whether an all sky survey to magnitude 15 would be of any use. The positive response encouraged me to set out on the attempt. Little did I know what an ambitious project I had undertaken.

Thus the idea of “The Amateur Sky Survey” (TASS) was developed. An early plan was to assemble an array of CCD cameras using camera lenses. The array was to consist of 10 lanes each covered by 3 cameras spaced at intervals of one hour in right ascension. This was to be accomplished by the design of a CCD camera that could be semi-mass produced at low cost. This design was to be done in real time on the internet so that the process could attract the inhabitants (many of them programmers) that happen also to be interested in astronomy.

As time went on, I realized that it might prove to be a better plan to give the camera triplets away to programmer/astronomers that showed interest in operating them and in helping to do the necessary science. Later, the work on the various prototypes was reported to the news group, and a following was developed. This was, at first, moved to the CCD mailing list [3] then later a TASS listserv [4] was established by a programmer who volunteered to do it.

While my goal was to do real science, the process had to be entertaining if it was to be done by amateurs using their own time and their own resources. Therefore, I accept the fact that it would not be possible to “control” this effort. It has been designed from the start with a “laissez faire” structure. The plan was to ship out the telescopes to those that seem to want them, and to simply hope that they will try to pursue some resemblance of the group scientific goals with the apparatus. The objectives of this project were not tightly defined and it was hoped that goals would be generated as time and accumulated data define them. NOTE: Professionals out there, just try getting a grant with this approach!

The original goal was to search for comets. As time went on, however, more and more professionals showed up on the mailing list. They would ask questions like “as long as you are searching for comets, why not look for xxxx (their favorite object).” Thus, over time, I lost interest in looking for comets, which is very hard, and turned instead to frequently mapping the stars through multiple filters to look for new variable stars [5] and other such objects.

With the operations and data are open to anyone by design, there is always the possibility that someone will analyze data improperly and make erroneous publications. We consciously take that risk, and hope that the advantages of the open structure outweigh this possibility. Anyone can publish from the TASS data – true to the scientific paradigm, it is open and available to all. However – no one, to my knowledge, can publish as TASS without proper authorization.

At first, it was my plan to require 24 hour internet publication of data that was taken with the cameras which I had provided for free. This rule has not been enforced, indeed there is no way to enforce such a rule even were it so desired. An even greater reason for lack of enforcement is that we are drowning in data. In time we hope to correct this, at which time quick turn around will again be desirable, and we will encourage prompt reporting of all measurements.


Chris Albertson

It is my hypothesis that when a researcher acquires a large astronomical catalog he or she never intends to actually read it in its entirety, front to back. What the researcher wants is either a small subset or summary statistics. The conventional method, however, is to publish the complete catalog. The researcher physically acquires the requisite computer data file and processes it locally to produce the small subset or summary statistics that are actually desired. This model is not well suited to the present state of computing technology or to the needs of the TASS project.

Today raw computing power is cheap and readily available. However, high speed long distance data transport is expensive and rare. The TASS data will never be complete as the project has no plans to ever stop collecting data. The project, being cumulative, has the potential to outlast any one individual. So with luck, there may never be a “final TASS catalog”. The TASS system operates continuously, so any snapshot of the data would be obsolete as soon as it was made. Given this, it seems reasonable to put a computer with software that can accept any manner of ad-hoc requests for subset or summary statistics physically close to the TASS data. This way only a small amount of data – the data that is actually desired – has to be transported over today’s relatively slow network. The problem of data obsolescence is also solved because the transmitted data is always up to date.

The TASS Database is our implementation of these ideas. It is, in itself, a research project to determine how best to solve the above problems and many more which are discussed below. “Database,” as we use the term here, refers to an active object. It is a combination of software and tabular data. A database exists on a network and accepts electronic requests of the form ”From your tables X and Y, compute a table with columns a, b, c, … where the data meets criteria M, N, and L.” These requests can become quite complex. It is the job of the database system software to figure out how best to compute the requested table from the information the system holds internally.

The paragraphs that follows will look at:

  • the “information the system holds internally”,20
  • how they are maintained and created from a TASS camera and/or other data, and
  • what can be computed from these base tables.

Then we look at various methods of “wrapping” this databases system with a set of user interfaces, and finally, we will address problems and plans for the future of the TASS Database.

The TASS system as a whole, in its first year of operation, has placed more than 18 million photometric measurements into the database. This number is expected to grow to the hundreds of millions soon. Sever factors influence this expectation. All of the first generation cameras are yet to be put into operation, while a new generation of much more capable cameras are due to begin operation in a few months. There is also a backlog of unprocessed data “in the system”. It is quite likely that over the life of the project a billion or more photometric measurements will be entered into the database.

[Math to be add in support of these claims]

Sources for software:

Mark III Web sites:

Software programs

The programs that are used by most TASS Mark III sites to process their data for contribution to the TASS Mark III database, are listed below in order of sequential use. The authors and rewriters of these programs are named in parenthesis. Details on these programs can be found via the references in another section of this note.

  • tm3get.exe – (Droege/Gutzwiller) data acquisition program, collects image data. Produces image with approximate RA, dec of center of image in image FITS header.
  • dark.exe – (Gutzwiller) Process raw FITS image Produces dark vector FITS data file.
  • flat.exe – (Gutzwiller) Process raw FITS image. Also uses dark vector from “dark”. Produces flat vector data.
  • star.exe – (Gutzwiller) Processes raw FITS image. Also uses dark and flat data from “dark” and “flat”. Also uses reference star catalog: SAO, GSC, or Tycho. Produces corrected image FITS file. Also produces star list of RA, dec, magnitude. starpost.exe – (Gutzwiller) Processes a number of star lists. Also accepts some input pass/fail criteria. Produces two sorted subset star lists, pass and fail.
  • flatcomp.exe – (Gutzwiller) Processes a set of star lists. Also uses “good Tycho” star catalog as reference. Produces in pass 1 a correction vector for magnitudes. Produces in pass 2 a modified set of star lists based on correction. (Output from flatcomp is suitable for entry into the TASS database.)
  • collate.exe – (Arne/Gombert) Processes up to 3 star lists from “star” or “sextr actor”. Produces list of stars found in same location in more than one list.
  • trnsform.exe – (Arne/Gombert)Processes star list from “collate” Also uses a list of standard stars. Computes the V and V-I transformations using all matched standards. Produces an output file of transformed magnitudes.
  • master.exe – (Arne/Gombert) Processes star lists from “transform” and “collate” . Finds all objects with 3 or more observations in each filter. Produces a list of all stars that still have 3 or more observation s Also produces a file with two tables: (1) a histogrammed list of magnitude bin, mean error in that bin, and number of stars in that bin; (2) those objects whose magnitude variation is greater than ‘ average’.
  • diffcal.exe – (Arne/Gombert) Processes lists from “master”. Also processes star lists from “transform” and “collate” Compares star list items to Master items. Produces differential photometry record for each item to output file.