Some information concerning data from the All Sky Automated Survey


Content:
1. Data taking
- Concept of Fields
2. Data reduction
- including taking darks, flats and applying other corrections.
3. Photometry
- aperture photometry; five apertures (2...6 pixels wide)
4. Astrometry
- ACT relative; accurate to 3-5 arcsec
5. Catalog
- Post-flat-fielding; 32x32 smooth "glue function"
6. Data retrieving
- Concept of Datasets
7. WWW interface
- plots

1. Data taking

The whole ASAS3 sky is divided into 709 fields named according to coordinates and using sequential number; e.g.
F0552-16_254 is Field number 254 centered on R.A. 05h52m00s and DEC -16o00'00"
Fields are 2048 x 2048 pixels (8.5o x 8.5o deg) each, so they overlap slightly. Telescope pointing is such that consecutive images of the same fields are shifted by tens or (sometimes) hundreds of pixels.
In ASAS3 we were taking three consecutive 1-minute exposures of the same field (instead of one 3-minute exposure).

2. Data reduction

Includes the following steps:
- taking DARKS.
- taking FLATS, (but same flats are used for very long time).
- subtracting darks, applying shift-reject-add procedure (to avoid planes, cosmics, etc.).
- cosmetic correction for bleeding.

3. Aperture photometry

- currently 5 apertures are used: 2,3,4,5 and 6 pixels wide.
- this gives better results then profile (due to time/space varying PSF).
- magnitudes are listed as MAG_0 (smallest, 2 pixels wide) ... MAG_4 (largest, 4 pixels wide).
- small apertures are better (consider S/N) for faint stars, large ones for bright stars.
- rule of thumb: is aperture_index=12 - V (MAG_0 for 12m and fainter; MAG_4 for fainter then 9m )
- be aware of Blending effects:
- if blended stars are closer then 15 arcsec:
--- they are not resolved at all!
- if separated by more then 15 arsec:
--- brightness usually increases when going from MAG_0 towards MAG_4.
--- small apertures may give magnitudes closer to "standard system".
--- small apertures may induce additional scatter (due to inclusion of the part of neighbor's flux).
--- best aperture may change in time (variable PSF, variable stars).
--- there are more effects to consider!

4. Astrometry

- currently done using ACT catalog.
- distortions included up to (x,y)3
- typical RMS 0.2 - 0.3 of the pixel = 3 - 5 arcsec.

5. Catalog

- ASAS catalog is maintained for each Field separately.
- each sub-catalog is first seeded using external catalog stars
- in V external catalog stars come from Tycho (using V=VT-0.09 BT calibration).
- resolution of the ASAS catalog is 15 arcsec.
- ASAS v magnitudes are converted to Johnson's V using V=v+const calibration only (no color correction).
- matches between v and external catalog stars are done separately for each 64x64 pixel subregion on the 32x32 grid.
- smooth matching glue is then applied to all v measurements: V=v+glue(x,y), separately for each aperture.
- Measurements are added to the already existing entries int the ASAS catalog, or new catalog entries are created.

6. Data retrieving

- for each Field within field-radius (14 deg):
--- subcatalog is searched for entries within object-radius 25 arcsec (default value) of the required position.
--- for each entry found - all measurements are listed.
--- Thus ASAS data is divided into datasets; e.g.:
#dataset= 1 ; 2 F0552-16_254
--- where:
------ first number (1 in above example) is sequential dataset number.
------ second number (2) is sequential entry number in subcatalog.
------ F... is Field (subcatalog) name.
- due to overlap up to 5 subcatalogs may contribute to data.
- several entries for the same star may exist in each subcatalog:
--- for faint stars - due to astrometric errors,
--- for bright ones - due to saturation,
--- also due to varying PSF, focus and tracking problems, etc.)
- "Grades" at the end of each data line correspond to the average quality of the frame. A - for the best frames, B - for worse, C for missing data.
- A and B grades are determined statistically, looking at distribution of ERR_x's
- also ERR_x are not individual photometric errors, but average for the frame.
- due to flatfielding and PSF problems same stars may differ in brightness in different subcatalogs.
- ergo: ASAS never pretended to be on any standard system.

7. WWW interface

- different colors on the plots correspond to different datasets.
- when searching for the object the object-radius can be specified;
- however: when plotting and retrieving data default radius of 25 arcsec is used again!
- you can extract exactly what you need using datasets.

Some older answers, extracted from my e-mail log

Due to problems with PSF we have decided to do aperture photometry only. Separate datasets come from overlapping areas of the separate fields on which ASAS cameras are centered. You can have up to four datasets. Five columns correspond to five apertures used for aperture photometry (with diameters 1 2 3 4 5 pixels). Usually for bright stars largest aperture (fifth column) gives smallest dispersion and for fainter stars smaller apertures are better. Usually all five measurements should be close to each other but in case of close companions this is usually not true.
Now "doubled" light curve.
These are due to differences between stars' brightness in different sets.
There are two reasons of that:
1) Flat-filed errors
2) PSF errors
In the first case you will see similar amount of offset in each aperture. In the second case the difference will disappear with larger apertures.
Five entries are due to the catalog structure and some astrometric errors:
Catalog is divided into "fields", and measurements are tied to the individual field catalogs. If there is an overlap - you may get several "datasets" of data (they are numbered as #dataset=1...,etc in the returned file). They are separated, since ASAS data are NOT ALWAYS properly calibrated and some photometric offsets within a field due to the improper flatfielding exist.
In some (?) overlapping areas magnitudes may differ by as much as 0.1 mag. On the other hand there are situations, when larger then usual astrometric errors occur - or alternatively coordinates are such, that cataloging software creates two separate entries in the catalog. In such cases it is better to list all possible entries around requested position, This can be recognized by a number after a semicolon, e.g.:
#dataset= 1 ; 1 F0552-16_254
string F0552-16_254 denotes field-name. In fact, if you click on ANY of the listed items you'll probably get photometry from all listed items (if you use default radius of 15 arcsec), because the search radius for photometry of a given star is set internally to 25 arcsec.
Final decision which data to can be used, which should be rejected belongs unfortunately to the user.
The magnitude printed in the lower left panel is for orientation only - it is so called "reference magnitude" - i.e. the first measurement recorded for the star in the smallest aperture.

First let me explain apertures: We get photometry with 5 different ones: with diameter of 2 pixels (MAG_0), 3,4,5 and 6 (MAG_4).
The best (S/N) you can achieve depends obviously on the stellar brightness and aperture diameter. For bright stars one should use wide apertures (to include more photons) for faint ones - small apertures (to avoid sky noise). My rule of thumb is
for magnitudes fainter then 12 use aperture_number = 0 (MAG_0)
for magnitudes brighter then 9 use aperture_number = 4 (MAG_4)
otherwise use:
aperture_number = 12 - magnitude.
This is usually a good choice, but can fail e.g. for blended stars. In close blends you'll include more companion light with larger aperture, so there is a trade-off between S/N and contamination. Usually it's better to use smaller apertures in such cases, or at least be very cautions. (One can easily detect blends comparing MAG_0...MAG_4 - MAG_(X+1) < MAG_X (for X=0,3) usually means blending)
If you use MAG_4 as default you are more prone to blending effects.
As far as periods are concerned I must admit that obviously there are some errors in our periods, but not too many. Some periods may be wrong due to shorter base-line of observations at the time of determination, some may accompany missclassification.
In any case if you have problems with individual stars, let me know details (ASAS-ID, period(s)). I'll try to look at the data myself to locate problems. Different datasets come from:
a) different fields. ASAS fields do overlap, so you can have as much as 4 datasets for each object (if it sits in the corner of the frame). Due to problems with flatfielding it is possible, that datasets do not match precisely. They can be offseted by as much as 0.1-0.2 mag. In most cases offset is minimal and indeed you can just combine datasets. If, however, you notice a "dual" or "dispersed" nature of your light curve it may suggest offset between datasets. In such case it is best to find this offset individually (e.g. comparing magnitudes of the datasets in several bins across the phased light curve) and apply them.
b) Native pixel size of ASAS catalog is 15 arcsec. Sometimes it happens, that one star has 2 entries in the catalog separated by more than 15". Thus any search for photometry includes all positions within 25 arseconds of the given coordinates. This means, that for any field you can have more than one dataset listed (sequential numbers within a field are given after semicolon). Sometimes however, there could be just another star within 25", in which case you'll get mixed data of two stars.
This should be easily noticed as doubled light curve.
Unfortunately there is no way to properly sort out these things automatically, and I cannot do that manually for 20,000,000 stars, so the only thing I can do is to provide access to largest possible set of data and let user do selection.

Our transformation is simply "V=v+constant", where constant is median form relation "V_Tycho = v + constant" for known Tycho stars.
No color terms are included.
As far as "I" transformation is concerned we tried to calibrate "(V-I) vs. (V-J)" relation using known standards (Landold and others), 2MASS and Tycho (inspired by:"http://www-int.stsci.edu/~inr/cmd.html Photometric properties of low-mass stars and brown dwarfs" ). Having this relation we have used I = V - (V-I) = V - f(V-J) calculated for Tycho stars to create our our reference system. Then, simple transformation I= i + const was applied.
Based on multiple checks on standard stars we think that our "standard" "V" system is accurate _ON_AVERAGE_ to about 0.05 and "I" to 0.15 mag.
Please note also, that there is blending in our data, which may cause any star with close companion to deviate by any amount from the true V or I value (this also affects amplitudes of variables). And there are some unsolved flat-fielding problems, that in several cases result in different average magnitudes (by as much as 0.1) of the same stars observed in the overlapping areas of the neighboring fields. As you see, ASAS was never intended to be a source of "standard" magnitudes, however we try to keep it close _on_average_ to such system.

which indeed came from the same frame, but refer to different centroids - as indicated by the number following the semicolon in "#dataset". The reason is that "find" algorithm in the photometry package can distinguish blended objects even if they are 1.5-2 pixels (>20 arcsec) from each other. This however depends on the quality of individual frames. Catalog software has fixed resolution of 15 arcsec.
RMS astrometric quality is around 0.2 pixel (3-5 arcsec), but individual errors could be larger - so search radius when requesting data is larger - usually 25 arcsec.
So - if it happens that on a given frame there are two close brightness maxima (real or e.g. due to a cosmic ray) they will be entered into the catalog as two objects, and retrieving them may give you two entries with the same HJD and slightly different coordinates.
Since this entries have different centroid position, photometry (especially in small apertures) is affected.
The original reason for inclusion of these secondary entries was for faint objects with larger astrometric errors - it is not always obvious, that measurement closest to the average position is the true one.

Shortly speaking:
MAG_0, ..., MAG_4 correspond to aperture photometry obtained with 2,3,4,5 and 6 pixels wide circular apertures. MAG_0 is usually best for faint stars, MAG_4 for the brightest. The rule of thumb is to use "12-mag" index, i.e. MAG_0 for 12 and fainter and MAG_4 for 9-th and brighter, This is to avoid excessive noise from background for fainter stars. Large apertures often catch blends. MAG_0... MAG_4 are calibrated in such a way, that they have the same median for all stars on the frame. Blending reveals itself as brightening with increased aperture.

"Grades" at the end correspond to the average quality of the frame. A- for the best frames, B - for worse, C for missing data. Also ERR_x are not individual photometric errors, but average for the frame.
"Datasets" come from different frames (overlapping areas). I do keep distinction, since sometimes flat-fielding errors may show up as systematic differences between "datasets". One must take care about that.
----------------------------------