All,
One of the issues that is coming up in discussions of database ingest is
the number of objects detected per image, and I'd like to comment on
that. First of all, we are considering here detections in single
images, as will be performed at base camp for generating alerts, not
detections on stacked images. Here are a few basic numbers:
1. Looking toward the galactic pole, there are roughly 5000 stars per
square degree brighter than V=25, which is roughly our single image limit.
2. Looking toward the galactic center, there are roughly 1000000 stars
per square degree brighter than V=25. If we look toward the galactic
anti-center, but still in the galactic plane, the number is about 50000
stars per square degree
3. There are roughly 100000 galaxies per square degree brighter than V=25
4. Looking in the plane of the ecliptic, there are about 300 main belt
asteroids per square degree to V=24.
5. Every galaxy has supernovae occurring about once per 200 years, and
they are detectable for roughly 20 days. Both these numbers are very
rough, with the rate being appropriate for intrinsically bright
galaxies, and the detection time number being about right for faint
galaxies near the detection limit. If we go with these numbers, then
there is a probability of 20/(200*365)=3e-4 that a given galaxy has a
supernova detectable in one image.
One can take these basic numbers and come up with very different numbers
of objects detected in a single LSST image of about 10 square degrees.
The total number of detectable objects in a single image ranges from
about 10^6 to about 10^7 depending on where we are looking relative to
the galactic plane. If we are looking well away from the plane, these
objects are nearly all galaxies, while if we are looking toward the
plane, or especially the center of the galaxy, stars dominate.
In alert mode at base camp, however, our current plan is that we are
operating on difference images, and detecting only objects which:
1. Vary in brightness significantly from the template image, or
2. Are on a list of objects to photometer all the time.
Note that category (1) includes not only real variable and moving
objects, but also false detections due to noise. We can get a
reasonable(?) estimate of a typical number of detected objects in alert
mode by asssuming:
0. We are looking away from the galactic plane, but near the plane of
the ecliptic
1. About 5% of stars vary detectably from the template
2. All the supernovae and moving solar system objects are detected
3. Detection threshold is set so that we detect about 50% noise and 50%
real objects
4. The "always photometer" object list in (2) above is empty.
These assumptions result in 2500 variable stars, 3000 asteroids, 300
supernovae = 5800 real objects, or 12000 objects including noise.
The number would be reduced by nearly 2x if we are looking well away
from the ecliptic, will be increased by about 4-8x if we are looking in
the galactic plane (but away from the center), and could grow extremely
large if the "always photometer" list includes all the objects on the
sky. I'd say we need to design for 10000 objects as a typical case, and
about 50000 objects as a less frequent (10% of the time) case. These
are for the full LSST, DC1 will be scaled down from this in a manner
still TBD.
Please take a look at my assumptions and identify any errors (arithmetic
or astronomical) I've made!
Tim
begin:vcard
fn:Tim Axelrod
n:Axelrod;Tim
org:;Large Synoptic Survey Telescope
email;internet:[EMAIL PROTECTED]
title:Data Management Project Scientist
tel;work:520-322-8735
version:2.1
end:vcard
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
