As many readers are aware, Dr. Tancredi is working on automatic software detections of meteor trails on CONCAM images. WOLF, the software written by Lior Shamir and running at all CONCAM sites, does flag 'novelty' images where something unexpected has appeared in a frame. These two can be used beneficially during the upcoming Perseid meteor shower which peaks on August 11.
The predictions from this year's shower indicate a possibility of two maxima, one from ejecta from the parent comet at its last perihelion passage and another being the regular, broad Perseid peak. This year, the Persieds are going to look good, because New Moon would be just one week after the Perseid peak. The Moon will, therefore, be rather thin and will not interfere destructively with the meteors. The predicted ZHR (Zenithal Hourly Rate) is 50200, which could be quite impressive for naked eye viewing!
I propose that , as done in previous years for the Leonids, we use a 'meteor setup' for this year's Perseids. In this setup we would have shorter exposures that would be repeated more often, so that we could better characterize the hourly rate of bright meteors.
FYI, we are going to have a public event in the vicinity of the Wise Observatory, with public viewing of meteors on the night of the maximum (1112 August). We were promised that the nearby settlement of Mizpe Ramon would turn off its street lights to allow for even better viewing; in this case, you would not see the light halo on our eastern horizon. This may happen also on the second night, 1213 August, as well.
Cheers,
Noah Brosch[/u][/b]
Perseids 2004

 Ensign
 Posts: 78
 Joined: Tue Jul 27, 2004 1:45 pm
 Location: Back at Tel Aviv University after a sabbatical

 Baffled Boffin
 Posts: 1574
 Joined: Sat Jul 24, 2004 1:58 pm
 Location: Michigan Tech
Noah,we would have shorter exposures that would be repeated more often, so that we could better characterize the hourly rate of bright meteors.
It is of interest to be able to calculate the "best" exposure duration for CONCAMs for meteors. If the goal is to sample quickly compared to an hour, then I think the current duration of 3 minutes with time between exposures of 3m56s is already fast enough. If the goal, however, is to find the most meteors, then the situation becomes more complicated. This question that was addressed generally in my "Tile or Stare?" (ToS) paper last year in the Astronomical Journal. An ADS link to it is here
ToS will tell you that the frame rate that discovers the most transients is usually the duration of the transient. In the case of meteors, this is about a second. Frame rates near one second aren't good for CONCAMs for two reasons. First, the frame readout time is about 35 seconds, so that repeated exposures of 1 second mean that most time is spend transferring data and not exposing to the sky. ToS has a formula that computes optimal rates including readout times, though.
The second reason is that ToS assumes a new field every exposure, and fisheye cameras like CONCAMs have essentially the same field per exposure. This will change things.
As ToS explains, the real answer depends on the luminosity function of the meteors and the brightness of the background, too. I think it would be really good for a student (for example) to go through this for wideangle cameras and meteors in general, and CONCAMs and Perseids specifically. This, in my opinion, would be an easy but useful little paper that could be vetted here but appear in a refereed journal!
 RJN

 Baffled Boffin
 Posts: 1574
 Joined: Sat Jul 24, 2004 1:58 pm
 Location: Michigan Tech
I have thought a little more about this and even gone back and looked at the Tile vs. Stare paper. It seems to me that most detected meteors are well above background, so that the background parameter beta is near 1 for CONCAM meteor observations. This is because the typical sky background for CONCAM is about 8th magnitude in a single pixel.
Now I seem to recall that the integral luminosity function for many meteor showers is about alpha = 1.5. If so, then alpha x beta is 1.5. So the integral number of new meteors discovered will be governed by N_meteors ~ t_exposure^1.5. In this case the best way to discover the most meteors is to stare as long as possible. This indicates that perhaps we should consider CONCAM exposures longer than 3 minutes instead of shorter.
This logic breaks down for meteors so dim that the background becomes important. At that limit, beta = 0.5. assuming that the same meteor luminosity function works for these dim meteors, then N_meteors ~ t_exposure^(alpha beta) ~ t_exposure^(0.75). Since this exponent is less than 1, this indicates that tiling will recover more meteors than staring.
The limit of tiling is to either infinitely fast exposures or a return rate at the duration of the transient (see ToS paper). But these short exposures mean that readout time is important and this must be incorporated. Equation (19) of ToS then indicates that the most productive exposure time is t_exposure = (alpha beta t_readout) / (1  alpha beta). Given alpha beta = 0.75 and t_readout = 35 seconds, the exposure time that will yield the most meteors is about 105 seconds. To get a more accurate number we would have to more precisely measure the readout time. The largest uncertainly is possibly in the luminosity function index alpha, though.
Thoughts?
 RJN
Now I seem to recall that the integral luminosity function for many meteor showers is about alpha = 1.5. If so, then alpha x beta is 1.5. So the integral number of new meteors discovered will be governed by N_meteors ~ t_exposure^1.5. In this case the best way to discover the most meteors is to stare as long as possible. This indicates that perhaps we should consider CONCAM exposures longer than 3 minutes instead of shorter.
This logic breaks down for meteors so dim that the background becomes important. At that limit, beta = 0.5. assuming that the same meteor luminosity function works for these dim meteors, then N_meteors ~ t_exposure^(alpha beta) ~ t_exposure^(0.75). Since this exponent is less than 1, this indicates that tiling will recover more meteors than staring.
The limit of tiling is to either infinitely fast exposures or a return rate at the duration of the transient (see ToS paper). But these short exposures mean that readout time is important and this must be incorporated. Equation (19) of ToS then indicates that the most productive exposure time is t_exposure = (alpha beta t_readout) / (1  alpha beta). Given alpha beta = 0.75 and t_readout = 35 seconds, the exposure time that will yield the most meteors is about 105 seconds. To get a more accurate number we would have to more precisely measure the readout time. The largest uncertainly is possibly in the luminosity function index alpha, though.
Thoughts?
 RJN

 Ensign
 Posts: 99
 Joined: Mon Jul 26, 2004 8:55 pm
 Location: Michigan Tech

 Baffled Boffin
 Posts: 1574
 Joined: Sat Jul 24, 2004 1:58 pm
 Location: Michigan Tech
Thanks, Dan. Interesting point.
Meteors only take a second to zip on by. Therefore longer exposures should have no effect on meteors. They will still look like streaks in the CONCAM frame. A longer exposure just means that there will be more streaks.
Longer exposures than 3 minutes will cause stars and planets to trail even more than they already do. This will also cause the loss of reliable photometry data as it will not be simple to compare it to 3 minute exposures photometrically. Routine opacity maps will not be able to be computed for this reason.
 RJN
Meteors only take a second to zip on by. Therefore longer exposures should have no effect on meteors. They will still look like streaks in the CONCAM frame. A longer exposure just means that there will be more streaks.
Longer exposures than 3 minutes will cause stars and planets to trail even more than they already do. This will also cause the loss of reliable photometry data as it will not be simple to compare it to 3 minute exposures photometrically. Routine opacity maps will not be able to be computed for this reason.
 RJN

 Ensign
 Posts: 78
 Joined: Tue Jul 27, 2004 1:45 pm
 Location: Back at Tel Aviv University after a sabbatical
Meteors
Bob et al:
Here is the magnitude distribution for meteors observed by us with intensified video cameras during the 2002 Leonid shower:
Magnitude bin Number of meteors
4...3 n=1
3...2 n=3
2...1 n=3
1...0 n=5
0...1 n=6
1...2 n=6
2...3 n=20
3...4 n=23
4...5 n=25
5...6 n=12
6...7 n=8
7...8 n=1
I did a cumulative distribution and plotted a regression for all the bins of log(N) vs. magnitude; the slope comes out as 0.17+/0.02.
We could assume that the Perseids are very similar to the Leonids and that this is what we'll see next week.
I think that your argument that we hould go to twominute integrations before readout is possibly what needs to be done. At the same time, perhaps it would be better to reduce the number of darks taken after 23:00 LT to increase the meteor observing efficiency.
Cheers,
Noah
Here is the magnitude distribution for meteors observed by us with intensified video cameras during the 2002 Leonid shower:
Magnitude bin Number of meteors
4...3 n=1
3...2 n=3
2...1 n=3
1...0 n=5
0...1 n=6
1...2 n=6
2...3 n=20
3...4 n=23
4...5 n=25
5...6 n=12
6...7 n=8
7...8 n=1
I did a cumulative distribution and plotted a regression for all the bins of log(N) vs. magnitude; the slope comes out as 0.17+/0.02.
We could assume that the Perseids are very similar to the Leonids and that this is what we'll see next week.
I think that your argument that we hould go to twominute integrations before readout is possibly what needs to be done. At the same time, perhaps it would be better to reduce the number of darks taken after 23:00 LT to increase the meteor observing efficiency.
Cheers,
Noah

 Baffled Boffin
 Posts: 1574
 Joined: Sat Jul 24, 2004 1:58 pm
 Location: Michigan Tech
Hi Noah,
I just input your meteor numbers into a handy Excel spreadsheet and computed alpha for every magnitude bin. For the whole sample I got an alpha of 0.46 but it was rising steadily for dimmer and dimmer meteors. alpha was about 0.65 at about magnitude 4 where the differential counts started to drop off. So I think 0.65 +/ 0.2 is a working estimate for alpha. (This number should be checked.)
Assuming beta = 1 in this range, then it is defintely better to tile than stare. The shorter the exposure the better as the rising background washes out more dim meteors than bright meteors are being discovered. But readout time will limit how short we can go. So with (alpha beta ~ 0.65) and given a read time of 35 sec, the exposure time that nets the most meteors is (ToS Eq. 19): t_exposure = 65 seconds (+133 36). So the uncertainty in alpha leads to a very large difference in the optimal exposure time. Therefore, to twosigma, any CONCAM exposure time might be optimal.
Thoughts?
 RJN
I just input your meteor numbers into a handy Excel spreadsheet and computed alpha for every magnitude bin. For the whole sample I got an alpha of 0.46 but it was rising steadily for dimmer and dimmer meteors. alpha was about 0.65 at about magnitude 4 where the differential counts started to drop off. So I think 0.65 +/ 0.2 is a working estimate for alpha. (This number should be checked.)
Assuming beta = 1 in this range, then it is defintely better to tile than stare. The shorter the exposure the better as the rising background washes out more dim meteors than bright meteors are being discovered. But readout time will limit how short we can go. So with (alpha beta ~ 0.65) and given a read time of 35 sec, the exposure time that nets the most meteors is (ToS Eq. 19): t_exposure = 65 seconds (+133 36). So the uncertainty in alpha leads to a very large difference in the optimal exposure time. Therefore, to twosigma, any CONCAM exposure time might be optimal.
Thoughts?
 RJN