Starship Asterisk*

Lior, this derivation is very nice. It is true that 3D trackes of meteors are analyzed elsewhere, but these are taken by an instrument with the potential of accurate photometry. Thus from your derivation it may be possible to associate the light production by the meteor with a specific path trough the atmosphere. The only problem is that there is no way of estimating the speed of the meteor. Other places, notably the European Fireball Network, use rotating choppers to cause meteors to show an interrupted trail. As the rate of chopping is constant and is well known, this shows immediately what was the projected angular velocity of the meteor. From this, and from a 3D analysis as you did, one can find everything about the meteor, including its orbit around the Sun. This IS VERY USEFUL, because only rather few meteors have orbits. Even more so, because of sensitivity issues the CONCAMs see the brighter meteors. These have the potential of dropping meteorites and later on can be recovered.

Noah Brosch

Here is a link where the kind of data that can be derived from proper two-station meteor observations is described:

http://www.asu.cas.cz/english/new/EN060402.html

Noah Brosch

And here is another link where a more detailed description of the network is given: http://www.molau.de/meteore/imc97-2.html. Note that the potential of on-line detection, measurement, and trajectory calculation has not been realized yet.

Noah Brosch

Thank you Dr. Brosch,
I could open the first link, while the second one gave me a blank page for some reason.
I noticed the puplications of the European fireball network at ADS. The data that they provide is very similar, except from the luminance. We currently do not have the ability to measure the true brightness of objects, and that is something that can weaken the analysis.

In their publications I looked for the way they perform the analysis, but couldn't find any reference to that. They just provide the numbers. I am most interested in the accuracy of their computation, as I don't believe that 100% of accuracy can be achieved. I'm pretty sure that the numbers above are accurate with an error of less than 1%. The error for the Irridium staelite was around 5%, and the error of HST was no more than 1 or 2 percents. These two objects are more difficult to estimate since the differences between the PSFs are just few pixels. Objects which are lower in magnitude have a much larger "wingspan", and therefore a single pixel error can not significantly affect the results.

Lior, the European network uses film cameras, and large format negatives, for their fisheye images. These have many more pixels than any CCD camera has. The pixels are extremely fine, perhaps a micron or so in size. This determines the astrometric accuracy.

Regarding the method, perhaps this Nature paper has it:
http://www.nature.com/cgi-taf/DynaPage. ... ynoptions=

If the link does not work, just look up vol. 423 (no. 6936), a paper describing this Neuschwanstein meteorite fall. Regarding the new model meteor stations, developed by the Czechs, you may want to look up whatever the Brits and the Australians are doing. This is at https://www3.imperial.ac.uk/portal/page ... PORTALLIVE. Another description is at http://www.blackwell-synergy.com/links/ ... 20.x/full/, which is Astronomy & Geophysics Volume 45 Issue 5 Page 5.20 - October 2004

Cheers,
Noah Brosch

These papers require some registration. I'll try to find them in the library.

A thought on error analysis. As we are all aware, the importance of a positional determination of a meteor is determined by the accuracy of the determination. If we were to predict that a meteor fell exactly into the trunk of my car, and no meteor is found in the trunk, how big a search will be needed to find the meteor? Should we look around the car, the surrounding cars, the whole parking lot, the campus, the town, the state, or the whole continent? Only an accurate estimate of the error could tell us. Now we may or may not be interested in meteor fall locations, but even the importance of orbit deteminations are based on accuracy and so of crucial importance.

So here are two ideas of how to get a quick and dirty accuracy on poisition including height.

1. Use the midpoint of the meteor streak. The midpoint should be clearly defined on the images of the meteor from both stations. It should also be invarient to projection effects. Once the midpoint is determined, compute meteor height and position with the midpoint mascarading as an endpoint (twice), and compare these two determinations with the determination using the two real endpoints. Curvature in the meteor trail might make the lower half of the determination significantly different than the upper half.

2. Assume a one pixel error in the determination of an endpoint and re-compute the height and position with this one-pixel difference. Do this several times, randomly offsetting a meteor endpoint a single pixel in a random direction, to get an ensemble and then use the standard deviation of this ensemble as an error estimate.

- RJN

Can meteor velocity be estimated?

First I didn't mention before that Lior's 3D meteor determination is really very impressive. Next, thanks to Noah for you comments and informing us of those references. I put a lot of thought last night into Noah's comment on the need for a CONCAM light chopper to determine meteor velocity. I agree that such a chopper would be useful and give a first hand determination of meteor velocity and acceleration.

Given our CONCAMs in the field, however, there might be some way to go forward with a velocity estimation. CONCAMs give accurate photometric determinations that are not available for those tracking fireballs on film. This might be useful.

The first position when the fireball becomes visible appears to be to be a function of meteor cross section, air density, meteor composition, meteor mass, and meteor velocity.

Following this logic, the brightness of the meteor at any position is also a function of these parameters. It might be possible to create a Monte Carlo simulation of an incoming meteor and try to simulate the light curve seen from the two stations simultaneously, assuming spherical emission. To do this meteor velocity might be an important parameter, possibly uniquely determined in a fit to the meteor brightness at all of the positions simultaneously. If so, we can recover velocity to some accuracy.

- RJN

Bob, thanks for your comments. As usual, nature is perverse in complicating things. Essentially, there are two kinds of meteors (or models of meteors). One assumes that the meteoroid (the incoming body) is a single entity, what would normally be called "a piece of rock". The other, that is becoming more popular because it explains many observational things, is that the meteoroid is a collection of small grains held together by glue. This model is callted, as one might expect, the "glueball". The assumption is that the grains are refractory material, with a melting temperature of 1500 degrees or higher, while the glue melts at a few 100 degrees. From here on we go into the wild guesses because for the glue, for example, I have seen proposals ranging from water ice, through organic materials, to sodium.

Glueball meteoroids produce wide radar echoes when detected at ~100 km because the grains spread out from the small incoming body upon the glue melting (probably above 200 km). They also produce jumps in the meteor brightness, "flares", which would mess up the determination of the photometric mid-point.

In order to be consistent, one would have to define an isophotal beginning and ending of the meteor trail and find the mid-point. But as we are talking about a phenomenon that is a few tens of km long, and is at 100 km or so away, issues of relative distance of parts of the trail begin to be important. One thing we should agree upon: if the meteor trail has a point that can be defined as the brightest, the "peak" of the trail light curve, this will be irrespective of the viewing location. In this case, the fiducial point to be used in the quick and dirty analysis would be unique.

Noah Brosch

Lior,
This calculations looks very promising. This would be really helpful in finding atleast approximate location of a meteorite fall on the earth. This would be another vantage point for NSL.

Did we hear anything about the "cosmic trail", detected in one of the dark frame?

Tilvi, it is very unlikely that the present CONCAMs will locate a meteorite. The main reason is that the only pair now in operation is in Hawaii, where a fall would most likely end up in the ocean. The Australians are putting up a camera network in their desert, where meteorites will stay on land, and are targeting their measurements so that the error ellipse of the fall would be lass than one mile. Best places to look for meteorites are, by the way, in the Arctic and the Antractic. The meteorites that are black stand up on th white ice...

Noah Brosch

Tilvi, the software already gives the estimted location of the meteor on the ground (had it made it that far). The longitude and latitude of the estimated hit are listed above. I assume that the accuracy is around 1 km, close to the 1 mile accuracy mentioned by Dr. Brosch.

As Dr. Brosch also mentioned, we will be able to collect the meteors only when they land on solid ground. In Hawaii they will most likely hit the ocean.

I liked the idea Monte Carlo simulation modelling. If we can tell by the light curve which type of meteor is which, then this might work. We still don't have enough data, but this can be changed with time.

Dr. Brosch, we still have the extra simple CCD at Wise. When it is installed, we can do the same thing at Wise. I guess it won't be usefull for collecting meteorites since the Negev or the Jordanian desert are not really areas where meteorites can be easily found.