Folks,

every month it takes me a lot of time to check and archive the video
observations of the IMO network. For the October data alone, it was about 15
hours! Part of it is because the number of cameras is growing, but part of it is
also because the data quality of a few observers is not as good as it could be.
There are some observers (like EVAST, STRJO, STORO and YRJIL) where I can almost
blindly take over the data because there is hardly anything to correct, and
there are others where it takes me a lot of extra time. Each hour that I spend
processing the data is not spend improving the software or doing some shower
analysis!

In order not to explain everything again and again, I started to list the most
common mistakes here. Please, read these comments carefully and try to apply
these rules when you are processing your data. I have saved the file in the file
section of our akm-network yahoo group
(http://groups.yahoo.com/group/akm-network/files/), and will append it every now
and then.

Best wishes,
Sirko

a.. What to do if I need to restart the observation?
If you observed no meteor so far, just delete the logfile, the dbase files and
the data directory and restart the observation.

If you observed already some meteors, just delete the logfile yyyymmdd.log (a
copy of it is stored in your data directory) and the PosDat header file
mmddhead.pdf (it will be created again), and restart the observation.
At the end of the night, you edit the logfile in your data directory, which
contains both observations. Copy the start time of observation and the effective
observing time

start of observation : 2006/11/18 20:16:30
effective observing time : 1 h 46 m 41 s

from the end of the first observation to the end of the file. There you replace
the start time at the end of the second observation, and add the effective
observing time from the first part.
Last but not least you delete everything from

19:19:00 End of recognition request
19:19:00 Saving flatfield ... ok!
19:19:00 Saving sensitivity image ... ok!
Freeing resources and finishing ... ok!
...

of your first observation, until

...
Initializing PosDat files '1118data.dbf' and '1118head.dbf'... ok!
Creating recognition threshold history file 'metrec.thr' ... ok!
Creating auto-subdirectory '20061118' ... ok!
19:20:19 Initialize flatfield
19:20:30 Start recognition

of your second observation, and add two new lines

19:19:00 Suspend recognition
19:20:30 Restart recognition

instead. You end up with a file that looks as if you did not interrupt the
observation, but just suspended it. You copy the modified logfile into your data
directory and post-process it as usual. The meteor counts will be updated
automatically, when you delete false detections. If there are no false
detections, use the procedure described later to correct the meteor counts.

If you restarted the camera several times, you can repeat this procedure, i.e.
you take the start time of the first observation, add the effective observing
times, and replace the begin of all following observations by the two suspend
and restart lines.

a.. What to do if it was clouded at the begin of observation?
From false detections showing only clouds and from long gaps where no meteor was
detected you can conclude, that it was overcasted at some time of night. If it
was just at the beginning of the observation, you correct the time of the begin
of observation

20:16:19 Initialize flatfield
20:16:30 Start recognition

and the observing statistics

start of observation : 2006/11/18 20:16:30
end of observation : 2006/11/19 05:18:59
effective observing time : 1 h 46 m 41 s

by replacing the observation start time (correct also the date if skies cleared
after midnight!) and subtracting the time from the effective observing time.
If you had false detections in the time when it was clouded, do not delete them
by hand from the logfile, but use PostProc. Otherwise these false detections
will not be deleted from the PosDat files, and the images will also remain in
the data directory.

a.. What to do if it was clouded at the end of observation?
Similar to the previous point, you correct the time for the end of observation

05:19:00 Maximum recognition time reached!
05:19:00 Saving flatfield ... ok!
05:19:00 Saving sensitivity image ... ok!

and the recognition end time as well as the effective observing time in the
statistics section

end of observation : 2006/11/19 05:18:59
effective observing time : 1 h 46 m 41 s

Remember not to delete false detections by hand, but always using PostProc.

a.. What to do if it was clouded once or more during the observation?
No problem. Just enter two lines

01:00:00 Suspend recognition
02:00:00 Restart recognition

after you have deleted all false detections, and subtract the time difference
from the effective observing time

effective observing time : 1 h 46 m 41 s

a.. How do I know that it was clouded?
If you have false detections, it is usually quite easy to find out when it was
clouded. But sometimes it is just foggy or the limiting magnitude goes down
dramatically. What I usually do is to look for gaps, when there was no meteor
recorded for more than one hour. These times are marked as clouded as described
above. Of course, this is just a rule of thumb: If my image-intensified camera
does not record anything within 30 min, I mark these times as well, because it
usually records a meteor every few minutes. If, on the other hand, a
non-intensified camera has a gap of two hours in the evening hours of spring,
when there is hardly any meteor activity at all, and if the meteor before and
after the gap show perfectly clear skies, I accept also longer breaks. What I
definitely not want are logfiles giving an effective observing time of ten hours
with only two or three meteors detected!

There is one exception: Ilkka automatically detects clouded skies with an IR
sensor - these data are not corrected by me.
a.. How do I correct the begin time, end time, and effective observing time at
the end of the logfile?
You have to do it by hand (see above) - there is no automatic procedure to
correct these times.

a.. How do I correct the meteor counts at the end of the logfile?
If the counts are incorrect (e.g. because you joined two observations, but there
was no false meteor do delete), there is a simple workaround: Create an
"artifical" meteor by duplicationg the first meteor of your logfile

20:38:59 Meteor #1 at (0.742,0.431)->(0.747,0.366) frames=5 dur=0.22s pixel=5
dir=274ø vel=13.6ø/s å=1.6
shower=STA bright=3.1mag raddist=0.7ø exp vel=13.2ø/s (1.008h,74.62ø)
-> (0.402h,76.08ø) acc=0.7'
20:38:59 Saving meteor data of #1 ... ok!
20:38:59 Saving image band of 11 frames of meteor #1 ... ok!
20:38:59 Saving sum image of meteor #1 made of 11 frames ... ok!
20:39:00 Restart recognition
and correcting the time (e.g. to 19:38:59) of all five lines. Then you start
PostProc again and delete the artifical meteor (which will have no image,
because you recorded nothing at 19:38:59). When saving the logfile, PostProc
will recompute your meteor counts!
a.. What if the computer was shut down early and did not write the observing
statistics to the logfile?
Take an existing logfile from a previous night and append the last few lines to
your logfile:

05:19:00 Maximum recognition time reached!
05:19:00 Saving flatfield ... ok!
05:19:00 Saving sensitivity image ... ok!
Freeing resources and finishing ... ok!

Observing Statistics
====================
start of observation : 2006/11/18 20:16:30
end of observation : 2006/11/19 05:18:59
effective observing time : 1 h 46 m 41 s

# active meteor showers : 4
# observed meteor showers: 3
# observed meteors : 15

Sporadics (SPO) : ä=7, í=1.6 mag
N-Taurids (NTA) : ä=1, í=-2.5 mag
Leonids (LEO) : ä=4, í=-1.0 mag
S-Taurids (STA) : ä=3, í=1.1 mag
alpha-Monocerotids (AMO) : ä=0

Correct the time of the three lines of the end of observation, the start and end
date and time, and the effective observing time by hand. Run PostProc with the
option -shower to delete false meteors and recompute the meteor counts. If there
is no false meteor, use the trick described above.
a.. What to do with double meteors?
Sometimes MetRec detects two meteors at the same time. Since it cannot save
images for both meteors, PostProc will report "No image available". That does
not mean, however, that you have to delete the meteor without image. Just
remember the x/y-position of the meteor

21:36:05 Meteor #20 at (0.343,0.758)->(0.385,0.688) frames=19 dur=0.74s pixel=7
dir=301ø vel=5.0ø/s å=2.3
shower=SPO bright=1.8mag (1.894h,47.96ø) -> (1.765h,51.36ø) acc=2.4'

which is roughly 0.3 / 0.7 is this case, go back to the previous image, and
check, whether there is indeed a second meteor at the given position (whereas
0.0 / 0.0 is the lower left, and 1.0 / 1.0 the upper right corner). If that's
the case, you can keep the meteor without image. If not, delete it.

In the case, that the previous meteor was much earlier, you need to delete the
meteor without image as well. In this case there was a slow moving target (like
a satellite) in the background, which prevented from saving the image, but was
discarted later on because of its velocity.

a.. What to do if one meteor was detected twice?
That happens sometimes, especially with bright meteors. Please, keep only the
record that consists of more video frames, and delete the other one. I don't
care if you keep a copy of the image for your own records, but the meteor should
appear only once in the archive.
a.. What to do if MetRec obviously detected the meteor at the wrong position?
When you plot the line of the detected meteor position in PostProc you sometimes
see, that the begin or end of the meteor was misidentified (i.e. the line
deviates significantly from the true meteor position). That happens frequently
with bright meteors which burn our the image completely. Please, delete this
meteors! Even if you think it might be pity for the bright meteor, you should
remember that the position you measured is simply wrong, and from a PosDat entry
you do not see anymore that this was a measurement error. You could try to
correct the position by hand, but is it worth the
effort when we have 200,000+ other in the database?
a.. I'm not sure whether what MetRec detected was a meteor or something else?
Look at the meteor frame by frame by pressing "M" in PostProc. When the meteor
seems to fly around the corner, or when it is moving irregularly - delete it. If
you are not sure - delete it. Remember the archive size given above, and that
quality comes before quantity.
a.. I like to save Flash images. Is that ok?
Yes. But, please, do not send these flash images to me at the end of the month,
because they will not be archived!
a.. I moved meteors to the MISC directory. Is that ok?
Yes. But, please, do not send the MISC directory to me at the end of the month,
because it will not be archived!
a.. I run the camera automated. Do I still have to stick to the naming
conventions?
Yes. Remember, that each data directory should contain the following files when
you send it to me:

- yyyymmdd.cfg -> the configuration file
- yyyymmdd.log -> the post-processed logfile
- mmddhead.dbf -> the PosDat header file
- mmdddata.dbf -> the post-processed PosDat data file
- yyyymmdd.ref -> the reference star file (whereby yyyymmdd is not necessarily
the current night)
- yyyymmdd.bmp -> the reference star image, if a new reference star file was
created this night
- hhmmss.bmp -> the sum image for each meteor
- hhmmss.bnd -> the meteor band for each meteor
- hhmmss.inf -> the data file for each meteor

If you run your camera automatically, you may need to rename your generic
configuration file metrec.cfg and logfile metrec.log before copying them into
the data directory.

a.. It was a moonlit night with clouds and I got hundred of false detections.
Do I have do delete them all with PostProc?
Yes, since otherwise the meteor files will not be deleted, nor the PosDat
entries. To save time, however, you can delete more than one meteor at a time
with PostProc by pressing "D".

a.. How can I check if my reference star file is still valid?
To find large errors, look for a meteor that begins or ends close to a known
star, and check the position in a star catalog. It should closely resemble the
position of the meteor.
In general you should repeat the measurement of reference stars even for fixed
cameras every few months.
a.. My reference star file was invalid! Is my observation now worthless?
No. You should create a new reference star file at the next possible occasion,
and re-run PostProc with the -ref option. This will recompute the meteor
position from the new reference star file. You can verify that by comparing the
position given in the meteor data window at left (that's the corrected one) with
the position given in the logfile window below (that's the original position).
Beware that the -ref option introduces rounding errors, so it should only be
used to correct for mistakes.

a.. My computer time was incorrect (e.g. I forgot to switch from / to daylight
saving time). What to do?
Don't correct anything - just tell me when you send me the data. I have some
little scripts that can easily correct all the incorrect values (config file,
logfile, PosDat data file, meteor data files, meteor file names).


--
************************************************************
* Sirko Molau * *
* Abenstalstr. 13b * __ *
* D-84072 Seysdorf * " 2B v 2B " *
* Germany * *
* phone: +49-8752-869438 * Shakespeare *
* email: sirko@... * *
* www : www.molau.de * *
************************************************************


[Non-text portions of this message have been removed]