Why are meteors sometimes missed - markmac99/ukmon-pitools GitHub Wiki
Sometimes, you may see a meteor on the timelapse that doesn't appear in the RMS data file. Or a meteor may be detected but not matched with other cameras. Why is that?
RMS requires at least 20 stars to be detected in the field of view otherwise it will ignore the detection completely. This is so that it can perform astrometry (to work out where precisely the meteor is) and photometry (to work out how bright it was, compared to the stars).
Bright moonlight, cloud, haze, fog or poor focus can cause RMS to see too few stars.
Take a look at the log file using the unix "grep" and "more" commands:
grep "stars:" logfilename | moreFor example
(vRMS) pi@uk001l:~/RMS_data/logs $ grep stars log_UK001L_20220813_083131.519539.log | more
2022/10/21 21:39:04-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 7
2022/10/21 21:39:14-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 7
2022/10/21 21:39:26-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 9
2022/10/21 21:39:36-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 14
2022/10/21 21:39:47-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 13
2022/10/21 21:39:57-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 20
2022/10/21 21:40:08-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 23
2022/10/21 21:40:18-INFO-DetectStarsAndMeteors-line:89 - Detected stars: 28
Its normal to see few stars detected at dawn and dusk, but here you can see that before 21:39:57, RMS could only find 14 stars or fewer and so it would not be able to search for meteors.
RMS needs to know where it is looking. The platepar is its starting point, but for each detection RMS will adjust the platepar to take account of any small movements caused by wind, temperature, etc. However, RMS can only automatically make _small_ adjustments. If the platepar is too wrong then the process fails. By the same token, if RMS can't match enough stars while recalibrating, then the process fails.
Look for a an image whose name ends in "calib_report_astrometry.jpg". This shows how well your camera is calibrated. On a typical night, the image should show a random scatter of yellow circles, each with a small stick pointing randomly. If all the sticks point in one direction, or form a pattern, or all point radially, and if the sticks are long, then calibration is bad and should be redone. Examples of five bad and one good platepar are shown in the image below.
You can also use the unix "grep" command to inspect the platepars_all file:
grep auto_recalibrated platepars_all_recalibrated.json
If you don't see at least some "true" messages and the sky was clear at least some of the night, then the platepar is bad and should be checked.(vRMS) pi@uk001l:~/RMS_data/ArchivedFiles/UK001L_20220813_201026_178620 $ grep auto_recalibrated platepars_all_recalibrated.json
"auto_recalibrated": true,
"auto_recalibrated": true,
"auto_recalibrated": false,
RMS can have problems with super-bright events. For more about this, see Why are Fireballs Missed
The solver that matches events can only use Calibrated data. If RMS has been unable to match the stars, the data will be marked as not recalibrated and will not feed the matching engine.
Look at the Calibration report on the Archive website: https://archive.ukmeteors.co.uk/reports/lastlog.html
As noted above, RMS performs photometry and astrometry on each individual detection, to work out where it is and how bright it is.
First RMS maps the visible stars onto the theoretical starfield, and from this it can work out the meteor's trajectory. The mapping process starts with the platepar file, but has to adjust the plate take account of any small movements due to wind, animals, cooling or heating and so forth.
If there aren't enough stars, especially near the edges of the image, then RMS can't perform this automatic adjustment. To help you, RMS produces an image each night that shows you how well this fit was done. Look for a an image whose name ends in "calib_report_astrometry.jpg".
Photometry is used to measure the brightness of the meteor. From the plate, RMS knows the magnitudes of each star in the field of view. It works out their brightness as sum of the values of all the pixels covered by each star, minus the average background brightness. This gives it a graph of brightness vs magnitude for the stars.
The same sum is then done for the meteor in each frame where its visible. The value is plotted on the same graph, and we can then read off the magnitude.
If not enough stars are visible, then RMS can't get a good estimate of brightness or position. This might seem picky but remember RMS is a *scientific* tool, aiming to generate data that is high quality. So for example if a reference star happened to pass behind thin cloud at the wrong time, you'd overestimate the meteor magnitude. By plotting a graph of 20 or more stars, you reduce the error bars.
