from: "WGN - Journal of the International Meteor Organization" 23-6 (1995), p.225

Systematic Errors of Visual Meteor Brightness Estimates

Video systems can deal with many different tasks in meteor astronomy. Due to their sensitivity and accuracy they are used to obtain data about meteor showers, telescopic meteors, orbits, meteor light curves, spectra and other properties ([1]). Visual like video records can also support the training of new observers to show them what to expect and what they have to look for. Last but not least we can check the reliability and accuracy of visual observations with them. Since our video system MOVIE ([2]) has almost visual characteristics it is very useful in this later field.
On the IMC '95 it was shown that plottings of even inexperienced visual observers result in quite accurate radiant positions ([3],[4]) even though the errors of individual meteors are relatively big. Another problem are meteor brightness estimates: As described in [5] we have recognised a constant shift of 0,5 to 1 mag of visual estimates based on double observations with MOVIE in summer '94 (i.e. visual observers underestimate the brightness significantly). I planned to analyse other sets of data to check this result and find out possible reasons for this considerable difference. Since all our video tapes are analysed now I could inspect the data from the 1993 Perseid maximum in detail and came up with some interesting results.

The described investigation was based on 213 meteor brightness estimates from 3 observers (Kathrin Düber, Sirko Molau and Mirko Nitschke) referring to 106 meteors recorded with MOVIE. I considered only events where the visual and video times agreed legibly. Especially from the morning hours I had to reject many double observations because the assignment was not sure anymore in times with several meteors per minute.
First of all the general trend occurred again: All three observers underestimated the meteor brightness by half a magnitude in average or even more. Table 1 gives the mean difference between the visual and video meteor brightness and the standard deviation. It should be mentioned that all visual estimates were done in 1 mag steps whereas the video brightness was computed with a resolution of 0,1 mag.

Table 1 - mean meteor brightness estimate errors for three visual observers _______________ observer | m(vis) - m(vid) | st. dev. | n -------------------------------------------------- DUBKA | +0,45 mag | ±1,08 mag | 70 MOLSI | +0,94 mag | ±0,95 mag | 69 NITMI | +0,68 mag | ±0,82 mag | 74 -------------------------------------------------- mean | +0,69 mag | ±0,95 mag | 213

There are differences between the observers but a part of the higher values for MOLSI and NITMI do result from only a few meteors in the strong morning twilight. In the absence of reference stars brightness estimates became especially difficult. Hence, the meteors were underestimated more strongly. Nevertheless there is a systematic estimaten error of about half a magnitude.
My first explanation for the effect was based on the properties of the estimate process: A visual observer watches an event which lasts only a friction of a second. He estimates the brightness later by comparing the remembered impression of the meteor with known stars. Thus, he considers unconsciously the appearance of the whole event whereas video systems measure the meteor brightness frame by frame and obtains the real maximum value. It is now important to know whether or not the estimate errors vary with the meteor velocity. We could expect that fast meteors are stronger underestimated than slower ones. The result of that analysis is given in Table 2 and Figure 1.

Table 2 - dependency of brightness estimate errors from the meteor velocity _______________ meteor velocity | m(vis) - m(vid) | st. dev. | n ------------------------------------------------------- <16 °/s | +0,75 mag | ±1,03 mag | 42 16..20 °/s | +0,68 mag | ±0,97 mag | 92 21..25 °/s | +0,60 mag | ±1,48 mag | 63 >25 °/s | +0,36 mag | ±0,63 mag | 16

The table gives the average values for all three observers to get better statistics from bigger numbers of events. Surprisingly it seems to suggest that brightness estimates are better for faster meteors, at least they do not become worse.

Figure 1 - dependency of brightness estimate errors from the meteor velocity

[Figure 1]

I would conclude from these data that the systematic estimate error is independent from the apparent meteor velocity. This is supported by the fact that the error development showed different trends for the three single observers. One has to consider the large scatter of the values, too. The standard deviation is generally very high. Especially the last row is based on only very few video meteors and should therefore not be taken into consideration. The standard deviation does not represent the accuracy of the difference in this case.
To cross-check the result I have changed the variables looked for dependencies of the meteor velocity from calculated estimate errors. The result is given in Table 3.

Table 3 - dependency of the meteor velocity from brightness estimate errors _ brightness error | v | n ------------------------------------- >+1,5 mag | 18,3 °/s | 43 +1,0..+1,5 mag | 19,5 °/s | 58 +0,5..+0,9 mag | 18,5 °/s | 34 0,0..+0,4 mag | 18,8 °/s | 33 <0,0 mag | 19,4 °/s | 45

It is obvious that variable meteor velocities do not influence the investigated systematic error. Another analysis dealt with the error dependency from the meteor brightness. I had the impression that especially fainter meteors are stronger underestimated. Table 4 and Figure 2 show the error distribution for different brightness classes averaged for all three observers.

Table 4 - dependency of brightness estimate errors from the meteor brightness _______________ meteor brightness | m(vis) - m(vid) | st. dev. | n --------------------------------------------------------- <-0,9 mag | +0,52 mag | ±1,12 mag | 38 -0,9..0 mag | +0,61 mag | ±1,09 mag | 59 +0,1..+1 mag | +0,73 mag | ±0,81 mag | 47 +1,1..+2 mag | +0,81 mag | ±0,82 mag | 50 >+2 mag | +0,67 mag | ±0,97 mag | 19

We should leave out the last row, because faint video meteors are difficult to measure. The SNR becomes that bad that slight fluctuations can have strong effects on the calculated meteor brightness. There is, as expected, a nice trend that bright meteors are better estimated than faint ones.

Figure 2 - dependency of estimate errors from the meteor brightness

[Figure 2]

However, this trend does not hold for all three single observers and the cross-check (Table 5) does not support that thesis either.

Table 5 - dependency of the meteor brightness from brightness estimate errors _ brightness error | m | n ----------------------------------- >+1,5 mag | 0,27 mag | 43 +1,0..+1,5 mag | 0,48 mag | 58 +0,5..+0,9 mag | 0,54 mag | 34 0,0..+0,4 mag | 0,83 mag | 33 <0,0 mag | -0,21 mag | 45 -----------------------------------

Big positive differences seem to result from brighter meteors whereas good estimates come in average from fainter ones. There is a sudden change in the trend for overestimated meteors which is hardly to explain. It might be influenced by errors of the brightness calculation routine due to 'burnt out' meteors even though this effect was already taken into consideration.

I can conclude from the analysis that there is a systematic error in visual meteor brightness estimates in the order of 0,5 mag which is independent from both meteors brightness and velocity. The suggested explanation for this effect could neither be proved nor rejected, but we have to consider such systematic errors if we want to derive physical showers properties based on visual brightness distributions.


[1] Hawkes, R. L. (1994), "IAU Report on TV Meteor Activity for 1991-1994",
Reports of Astronomy, p.216

[2] Molau, S. (1993), "MOVIE - Meteor Observation with VIdeo Equipment",
Proceedings of the International Meteor Conference 1993, p.71

[3] Koschny, D. (1995), "Positional Accuracies of Simulated Meteor Observations",
Proceedings of the International Meteor Conference 1995, in print

[4] Molau, S. (1995), "MOVIE - Actual Observations and Latest Results",
Proceedings of the International Meteor Conference 1995, in print

[5] Molau, S. (1994), "MOVIE - Analysis of Video Meteors",
Proceedings of the International Meteor Conference 1994, p.51

Sirko Molau; last change: October 7, 1995