RMS cameras as alternative for Watec in CAMS

Abstract: The new RMS cameras have been tested in Belgium since December 2018. From March 2019 onwards RMS cameras got fully operational within the CAMS BeNeLux network. The conclusion after months of testing is that the RMS is a valid alternative for the Watec camera. The RMS is significant cheaper, has a larger FoV and better positional accuracy, produces less false detections and has a slightly better score in percentage of orbits while the imaging quality is much better than that of the Watec camera.

 

1 Introduction

The CAMS BeNeLux network started in March 2012 with as standard equipment the relative expensive Watec H2 Ultimate cameras. Since all participants had to finance their own equipment, these costs seriously refrained amateurs to join the network. A camera with the required f/1.2, 12mm lens, frame grabber, power supply, video cables and camera housing requires a budget of about 650 Euro for each camera. A dedicated PC is also required for CAMS, regardless if one operates one single or several cameras. Since 2012 amateurs bought over a hundred of these camera units installed at about 25 camera stations of the BeNeLux network. A number of sites were equipped with 8 Watec cameras. Unfortunately, some amateurs quit and about 20 of these cameras are no longer used.

Since the CAMS standard equipment is based on 15-year-old technology, the hardware risks to become unavailable on the market. The EzCap frame grabbers tend to fail rather rapidly when being used permanently, but replacement becomes difficult to find. Windows 10 with its unavoidable updates caused some problems too. In recent years it became clear that we have to look for an alternative for the standard CAMS equipment. One possible alternative are the RMS cameras, introduced by Denis Vida and sold via the Croatian IStream. The first RMS cameras were offered for sale in October 2018 and the first such camera arrived in Belgium for testing in late November 2018. After some preliminary tests the first RMS got integrated in the CAMS BeNeLux network from 17–18 March 2019. After more than one and a half year of practice and testing, it is time for conclusions.

 

2 About the RMS camera

RMS stands for Raspberry Pi Meteor Station and has been developed in Croatia since 2014 when a Raspberry Pi was first used to record meteors (Zubović et al., 2015). Since then the project developed further into a more performant system that could be exported beyond Croatia. The software for the RMS cameras has been designed by Denis Vida, the registered meteor data is being collected and analyzed within the Global Meteor Network (Vida et al., 2019a; 2019b). Historically, the CAMS software has its roots in the video capture and detection methods and algorithms developed since 2006 in the Croatian Meteor Network project (Gural and Šegon, 2009). The new RMS software is to a large extent compatible with CAMS which is a major advantage on any other alternatives for the standard CAMS equipment. Some aspects were adapted in function of CAMS for instance the config file allows to define a CAMS ID number.

 

 

The first advantage of RMS is the required budget, purchased as plug and play, 450 Euro (price as advertised August 2020), ordering and assembling the components costs only ~200 Euro against 650 Euro + PC for a classic CAMS camera. Figure 1 shows the installation setup and Figures 2 and 3 display the camera components in detail. The RPi replaces the PC, no expensive video cables and no fragile frame grabbers are required. The images from RMS cameras show many more stars than those from Watecs (see Figure 6).

An excellent guideline can be found online how to assemble all components to build your own RMS camera with a budget of about 200 Euro.

 

 

 

 

While CAMS is running at most camera stations with a battery of Watecs, often 8 or even 16 cameras operated by a single PC, each RMS comes with its own small computer, the RPi. If a CAMS PC has a failure, all its cameras are affected, while a failure on an RPi affects only a single camera. The data reduction, removal of false detections and reporting of the data can be done for CAMS in a single procedure. Doing the same number of RMS cameras one by one takes significant more time.

The CAMS Watecs with the f/1.2, 12mm lenses and FoV of 22° × 30° register many short faint meteors which are at the limit of detectability with poor chances to be multiple station.  The RMS is a bit less sensitive and missing the faintest magnitude meteors captured by Watecs. The advantage of the RMS camera is that it can be combined with larger FoV optics with a good resolution, for instance at dark sites the f/0.95, 3.6 mm lens covers more than 4 times the standard FoV of the CAMS Watecs. For light polluted areas the f/1.0, 8mm les is recommended which has a FoV about 1.5 times that of the standard CAMS Watec.

The CAMS detection algorithm measures twice as many points on a meteor trail than the RMS software. However, the positional accuracy of the RMS is better because of the calibration for each detection while CAMS extrapolates from a single reference calibration for the entire night. Since the RMS requires the presence of a minimum of stars for each detection, it will not detect anything when the minimum of required stars is not available. While CAMS often detects meteors through thin clouds without any stars being visible, the RMS will detect nothing in such situation.

 

3 RMS in Belgium

The first RMS got installed in December 2018, with the large FoV f/0.95, 3.6mm lens at the light polluted camera station in the city of Mechelen. After initial tests this camera was replaced in February 2019 with a f/1.0, 8mm lens, identified as BE0002 in the Global Meteor Network and 003830 in CAMS. The first RMS got operational in Grapfontaine from 15–16 May 2019 identified as BE0001 with CAMS ID 003814. A third RMS followed 17–18 July 2019 in Genk with codes BE0003 and 003815. The fourth RMS had its first meteors 22–23 August 2019 in Mechelen, labelled BE0004 and 003831. Figure 4 shows the field of view (FoV) of the 4 cameras projected at an altitude of 100 km in function of the coverage required for the other cameras of the CAMS BeNeLux network.

Although each camera was focused, configured and tested in Croatia before being shipped as plug-and-play, all four RMS cameras required some finetuning after being installed. Without the assistance by Denis Vida, I would not have managed to get the RMS functioning. Luckily for me, Denis Vida solved the problems remotely or provided me with precise instructions what to do to help me out. Meanwhile the system got regularly updated and became far more stable than it was in the very beginning.

 

 

4 Watecs versus RMS

The four RMS systems in Belgium have been successfully operational for a relevant period of time. There is enough data to compare some basic statistics to compare both systems, CAMS with Watecs versus CAMS with RMS, in order to answer the question whether or not the RMS cameras can be used as a valid alternative for the old Watec configuration for CAMS.

 

Table 1 – Comparison between RMS and Watecs in 2019. RMS BE0002 started 17–18 March 2019 (271 nights), BE0001 started 15–16 May (222 nights), BE0003 started 17–18 July (154 nights) and BE0004 started 22–23 August (132 nights). All 6 Watecs were running entire 2019 (365 nights).

Camera	Detect.	Meteors	%	Orbits	%
BE0001 3814	9993	7830	78.4%	5573	71.2%
BE0003 3815	11622	2871	24.7%	2021	70.4%
BE0002 3830	22504	4973	22.1%	3512	70.6%
BE0004 3831	22070	2947	13.4%	1098	37.3%
Total RMS	66189	18621	28.1%	12151	65.3%
Watec 383	39518	3654	9.2%	2322	63.5%
Watec 384	39520	4387	11.1%	3023	68.9%
Watec 388	39520	3462	8.8%	2503	72.3%
Watec 389	39516	3118	7.9%	1245	39.9%
Watec 399	39517	3575	9.0%	2724	76.2%
Watec 809	39500	3981	10.1%	1997	50.2%
Total CAMS	237091	22177	9.4%	13814	62.3%

 

Table 2 – Comparison between RMS and Watecs in 2020. RMS BE0002 (177 nights), BE0001 (177 nights), BE0003 (178 nights) and BE0004 (177 nights). All 6 Watecs were running during the entire period 1 January until 30 June 2020 (182 nights).

Camera	Detect.	Meteors	%	Orbits	%
BE0001 3814	6943	3880	55.9%	2710	69.8%
BE0003 3815	33513	2659	7.9%	1857	69.8%
BE0002 3830	23246	1753	7.5%	1344	76.7%
BE0004 3831	16617	1807	10.9%	1107	61.3%
Total RMS	80319	10099	12.6%	7018	69.5%
Watec 383	14989	1285	8.6%	866	67.4%
Watec 384	14989	1288	8.6%	884	68.6%
Watec 388	15001	1109	7.4%	807	72.8%
Watec 389	15002	1056	7.0%	409	38.7%
Watec 399	14996	1144	7.6%	892	78.0%
Watec 809	14999	1334	8.9%	794	59.5%
Total CAMS	89979	7216	8.0%	4652	64.5%

 

We split the available information in two sets, the first with the earliest RMS data mainly obtained during the second half of 2019 with the meteor rich season (Table 1) and the second during the first 6 month of 2020 (Table 2). Two trends can be spotted, RMS seems to have less false detections and a higher percentage of meteors that prove to be multi-station with a valid orbit. However, the results differ a lot between the cameras and requires a look at each of them separately.

BE0001 f/0.95, 3.6mm lens (CAMS 003814)

This camera was first installed end 2018 in Mechelen but its optics proved to be unsuitable for light polluted sites. Therefore, the camera was moved to Observatoire Centre Ardennes, a public observatory in Grapfontaine, a dark region in the south-east of Belgium. The camera got reinstalled in April 2019, but it took a while before some technical issues were solved. From May till December 2019 this camera had 9993 detections of which 7830 were confirmed as meteors or 78.4%, good for 5573 orbits or 71.2%. The first six months of 2020 confirmed this trend with 3880 meteors out of 6943 detections or 55.9%. Stormy winter weather with fast moving clouds caused unusual numbers of false detections, but still the proportion remained much in favor of RMS compared to the Watec scores. The number of 2710 meteors (69.8%) with orbits is slightly lower due to poor coverage from the northern part of the CAMS network which had bad weather.

With these scores this camera performed as the best camera of the CAMS BeNeLux network. The camera is pointed low at 37° elevation so that its large FoV overlaps with almost 2/3^rd of the network because of its large 47° × 88° FoV. The lens is very efficient at a dark sky, BE0001 often detects only meteors without any false detections which is a great advantage for a video meteor camera.

BE0002 f/1.0, 8mm lens (CAMS 003830)

This camera was purchased as replacement for BE0001 in Mechelen and got operational 17–18 March 2019. The lens was chosen because of the problematic light pollution in the city of Mechelen. Moreover, the camera is pointed low at 30° right into the worst light polluted part of the sky. With 22504 detections of which 4973 meteors or 22.1% the camera has substantial more false detections mainly caused by planes from the nearby Brussels airport. With 3512 orbits or 70.6% the camera scores very high. First three months of 2020 had huge numbers of false detections caused by rapid moving clouds in stormy weather. April, May and June had almost no false detections because of the Covid lockdown with almost no air traffic. With low meteor activity 92.5% of all detections were false. 1344 meteors of the 1753 resulted in an orbit, or 76.7% and that makes this camera one of the best performing in the CAMS network. With its f/1.0, 8 mm lens and FoV of 22° × 41°, this proves to be an ideal camera for light polluted areas.

BE0003 f/0.95, 3.6mm lens (CAMS 003815)

BE0003 got installed in July 2019 on the roof of Cosmodrome, a public observatory in Genk. The camera had to be pointed south and to avoid over exposure by moonlight a f/0.95, 6mm lens was ordered. Unfortunately, the camera was delivered with a wrong lens, the f/0.95, 3.6 mm. As a change would take several weeks, it was decided to try the camera with this lens. However, all nights around Full Moon proved to be ruined as no calibration is possible when the Moon is in the FoV and no detections can be recorded. Capturing since 17–18 July 2019, BE0003 had 11622 detections of which 2871 were meteors or 24.7%. Light pollution and planes caused many more false detections in Genk than at the darker location of Grapfontaine (BE0001). With 2021 orbits or 70.4% the camera still scores very well. The first 6 months of 2020 confirm these scores with 7.9% of all detections being meteors and 69.8% of all meteors resulting in an orbit.

One reason why the total number of meteors remained far less than that of an identical RMS in Grapfontaine was caused by humidity. During humid nights the camera housing gets covered with dew above the flat roof. To avoid dew the camera will be moved to a position at the edge of the roof. To reduce the problem with light pollution and the Full Moon in the FoV, the RMS will be replaced by a new camera with a f/0.95, 6mm lens and FoV 30° × 54°, a change that got postponed due to Covid and lockdown measures. We strongly recommend not to use the RMS with a f/0.95, 3.6mm lens with its 47° × 88° FoV for cameras that must be pointed in Southern direction because the calibration fails during a few nights around Full Moon.

BE0004 f/1.0, 8mm lens (CAMS 003831)

Since BE0002 proved to be very efficient being pointed low to give coverage on a large part of the network in the Netherlands, an identical RMS was purchased to give coverage over Luxembourg and the south-eastern camera fields of the network. BE0004 got installed with some delay as it took a while to solve some technical issues. This camera got pointed South East at 32° elevation. The smaller FoV, 22° × 41° is more suitable in the light polluted city of Mechelen. Even with Full Moon in the FoV, the camera registers meteors. Started 22–23 August 2019, BE0004 had 22070 detections of which 2947 were meteors or 13.4%, many of the false detections being caused by planes and moonlight reflected on the edges of clouds. 1098 of the meteors combined with some other stations to obtain an orbit, or 37.3%. This percentage is much lower that for the three other RMS cameras since the region covered by this camera had poor coverage from other camera sites. The first 6 months of 2020 confirm the statistics for the camera. The significant increase in percentage of meteors with orbits happened because of some adjustments in the camera network to improve camera coverage on this area (Table 2).

 

 

5 Advantages of the RMS

CAMS is using the Watec H2 Ultimate with a small FoV of 22° × 30° f/1.2, 12mm lenses (Jenniskens et al., 2011). These are very efficient with severe light pollution. When considering to use RMS cameras different optics can be chosen. The larger the FoV, the more meteors the camera may capture if the sky is dark enough. The choice for the optics depends on the light pollution.

Another important aspect is the resolution. A Watec H2 with f/1.2, 12mm lens has a resolution of 2.8 arc/pix in NTSC format and 2.5 arc/pix in PAL. The RMS with f/1.0, 8mm lens has a resolution of 1.9 arc/pix with a FoV of 22° × 41° which is better than the Watecs. The option with f/0.95, 6mm lens has a resolution of 2.5 arc/pix, identical to the Watec in PAL format, but the FoV is 30° × 54°, significantly larger than that of the Watec. The f/0.95, 3.6 mm has a resolution of 3.9 arc/pix but a FoV of 47° × 88° which is huge compared to the Watecs.

The positional accuracy on video meteor cameras depends on the astrometric accuracy of the calibration which is based on a single calibration for an entire night or series of nights in the case of CAMS assuming that these remain stable for a fixed camera. However, the calibration parameters change during the night and the deviations are far larger than the resolution of the camera. The RMS comes with a detection and calibration algorithm which adjusts the general calibration for each single detection. This correction requires standard minimum 20 stars. In light polluted regions this can be lowered to 12 stars while the theoretical minimum to have a solution is 5 stars. This requirement means that RMS ignores meteors detected through clouds when not enough reference stars are present for the calibration correction. This is a major advantage in favor of the RMS with significant better-quality positional accuracy than the Watecs with the CAMS calibration.

Very bright meteors are problematic with CAMS as pixels get randomly detected in overexposed parts, RMS ignores overexposed detections in its standard detection algorithm. Fireballs got a separate solution which requires a manual procedure to determine the positions. This way unreliable positions are banned from the DetectInfo file for further automated data processing.

The significant better positional accuracy, less false detections with a much larger FoV with a good resolution definitely all favor the RMS above the Watecs of CAMS.

 

6 Disadvantages of the RMS

The concept of the RMS offers important advantages, but during the tests we encountered some problems too:

• While setting up a Watec to start recording meteors with the CAMS software was really plug-and-play, installing the RMS was less straight forward. These IP cameras connected with the RPi based on Linux had specific network problems that had to be solved. Resetting the router of your internet provider, changing a switchbox or adding some new device may interfere with the local network and its IP addressing. If for some reason the camera gets another IP address allocated than the one foreseen in the config file, the RPi fails to connect to its camera.
• The RPi freezes every now and then. Without anyone checking the system, it would remain idle until someone reboots the RMS. Until 31 December 2019 during the 289 available nights, 46 incidents occurred that one of the four RMS cameras could not function due to some failure. The CAMS Watecs had zero incidents during this period. During 182 nights in 2020, each RMS camera lost 5 nights due to failures of the RPi while the Watecs functioned all nights without incidents. One way to reduce the number of failures is to reboot the RPi each day.
• While both Watecs and RMS capture 25 frames per second, the RMS detects one position for each frame while the Watecs with CAMS software detect two positions for each frame.
• CAMS video camera stations in most cases have 4, 6 or 8 Watecs running on a single computer. The DetectInfo files for all cameras are combined into a single DetectInfo file with a unified Archived folder for all detections of all cameras. The confirmation of meteors and elimination of false detections happen in a single procedure that may take 5 up to 10 minutes of work if no excessive numbers of false detections are caused. With the RMS cameras this procedure has to be repeated for each camera separately, which takes for each single RMS camera about the time required as for a whole battery of Watecs. To replace eight Watecs with f/1.2, 12mm lenses, six RMS cameras are required with f/1.0, 8mm lenses to have the same coverage at the sky. So far, I could not test with more than two RMS cameras installed within a single local computer network. To keep the confirmation routine workable an app is required to merge the archive folders and DetectInfo files of all RMS cameras at a camera site into a single DetectInfo which allows to do the confirmation for all cameras in a single procedure. Of course, this concerns only those who use RMS cameras within CAMS. GMN participants not involved with CAMS don’t have to bother about confirmation procedures, for them all is running fully automated.

Like for every new project some child diseases occurred with the RMS cameras, most of which got solved meanwhile. The GMN is an open source project supported by a growing community, a concept which offers more flexibility than any existing video meteor observing project. Any problems encountered with the RMS cameras may be solved by this community itself. The purchase of an RMS is far cheaper than the CAMS set-up. At dark sites a large FoV can be applied with a single RMS replacing the FoV of more than 4 Watecs. Having these RMS cameras successfully operational as part of the CAMS BeNeLux network, we can safely conclude that these cameras offer a decent alternative for the meanwhile old CAMS technology based on the Watecs.

 

 

 

7 GMN trajectory and orbit data

Although the topic of this report is about using RMS cameras within a CAMS network, the RMS and GMN offer a number of important extras, regardless whether these cameras are used within CAMS or not.

By using the RMS cameras with the GMN software, the user contributes video meteor data to the Global Meteor Network which computes trajectories and orbits that are made publicly available. All the final results can be downloaded from the website. For each RMS camera a status report is compiled for each night including stacked images, thumbnails, radiant distribution, calibration report, astrometry report, photometry report and a time lapse of the night sky. The standard RMS output also provides the detection data in UFOCapture format (R91) of the SonotaCo network. To add the CAMS format output the only requirement is to define the CAMS camera ID in the RMS config file. With overlapping neighboring networks, the same meteors were often registered by different cameras from different networks. With RMS the data of a single meteor camera can now be delivered to different networks in the appropriate format. For the BeNeLux area this means that many meteors registered for CAMS, also got analyzed by the GMN as well as the French BOAM network which uses UFOCapture format.

One of the biggest advantages of the GMN is the public availability of the results. The trajectory and orbit data are available for anyone interested to make analyzes. CAMS orbit data is made public every few years, with the data until 2016 being public now. All CAMS data from 2017 and later is still under embargo which means that even the amateurs participating in the CAMS BeNeLux network are denied access to the orbit data obtained by their own cameras. For some amateurs the lack of feedback within CAMS has been a reason not to participate. It is a challenge to keep amateurs motivated when no results can be shared and feedback remains restricted to a minimum. Using the RMS gives direct access to its results which is far better for motivation than the data black hole policy applied for CAMS.

 

 

8 Conclusion

Since my first RMS camera got installed in Mechelen in December 2018, three more RMS cameras have been installed at three different sites in Belgium. After initial tests the RMS were successfully used for the CAMS BeNeLux network from 17–18 March 2019 onwards. Several shortcomings were solved during the testing period. The weak point in the RMS system remains the RPi which freezes too easily for no reason. A new OS for the RPi is expected to solve these problems.

Despite the encountered technical problems, the Belgian RMS cameras rank at the top as the best performing cameras in the CAMS BeNeLux network. Despite some frustrations with technical issues, the overall experiences are definitely positive. In my opinion the RMS cameras provide a valid alternative for the currently used Watecs in the CAMS networks. RMS cameras with 6 mm lenses are ideal to function at remote stations. The author hopes that more amateurs beyond the BeNeLux CAMS network will invest in video meteor work to expand the coverage of the Global Meteor Network. The RMS cameras have been developed for this purpose and will continue to expand our knowledge of meteor shower activity for the years to come.

Acknowledgment

The author thanks Denis Vida for his continuous support with the installation and operation of the RMS cameras. I thank Pete Gural for providing a conversion app to adapt the GMN DetectInfo to the CAMS format. Last but not least, I thank Martin Breukers and Damir Šegon for checking this article and for their valuable comments.

References

Gural P. and Šegon D. (2009). “A new meteor detection processing approach for observations collected by the Croatian Meteor Network (CMN)”. WGN, the Journal of the IMO, 37, 28–32.

Jenniskens P., Gural P. S., Grigsby B., Dynneson L., Koop M. and Holman D. (2011). “CAMS: Cameras for Allsky Meteor Surveillance to validate minor meteor showers”. Icarus, 216, 40–61.

Vida D., Gural P., Brown P., Campbell-Brown M. and Wiegert P. (2019a). “Estimating trajectories of meteors: an observational Monte Carlo approach – I. Theory”. Monthly Notices of the Royal Astronomical Society, 491, 2688–2705.

Vida D., Gural P., Brown P., Campbell-Brown M. and Wiegert P. (2019b). “Estimating trajectories of meteors: an observational Monte Carlo approach – II. Results”. Monthly Notices of the Royal Astronomical Society, 491, 3996–4011.

Zubović D., Vida D., Gural P. and Šegon D. (2015). “Advances in the development of a low-cost video meteor station”. In: Rault J.-L. and Roggemans P., editors, Proceedings of the International Meteor Conference, Mistelbach, Austria, 27-30 August 2015. IMO, pages 94–97.