- Address: Réservations par internet seulement
- Email: infoSPAMFILTER@spaceobs.com
- Phone: No telephone
- Cellphone: No telephone
- Website: www.spaceobs.com
The name MAP comes from the initial of the last name of the three participants in this project.
Alain M aury, Georges A ttard and Daniel P arrott.
MAP is a project to detect small bodies of the solar system in the southern sky using wide field telescopes, CMOS cameras and software detecting moving objects using synthetic tracking technique, written by Daniel Parrott, tycho-tracker
The scientific objectives are multiple:
Alain Maury financed the current instrumentation and wrote the acquisition script (as part of the prism software), Daniel Parrott wrote the Tycho-tracker software, and improved it in collaboration with us to make it compatible for use automated, and Georges Attard wrote the scripts which allow the automatic processing of the images as they arrive on the processing PC, and facilitate the extraction of the discovered objects. Alain Maury can be considered as professional (member of the UAI having previously worked in professional observatories) but acts as an amateur, not being linked to or funded by any professional institution, Georges Attard and Daniel Parrott are amateur astronomers, but their professional field is computer science.
Alain Maury is a passionate astronomer and a photographer by training (National School of Photography and Cinema). Self-taught, he is a member of the International Astronomical Union, initially in France and today in Chile where he lives since 2000. He discovered his first near-Earth asteroid in 1983 on photographic plates taken with the Schmidt telescope at the observatory of the French Riviera in France.
He discovered several other near-Earth asteroids and 2 comets between 1985 and 1988 while working at the Mount Palomar Observatory in California as part of the second Palomar survey. Back in France, he mainly worked on the correct functioning of the Schmidt telescope of the OCA, then on his equipment of a CCD camera allowing the automatic detection of asteroids. The telescope measured 90cm in diameter, and the camera was at the time of its construction the largest CCD camera in Europe, with 4 megapixels. At the cutting edge of technology at the time, it was controlled by a Pentium PC, with 32MB of memory, and a huge HP workstation that had 64MB of RAM (the good old days). As part of the ODAS program, a collaboration with the German Space Agency (DLR), in two and a half years 5 near-Earth asteroids were discovered. Since some astronomers believe that modern astronomy is only done through models and simulations and that there is no point in getting real data (sarcasm), the telescope was shut down by decision of the director of the observatory assisted by its deputy director.
Alain Maury changed continent and then worked at the Southern European Observatory within the framework of the DeNIS programs, then the EROS2 program. Then he left CNRS (the public institution of scientific research in France, also known as the National Center for Ridiculous Salaries) to set up a public observatory in the Atacama desert, with the aim of showing the southern sky to the public, but behind the scenes with the idea of discovering again some asteroids in the future.
In 2015, he conducted an automated research program in collaboration with Joaquin Fabrega from Panama, using a 40cm Newton Telescope (ASA) and a 16-megapixel FLI PL16803 camera. This program was carried out during a year and gave no interesting discoveries. This program detected many interesting objects, but the limiting magnitude was relatively low and these objects were already discovered in the previous weeks by more powerful telescopes. During this project the basis of the observation script being used right now was written
In 2014 the American firm Celestron marketed wide-field telescopes, the RASA (Rowe Ackerman Schmidt Astrograph) having 28cm in diameter, but only 62cm in focal length, giving a very large field of 3.3x3.3 ° with an ML16803 camera. A first telescope was purchased, mounted on a Paramount ME mount, and it showed that it was possible to cover a large area of the sky in a very short time. But with a 3'' pixel size, the limiting magnitude was not very impressive. Plus the CCD camera has a substantial readout noise and relatively low quantum efficiency.
First prototype, RASA 11, ML16803 CCD camera on Paramount ME mount
Since then 3 other telescopes of this type have been purchased. In 2019, the Chinese firm ZWO started to market large format cameras (24x36mm) with 60 megapixel Sony IMX455 detectors, very high sensitivity, very low noise, almost instantaneous readout time, and it seemed obvious that a lot would be gained by using these cameras at the focus of RASA telescopes.
2019: 4 RASA, 2 cameras
A solid mount was also purchased (VMA200 from the French manufacturer Valmeca) to support the 4 telescopes. The whole installation is located inside a clamshell-type dome. A mounting system for each telescope has been made in Chile to correctly offset the 4 telescopes so as to cover contiguous fields. It was also necessary to make adapters allowing the cameras to be mounted without producing vignetting. Last addition, to save time and spend less time focusing to compensate for the expansion of the original aluminum telescope tube, 4 carbon fiber tubes were purchased.
With the 4 telescopes, an image represents a total of 240 megapixels, therefore comparable to the largest cameras used on professional telescopes, with the same cosmetic quality if not better (no gaps between the different fields). The Megacam camera from CFHT in Hawaii is 288 megapixels, OmegaCAM at ESO's VST is 250 megapixels.
At the software level, it seemed obvious that the ancestral technique (dating from 1989, Spacewatch project) consisting of photographing the same field several times with a given time interval was no longer necessarily the ideal method, we had, 30 years later, to have progressed a little and to use something a little bit smarter and efficient.
Alain Maury in 2018 got in touch with Michael Shao of JPL (NASA) who had, with his team, developed the first software using synthetic tracking. The idea is to do a series of short exposures, then shift and add them by following the movement of the asteroid that we want to detect. The problem is that when we try to detect new asteroids we do not know the speed or direction of the asteroid. In this case the brutal method consists in making all the possible additions in all possible directions and seeing if we detect a possible asteroid. This is now possible thanks to GPU (Graphics Processing Unit) cards which increase the computing power of a PC by several orders of magnitude. Alain Maury having left a comment on RASA telescopes on a youtube video, he was contacted by Daniel Parrott who told him that he was developing a synthetic tracking software, and the two began to work together, quickly joined by Georges Attard who is with Alain Maury a member of the Groupement d'Astronomie Populaire de la région d'Antibes ( GAPRA ), the best astronomy club in France :) (in all modesty).
Initially, not at all usable in automatic mode, tycho tracker was gradually modified to be able to be scriptable, and has undergone many improvements, including the use of the GAIA DR3 astrometric catalog (GRAPPA version produced by Marc Serrau, a French amateur astronomer). A dll (dynamically linked library) kindly provided by Raoul Behrend (Geneva observatory) allowing the calculation of the precise position of already known asteroids was also implemented in Tycho. Georges then wrote a control panel which allows the management of detections made during the night and we will work very soon in the automatic management of the confirmation of objects discovered using a 50cm telescope at F / 4 whose construction is almost finished in early 2021.
2 RASA on the VMA mount
The pandemic has made financing the operation much more difficult. For the moment we have the 4 RASA telescopes on their mount, 4 zwo 6200 cameras (but one is used on the tracking telescope, so at best we can put 3 RASA in operation), the 5 data acquisition PCs, and only 2 image processing PCs. Ultimately, it should look like this:
The detections made by each GPU PC land on PCGPU1 for confirmation, and it sends the requests for confirmations to NUC5 (confirmation telescope). There is a whole synchronization procedure between the master NUC1, and the NUC2, 3 and 4 which are slaves and which receive the commands from the master PC NUC1.
The acquisition software runs with the prism software. Once launched, it calculates the height of the sun. At -5 ° altitude, it opens the dome, and cools the camera. We are working at only -10 °C, the zwo camera do not allow to descend much lower in a stable way in summer (30 ° of delta T is already a lot). Once the sun is at -12 ° altitude, the observations begin automatically. There is a text file of feasible fields in the sky, from 0 ° declination to -90 °, in bands 2.1 ° high, and each band is divided into 3.2/cos(declination) fields. Over time we will therefore always come back to the same fields. Initially (say just after the full moon), the file is filled with 0's, one 0 per field, separated by a space from the next field. When a field has already been observed during the lunation, the 0 is replaced by the number of the day on which the field was taken. This means that the next day an already observed field is ignored. We only observe between -2h and + 2h from the meridian. And so the telescope goes from one field to the next, as long as they haven't been done. If it encounters a field already made, it switches to the lower declination band. If it arrives 2 hours after the meridian, it also passes to the lower declination band.
With a single camera (3.3x2.2 ° field, 1.2 "pixel) there are 3087 feasible fields in the southern hemisphere, on 42 bands of different declensions (whose field centers start with -1.05 °, -3.15 °, -5.25 °, etc ....). There are 113 fields on the first declination zone (at 0 °: 360 / 3.2 = 113 fields) and only 6 fields on the -87.15 ° zone
The software allows the detection of asteroids and recognizes the asteroids already known (present in the file mpcorb.dat), and allows to classify these asteroids according to the level of confidence which one can grant to the detection. Because of the large field offered by the camera, using (for now) 36 exposures of 30 seconds (a total of 18 minutes) we can detect sources of magnitude 20.5 with relative confidence.
For now, we are observing with one telescope, the second will arrive quite quickly. We select a starting declination, and the software manages according to the sidereal time to remain observing between plus or minus 2 hours around the meridian while avoiding the presence of the moon.
During an exposure, the previous exposure is preprocessed (dark, flat) refocused to within a pixel (no interpolation), binned by a factor of 2 and sent to the processing PC which is equipped with one or two nVidia RTX2080ti graphics cards . The processing is done in real time (that is to say such that the processing of an image stack is done in a time at least less than the shooting time of the next exposure). At the start of the series, the program performs an astrometric refocusing, and a focusing. We dithering in a 20-pixel window between each exposure, so as not to have elongated residual hot spots. The telescope is not self-guided, given the short duration of the individual poses. The master program (the one that also controls the mount) sends information to other PCs in acquisition and there is a protocol allowing to synchronize all this small world, by exchanging files in common directories. The slave PC program only works by reading the files containing the observations to be made (coordinates, so as to fill in the file header correctly, exposure time, number of exposures, etc.) and once the telescope has been astrometrically recentered , it just sends an empty "go.txt" file which causes the slave PC to launch its acquisitions. Once finished, it is the slave PCs which send a "finished.txt" file in a directory of the master PC which indicates to the latter that it can move. Synchronization is at worst 1.5 seconds long, but often less, so you don't waste much time overnight.
We use the latest version of tycho-tracker, and Georges has written on the one hand scripts that allow tycho-tracker to be automatically launched with different options, and a control panel that allows on the one hand to manage the processing of images by tycho as soon as the series is complete (the acquisition program sends a file to a synchronization directory giving the location of the images to be processed once the 30 images are finished) and has developed many tools allowing to sort the detections and the management of confirmations if necessary.
Given that with synthetic tracking we can adjust the search to the type of object we want to detect, also knowing that if we really want to explore all the speeds and all the search angles, the computation time increases exponentially , we perform two detections, one on the image in bin1 for slow objects (up to 1.2 '' / min) and one on the images in bin2 for fast objects (up to 20 '' / min). In these conditions, we manage to maintain a real-time rate.
Georges also wrote a "dashboard" application which allows on the one hand to examine the triplets of images generated by tycho according to their characteristics (known or unknown objects) and their level of confidence (parameter calculated by tycho, which can be "high", "medium", "low" or "none"). In known objects, even the "low" are well identified, and in unknown objects, normally a "high" object has all probability of being real, medium, it is necessary to sort and it is rare that a "low" object is real. Once an object is selected, we can see its speed and angle, the three astrometric positions generated by Tycho, we can generate the message to send to the MPC, we can also extrapolate its position in the hours that follow, and you can click on a "to confirm" button which sends the astrometric measurements to a directory that the confirmation telescope reads every minute. If it finds a file of an object to be confirmed, it performs the observation. The number of confirmation files is incremented overnight. We also have to implement a series of "to follow" objects which are the objects discovered during the previous nights for which we must make an observation to improve the orbit.
Georges also implemented the ndigest2 software of the MPC which allows for each object that we control to see the probability that it is an NEO, a Mars cruiser, or other category of asteroid. It also gives an estimate of the quality of the 3 measured points which makes it possible to immediately see whether the object is real or not.
We now number the discovered objects as follows:
We use the numbers 1 to 9, then the letters A to Z then a to z, which allows to encode up to 62 using only one character. This protocol evolved a bit initially, but now it seems relatively stable.
First character: the last number of the year, so 1 at the moment, we can a priori use this coding until 2082 :)
Second character: Month of the year from 1 (January) to C (December)
Third character: Day of the month, so from 1 to V (31 of the month)
Fourth character: Camera number, from 1 to 5, recently added, it was not present in the first detections. The 5 corresponds to the confirmation telescope
Fifth character: Number of the image taken during the night, possibly from 1 to 62 (from 1 to z)
Sixth and Seventh character: Number of the object discovered during the night. As we observe very far from the ecliptic (it is a choice, to go to the southern sky where no one else is scanning) we find very few objects but when we find an unknown object it is all the time interesting, either an NEO, or an areocross (Mars cruiser), a Hungaria, Phocea, etc ... necessarily a very strong inclination.
Currently on a given field we carry out 36 exposures of 30 seconds (ie 18 minutes in total) which makes it possible to detect certain objects beyond the magnitude 20.5. The proportion of asteroids detected changes with the magnitude. On a given field (calibrated on the ecliptic) we detect 80% of asteroids of magnitude 20 and only 20% of asteroids of magnitude 20.5.
The San Pedro de Atacama site (actually south of San Pedro) is a very good site with very dark skies and over 300 clear nights per year.
The following image gives the 2021 calendar (in blue the nights observed, even partially) and in gray the nights when no observations have been made. The observations started in 2020 but rather in test mode, now the software runs on its own and the observations resumed on January 7, 2021.
Discoveries can be found on this page
Our strategy is to clearly observe in negative declensions where for the moment no one is observing beyond magnitude 19. We are optimizing the search for asteroids towards fast asteroids. Being at very negative variations we hardly find any object of the main belt, except if they have a very strong inclination.