Digitization of Astronomical Photographic Plates René Hudec, Lukáš Hudec, Matěj Pur Astronomical Institute, Academy of Sciences of the Czech Republic, Ondřejov, Czech Republic Czech Technical University in Prague IBWS2015 1
Motivation Many objects in HE astrophysics are also emitters of optical light They are variable The astronomical plate archives represent the only method how to study the behaviour of the objects over very long (100 years or even more) time intervals, and the only method to go back in time In addition, huge monitoring times allowing to detect and to study rare events such as flares IBWS2015 2
The Plate Archives The recording medium in astronomy (and not only astronomy) for recording images was photographic emulsion until about 1980, i.e. for about 100 years Hence all scientific Institutes involved in the past in imaging are expected to have archives of recorded images and/or spectra (as photographic negatives) IBWS2015 3
Importance of astronomical plates in HE astrophysics Providing supplementary observational data for objects with optical counterparts on: long-time evolution up to > 100 years flaring activity In some cases fine sensitivity (mag 20-23) In some cases spectra (up to mag 18) In some cases ability to study short-term time variations ~mins Huge amounts of measurements: typical plate has ~ 1 mil. objects, 10 million of plates ~ 10 13 data points brightness/time IBWS2015 4
IGR J12349-6434 = RT Cru hard X-ray symbiotic star on historical Leiden Franklin Adams Plates INTEGRAL gamma-ray source visible on astronomical plates taken by 100 years old optical talescope 90 years ago Violent (amplitude 3 magnitudes) optical brightness variations identified on the historical plates. Confirmation of classification by Masetti et al. We can follow the evolution of the source over almost 100 years IBWS2015 5
Various types of astronomical plates Multiple exposure LD Spectral image Direct Image HD spectrum IBWS2015 6
New: The digitisation of the astronomical plates, the novel software and the use of powerful (1 plate can represent up to 1 GB) computers allow the automated data mining and scientific evaluation of the plates for the first time Interface between astronomy and informatics (excellent task for informatics students!) IBWS2015 7
Plate Digitization Why to digitize the plates? The photographic emulsions can be damaged To enable evaluation by dedicated software and powerful computers The access to the info in the plates is otherwise very difficult What positional accuracy is needed? What resolution (pixelsize in microns) is needed?
Plate Digitization II Very high astrometric positional accuracy 0.1 microns only with custom made scaners, very expensive Most applications do not need such high accuracy Pixelsize: emulsion 5 microns, Sonneberg 25 microns. Smaller pixelsize = very large size of image files = more difficult to store, to access, and to evaluate
3 basic digitization techniques Custom made scanner. Very accurate but very expensive, expensive use, problems with maintenance and service later Commercial Plate/Negative flatbed scanner. Digital camera
Plate Digitization Cost and Time Factor Dedicated custom made accurate scanner very expensive, difficult to move, 1 plate > 100 Euro Commercial flatbed scanners moderately expensive, 1 plate >> 10 Euro Digital camera inexpensive, fast 1 plate < 1 Euro Cost of purchasing instrumentation not included
Comparing Digital Camera vs Scanner Digital Camera Very fast (2 sec/plate) Very low cost scanning No waves caused by irregular movement Easy transportable Satisfactory resolution for small and medium sized plates (5 20 microns) Need to apply automated algorithms for lens distorsion and edge cutting solved
Transportable plate scanning device Plate collections visited with test scanning: Carnegie Observatories Pasadena Lick Observatory Yerkes Observatory Mt Palomar Observatory PARI NC KPNO Tucson CFHT Waimea, Hawaii IfA Manoa, Hawaii USNO Flagstaff, AZ USNO Washington, DC & 20 more About 3 million plates in these archives Preferences: transportable, very fast scanning, high Less expensive alternative to repeability (no moving plate scanner cost reduction > 50 scanner parts) Used equipment: Camera: 39 Mpx high quality optics, tripod, light table, control s/w, IBWS2015 Lens distorsion treatment etc 13
Test plate scanning of Takubaya and Tonanzintla Mexico IBWS2015 14
CFHT Waimea Hawaii Valuable plates taken by 3.6 m CFHT telescope Very deep lim magn
Bologna Unproperly stored Some badly damaged Historical value Bologna tower telescope 1 st MMT in the world IBWS2015 16
Asiago Plates IBWS2015 17
Catania Mt Etna Comet Halley 1910 Unique spectra 1905-1912 Plates taken by CdC Catania Telescope ~ 10 000 plates CdC and spectra Rare set of Nova Geminorum 1912 spectra IBWS2015 18
UNAM Mexico Takubaya CdC plates CdC Carte Du Ciel project Large sky survey Plates at many observatories Mexico Takubaya completely digitized in April 2014 (~2 000) IBWS2015 19
Bochum Observatory Germany IBWS2015 20
Heidelberg Germany These small plates were not scanned within the Heidelberg scanning project IBWS2015 21
Bern Switzerland Plates stored in bank cabinet (tresor) IBWS2015 22
Solar Eclipse Einstein relativity theory tests 1919 (Eddington) original plate digitized by us at Yerkes IBWS2015 23
Astrometry Tests of digitized plates Center (RA, Dec): (286.397, 25.253) Center (RA, hms): 19h 05m 35.189s Center (Dec, dms): 25 15' 09.472" Size: 14.7 x 14.7 deg Radius: 10.416 deg Pixel scale: 10.8 arcsec/pixel
Comparing Camera vs. Scaner (the residual plots) Scaner Camera Bamberg Southern Sky Survey Plate NZ 11844
The procedure (Taavi Tuvikene) The images were run through solver script, which calls SExtractor for extracting all stars, then astrometry.net for an initial solution, and finally does solving in sub-fields recursively. Two sets RA and Dec were obtained: one from astrometry.net (with SIP order 3) and second from the SCAMP software in subfields. The resulting coordinates were matched with the UCAC4 catalogue in TOPCAT and plotted the residuals (in arcsec). The SIP3 solution shows wobbles from the scanner arm movement (in Dec). These wobbles are missing from the digital camera image.
The "recursive" solution eliminates the distortions. The standard deviation of the residuals with the scanner method are 1.8 and 1.6 arcsec (plates 11844 and 11855, respectively). Stddev with the digital camera are 2.4 and 2.2 arcsec. Pixel scale is 1.77 arcsec with scanner and 5.36 arcsec with camera. The photometric accuracy was also investigated and will be presented by Petra Skala Comparison with scaner not trivial as similar studies are, to our knowledge, not available Details on photometric accuracy see next talk by P. Skala. Not trivial to compare with scaner as no such study available.
Damages to plates: need to digitize fast Gold disease and damage by humidity Collaboration with IChT Prague IBWS2015 28 in recovery damaged plates
Gold disease Worlds unique images Halley Comet 1986 Klet Observatory Czech Republic IBWS2015 29
Open Science Project (high school student) Digitized plates(images) analzsed with ImageJ IBWS2015 30
Astrometry analyses IBWS2015 31
Carte du Ciel Plates astrometry Center (RA, Dec): (178.349, 11.387) Center (RA, hms): 53m 23.650s 11h Center (Dec, dms): 11 23' 12.371" Size: 1.79 x 2.01 deg Radius: 1.345 deg Pixel scale: 1.96 arcsec/pixel IBWS2015 32
Conclusions Astronomical Plate Archives represent valuable data source for all optically variable sources including binary blazars Most important: long time evolution, dense sampling, rare flares, spectral changes Recent wide digitization and evolution of dedicated software enables evaluation by computers, for the first time Fast (<20 sec) and inexpensive (~0.25 $) plate digitization technique developed and tested able to convert glass plates to computer files effectively IBWS2015 33
The End IBWS2015 34
Plate Digitization Why to digitize the plates? The photographic emulsions can be damaged To enable evaluation by dedicated software and powerful computers The access to the info in the plates is otherwise very difficult What positional accuracy is needed? What resolution (pixelsize in microns) is needed?
Plate Digitization II Very high astrometric positional accuracy 0.1 microns only with custom made scaners, very expensive Most applications do not need such high accuracy Pixelsize: emulsion 5 microns, Sonneberg 25 microns. Smaller pixelsize = very large size of image files = more difficult to store, to access, and to evaluate