Data analysis tools for the DFBS

1. Data analysis tools for the DFBS Roberto Nesci University La Sapienza, Rome Italy Overview of the DFBS What can be retrieved from the web interface Instrumental spectral response Instrumental magnitudes and calibration COMPARE: a tool for search of variables CLASSIF: a tool for spectral classification Jenam 2007, Yerevan, Armenia

2. A portion of a DFBS plate Jenam 2007, Yerevan, Armenia

3. Overview of the DFBS • The DFBS is the digitized version of the First Byurakan Survey. • The plates were taken with the 102/120/240 cm Schmidt telescope of the Byurakan Observatory and a 1.5 degree objective prism, covering 4.1x4.1 degrees. • the emulsions used are mostly IIaF and IIF, giving a spectral coverage from 3800 to 7000 A. The emulsion cutoff and the wavelength compression of the prism in the Red produce a sharp Red-Head in the spectra, useful as fiducial mark. • For most plates an acceptable S/N is obtained up to B=16.5; stars brighter than B=13 are at least partially saturated. • The DFBS fills therefore the magnitude range between the Tycho catalogue and the Sloan Digital Sky Survey • The DFBS is presently available on line from the web page http://byurakan.phys.uniroma1.it/ Jenam 2007, Yerevan, Armenia

4. What is the DFBS • The DFBS is a database containing the digitized plates of the First Byurakan Survey, the individual spectra of the sources and their B and R magnitudes. • All these data can be retrieved using a dedicated WEB interface. • The spectra of the objects recorded on the plates were extracted using a catalogue driven procedure, based on the USNO-A2, cut at B=17.0 • In principle, therefore, no objects fainter than B=17 should be present in the DFBS database, but the accuracy of the USNO magnitudes is about 0.4 mag, so that objects actually fainter than the formal limit may be present, as well as objects actually brighter can be missed. Jenam 2007, Yerevan, Armenia

5. Spectra extraction • The extraction software, bSpec, looks for each USNO-A2 object according to its coordinates, finds the position of the Red-Head of the spectrum, compute the local background, finds the spectrum direction and extract the spectrum summing over a 5 pixels wide strip. • Due to the sensitivity function of the emulsion, and the variable dispersion of the prism with wavelength, the spectra have a typical “double bump” shape, with the red portion better exposed than the blu-violet one, and a marked dip in the green. Jenam 2007, Yerevan, Armenia

6. Typical spectra Jenam 2007, Yerevan, Armenia

7. Instrumental magnitudes • The red side of the spectrum has a sharp cutoff, due to the emulsion sensitivity and compression of the spectrum by the prism at long wavelengths. • This cutoff (Red border) is used as fiducial point on the extracted spectrum for wavelength calibration. • For a point-like source this cutoff is a few pixels distant from the light baricenter, depending on the seeing. • Instrumental Blue and Red magnitudes are computed integrating the spectrum over predefined pixel intervals counted from the cutoff. Jenam 2007, Yerevan, Armenia

8. Calibration of the DFBS magnitudes • These instrumental magnitudes are linked to the USNO scale using the objects in the central square degree of the plate: a linear fit is made in the magnitude range 12 to 16 both for the B and R magnitudes. • Overlapped objects and most discrepant objects are excluded from the fit and a second iteration gives the final calibration. • The typical slope of the fit is nearly 1, indicating that the transformation from plate transparency into intensity is generally good. • The rms deviation of this fit is 0.4 mag, mostly due to the intrinsic uncertainties of the USNO magnitudes. Jenam 2007, Yerevan, Armenia

9. Example of calibration plot • The calibration plot for each plate is available form the web interface • The plot gives also: the coefficients used to transform instrumental magnitudes into B and R (DFBS magnitudes); the peak of the luminosity function from the DFBS magnitudes in B and R Jenam 2007, Yerevan, Armenia

10. B mag calibration plot Jenam 2007, Yerevan, Armenia

11. R mag calibration plot • bb Jenam 2007, Yerevan, Armenia

12. Accuracy of the DFBS magnitudes • The internal accuracy of the DFBS magnitudes is better than the USNO one. • To measure this accuracy we used some sky areas covered by 3 or more plates and made a comparison of the magnitudes for all the stars. • Overlapped stars, or stars with a bad determination of the Red-Head, were excluded from the comparison. Jenam 2007, Yerevan, Armenia

13. Internal consistency from 6 plates, B band • FBS0150 • Fbs0913 • Fbs1296 • Fbs1366 • Fbs1393 • Fbs1399 Jenam 2007, Yerevan, Armenia

14. Internal consistency from 6 plates, R band • Fbs0150 • Fbs0913 • Fbs1296 • Fbs1366 • Fbs1393 • Fbs1399 Jenam 2007, Yerevan, Armenia

15. Reliability Limit The peak of the luminosity function of a plate can be regarded as its photometric reliability limit. Its value is given in the calibration plot for each plate. Typically it is about B=16.2 • Crosses: USNO-A2 • Line: DFBS • Only not overlapped stars are included Jenam 2007, Yerevan, Armenia

16. Magnitudes of AGNs For AGNs with detectable host galaxy on the POSS plates, the USNO (or GSC2) magnitudes are largely overestimated because are “isophotal diameter” sensitive. DFBS magnitudes are mainly “core sensitive” and therefore better indicative of the AGN luminosity. Jenam 2007, Yerevan, Armenia

17. How to query the DFBS • You can get data from the DFBS with any (recent) browser at http://byurakan.phys.uniroma1.it/ • You can ask for: A given plate number A given position in the sky A list of object coordinates • You can get: A quick look at a sky area, compared to the POSS, and at any spectrum extracted in that area (Explore); A FITS file containing a portion of a plate (GetImage) and all the extracted spectra in that sky area; A text file with a list of extracted objects and their B and R magnitudes derived from their spectra (GetSpectra); A text file with all the spectra in given sky area, from one plate or from all the plates covering that area (GetSpectra). Jenam 2007, Yerevan, Armenia

18. Tools for the DFBS data • To help the user (and ourselves!) managing the DFBS data, we prepared some software tools, written in FORTRAN77, which use the data as downloaded from the WEB interface: • COMPARE • CLASSIF • COMPARE is a tool to look for variable sources in a list of sources detected in at least 2 plates. • CLASSIF is a tool to perform an automatic spectral classification aimed at finding “peculiar” objects. • Both codes can be downloaded from the DFBS web page. Jenam 2007, Yerevan, Armenia

19. Search for Variable Objects the FORTRAN program, COMPARE, has been developed, which uses the output of the DFBS, and makes the following operations: • matches the objects by name • For each object computes: • the average B and R magnitudes , • the rms deviation (dev) of the B and R magnitudes, • the correlation between B and R magnitude. • Then computes, for each magnitude range, the average rms deviation (DEV) and the spread (SIG) around this value • For each star computes a variability factor • sigma=(dev-DEV)/SIG • Stars with sigma > 3 AND with positive (>0.7) correlation between B and R are selected as variable candidates. Jenam 2007, Yerevan, Armenia

20. Trial area • We made a search for variable objects using a trial sky area of about 1 square degrees (RA 01h26m, DEC +27d16m), covered by 6 plates (0150, 0913, 1296, 1366, 1393, 1399). • Sources where collected from the DFBS web interface (GetSpectra). • Two variables are known in this area EH Psc (B = 12, a semiregular Red Star) and BN Psc (B = 16, an eclipse variable). Jenam 2007, Yerevan, Armenia

21. Rms deviation for magnitude bins • Dots: average rms deviation for each magnitude bin • Line: 3 sigma limit for each magnitude bin • Stars must be above the line to be candidate variables Jenam 2007, Yerevan, Armenia

22. The sigma-correlation plot Stars in the upper right area are selected as candidate variables. Jenam 2007, Yerevan, Armenia

23. Results • EH Psc showed a variation in B of 0.5 mag, while it was saturated in R. • BN Psc was too faint to show reliable variations in B. • Two other stars showed variations, but were recognized as plate defects. Jenam 2007, Yerevan, Armenia

24. CLASSIF • a FORTAN code, CLASSIF, makes spectral templates from the spectra of a given plate. The following operations are performed: • a selection is made of well-exposed spectra both in the Red and the Blue • the B-R color index is used to average similar spectra in bins 0.2 mag wide; • the most discrepant spectra for each B-R bin are excluded and a new average spectrum is computed; the rms deviation for each bin is computed; • each spectrum of the plate is compared with the grid of templates and the best fitting is selected: if the deviation from the template is too large the spectrum is flagged as peculiar. Jenam 2007, Yerevan, Armenia

25. Emulsions and classification • Due to differences in the spectral sensitivities of the emulsions used for the FBS plates, separate templates must be made for each emulsion type. • The most frequent types of emulsions are: • IIaF • IIAF • 103aF • II F Jenam 2007, Yerevan, Armenia

26. Spectral response Average spectra of stars with B-R=0.5 in plates of different emulsions IIaF and 103aF are very similar. IIF and IIaF are similar, but IIF are more red-sensitive. IIAF emulsions are markedly different in spectral shape. Jenam 2007, Yerevan, Armenia

27. Spectral shape of templates • Comparison of templates of a K-type star and an A- type star. • The most prominent features are due to the spectral sensitivity of the plate, NOT to real stellar features. • Narrow band indices must be built “AD HOC” to select a well defined class of object. Jenam 2007, Yerevan, Armenia

28. Check of spectral classification • A template is made averaging normalized spectra in a bin of Delta(B-R)=0.2 • The rms deviation between the template and each spectrum, used to build it, is about 15% • CLASSIF looks for the best fitting template of each spectrum on the plate, trying all the templates. • The spread in (B-R) within each class is consistent with the intrinsic spread in color and spectral noise Jenam 2007, Yerevan, Armenia

29. Templates stability • Templates for a given color index from different plates of the same emulsion are very similar, but not identical. • If the inclination of the spectrum is appreciably different for two plates, the length of the spectra is also different and produces a systematic effect on the templates. Jenam 2007, Yerevan, Armenia

30. Spectral consistency check The spectrum of the same object in two plates with the same emulsion Emulsion IIaF Plate fbs1387 (line) Plate fbs1476 (crosses) DFBSJ063002.4+690503 Jenam 2007, Yerevan, Armenia

31. The case of an AGN Mkn 1502=I-Zw 1 • Low-redshift AGN have some emission lines (H-beta, H-gamma, H-delta, [OII]3728) detectable. • H-alpha is missed in the Red-Bump • The UV excess is also detectable by comparison of a template with the same B-R color index. Line: template Squares: AGN Jenam 2007, Yerevan, Armenia

32. Conclusions • The computer codes presented are aimed at helping the general user to manage the data retrieved from the DFBS PHOTOMETRY • Large magnitude variations (about 1 mag) can be safely detected for objects < B=16. • AGN magnitudes are closer to the real nuclear value than those available from present public POSS-based catalogues. SPECTROSCOPY • Very strong emission line objects can be recognized • UV excesses can be detected in objective manner • Reliable classification can be obtained down to B=15.5 Jenam 2007, Yerevan, Armenia