Calibration of the Arcetri Solar Archive Imagessait.oat.ts.astro.it/MmSAI/76/PDF/977.pdf · Mem. S.A.It. Vol. 76, 977 c SAIt 2005 Memorie della Calibration of the Arcetri Solar Archive

Mem. S.A.It. Vol. 76, 977c© SAIt 2005 Memorie della

Calibration of the ArcetriSolar Archive Images

F. Giorgi1, I. Ermolli1, M. Centrone1, and E. Marchei1,2

1 INAF Osservatorio Astronomico di Roma, Italy;e-mail: [email protected]

2 CVS Centro per lo Studio della Variabilita del Sole, Italy

Abstract. The non linear characteristic curve representing the relation between intensityand the measured density obtained by the digitization process constitutes one of the majorproblems associated with quantitative photometry based on the use of photographic plates.This paper briefly describes the pre-processing and photographic calibration proceduresapplied to the images obtained by the digitization of the Arcetri solar archive. This archivecontains about 13000 plates of full-disk Ca II K2,3 and Hα spectroheliograms acquired at theArcetri Astrophysical Observatory during about 5000 observing days from 1925 to 1974.

Key words. Sun: historical data – Sun: variability – Sun: magnetism

1. Introduction

The results obtained by the synoptic obser-vations of the solar disk stored in historicarchives provide a precious temporal baselineof several decades for the study of many so-lar properties. For example, these observationspermit a variety of retrospective analyses of thestate of solar magnetism, as well as they allowto deepen the role of the Sun on the climatechange noticed during the last century.In the last few years the increasing inter-est toward the recovery of early observationsand new hardware resources drived on sev-eral projects to digitize existing solar archives.These projects permit the preservation of thearchives content, but to allow the scientificexploitation of them, it is also indispensableto develop a dedicated and optimized pro-

Send offprint requests to: F. Giorgi

Table 1. Transmission values of the Zeiss K 58step filter.

λ= 6000Å 100 72 45 28 19 11λ= 4000Å 100 68 42 29 20 14

cessing for the images produced. This pa-per summarizes the pre-processing and pho-tographic calibration procedures applied tothe images obtained by the digitization ofthe Arcetri Astrophysical Observatory solararchive. Details about the digitization workcarried out by the CVS (Centre for Study theVariability of the Sun) project at the RomeAstronomical Observatory, with the sponsor-ship of Regione Lazio, can be found inMarchei et al. (2005).

978 Giorgi et al.: Calibration of the Arcetri Solar Archive

2. The Arcetri Solar Archive

The Arcetri archive contains 12917 plates offull-disk CaII K2,3 and Hα spectroheliogramsacquired during 5042 observing days at theArcetri G.B. Donati tower from 1925 to 1974(Godoli e Righini 1950). One or more spec-troheliograms were daily obtained in the twospectral bands, with an average number ofobservations of more than 100 observationsper year for each spectral band. A descrip-tion of the Arcetri instrument charactersis-tics at the time of observations is given byMarchei et al. (2005). In brief, the archivemainly contains plates with a solar disk im-age of ≈ 6.5 cm corresponding to an imagescale of 0.033 mm/arcsec. The spectral win-dow was 0.3Å centered at the CaII K line core.However, it is worth noting that several instru-mental changes occured during the about fiftyyears of the spectroheliograph utilization. Forexample, on 23 Jul. 1938 started the testing ofan additional field lens to improve the imagedefinition. This lens has been definitively uti-lized starting from 20 Jan. 1939. In the sameperiod also started the utilization of filters setupon the first slit. An UG2 filter has been uti-lized for the K2,3 observations from 15 Oct. to30 Dec. 1938, then replaced with an UG3 filter.A RG2 filter has been utilized for the Hα obser-vations from 15 Oct. to 29 Jul. 1939. Moreover,on 25 May 1953 the position of the 2nd slit hasbeen modified, decreasing its position respectto the plate from 1.3 cm to 2 mm. This changeallowed a remarcable increase of the observa-tions monocromaticity and a decrease of thestray light.We note down all the improvements occuredbecause these mark a few discontinuities on thedata collection, that anyway exhibits a ratherremarkable homogeneity.More than 30 different emulsions were usedalong the whole period of observations. Thespectroheliograms obtained since 22 Feb. 1938have been calibrated through a Zeiss K 58 stepfilter, whose transmissions are listed in table 1.At first, only one photometric mark was storedon the plates, by placing the Zeiss filter uponthe first slit and gathering the radiation emittedfrom the solar disk center. Starting from Sept.

1947, two photometric marks were stored onthe plates by changing the width of the first slit.From Aug. 1948, there have been three pho-tometric marks stored on each plate: the firstobtained with the radiation emitted in the CaIIK2,3 or Hα from the solar disk center, the sec-ond one in the radiation of the near-continuumto the line at issue, and the third one alwaysin the same continuum, but by doubling thewidth of the first slit (Fig. 1). The velocityof driving motors which translate the photo-graphic plate during the wedge exposition wasequal to that utilized during the solar imageexposition. Whether more observations wereobtained during the same day, the calibrationwedge was stored only on the lattest plate ex-posed. However, being all the plates obtainedin the same day photographically processed atthe same time, a sole calibration wedge can beused for all the specthroheliograms of the sameday and photometric measurements can be per-formed as well through all the plates obtainedon the same day.The digitization of the archive has been car-ried out through the scanner Epson Expression1680 Pro, with the setting 1200 dpi and 16bit significant data. The instrument and set-tings allowed the scanning of plates in groupsof four for most of the archive; each scanningproduced a tiff format 8435×11153 pixel im-age (179 Mb). The spatial resolution due to thedigitization is 0.62 arcsec/pixel. To get an ho-mogeneous set of data, both the solar disk andthe calibration step wedge, when stored on theplate, were digitized on the same image.

3. Image pre-processing

We describe in the following the proceduresdeveloped to analyze the images obtained bythe digitizztion of the K2,3 spechtrohliograms.For all these images, we ran a pre-calibration,semi-automatic procedure which produces im-ages corrected for the scanner flat field. The flatfield response of the scanner has been obtainedat the beginning of each digitization session bytaking an average image of the scanner fluores-cent screen.Then the sub-image corresponding to eachplate has been singled out; couples of im-

Giorgi et al.: Calibration of the Arcetri Solar Archive 979

Fig. 1. Example of the images obtained by thedigitization. Notice the lack of homogeneity inthe upper calibration wedge, due to the inten-sity pattern on the solar disk center.

ages for each individual spectroheliogram,both in .fits (1020x1360 pixel) and .tiff formats(510x680 pixel) have been stored. The storingof 2k×2k fits format images is now being com-pleted.Whenever the calibration wedge is includedin the plate, the pre-processing procedure alsoprovides from 7 to 21 triplets of average andmedian transparency (or transmission, T ) val-ues, and the related standard deviations, on thewedge steps. These values are used to evaluatethe calibration curve.Given the large quantity of digitized plates wefirst tried to fully automatize the wedge iden-tification procedure. However, the plates arevery different one from the other: the Sun andthe wedge show different shades and positionsagainst the plate background, and their dimen-sions differ. We can notice a series of defects,such as lines, large-scale lack of homogene-ity, scratches, and over-exposition, and all thismakes it difficult to run successfully a com-pletely automated procedure. So, the proceduredeveloped for the identification of the subarray(3500x3500 pixel) containing the solar disk in

Fig. 2. The calibration curve is obtained as a200×2 element matrix showing the correspon-dence between density D = Log10(1/T ) (or-dinate) and intensity (abscissa), as computedthrough a least squares third degree polynomialfit (bottom) of the suitably re-arranged (mid-dle) measured density values D for the variousexpositions of the wedge steps (top).

980 Giorgi et al.: Calibration of the Arcetri Solar Archive

the image, and for the calculation of median,average and standard deviation of transparencyvalues in each step of the calibration wedge re-quires a check control.The procedure single out a sub-array for eachstep, then calculates the transparency valueslisted above.

4. Photographic calibration

The determination of the linear characteristiccurve representing the relation between inten-sity and the measured density obtained by thedigitization process (Dainty and Shaw 1974)is considered in the following.Whether the aim of the image analysis is an ac-curate measurement of the location and extentof spot and bright plages, a sophisticated pho-tometric calibration is not necessary. To carryout these measurements we need only to beable to define edges of spot and plages by sometransparency thresholds and to be sure that thelocation and area measurements derived are abit sensitive to the possible calibration errors.However, while spots (as well as false spotsproduced by dust effects) and plages can berather easily detected by the way of few thresh-olds in ’transparency’, with respect to the lo-cal background of the solar disk, the contour ofplage remnants and of active network are verysensitive to the ’transparency’ threshold. Thenthe photographic calibration constitutes a keystep for an accuate image processing.Following Caccin et al. (1997), we evaluatethe calibration curve, through the average val-ues of transparency T and its standard devia-tion on the wedge steps. The calibration curveis obtained as a 200×2 element matrix show-ing the correspondence between density D =Log10(1/T ) and intensity, as computed through

a least squares third degree polynomial fit ofthe suitably re-arranged measured density val-ues D for the various expositions of the wedgesteps (Fig. 2). The curve computation has beenapplied to each plate that includes a calibrationwedge. An average curve obtained by severalcalculated curves, has been applied to all theplates stored without calibration wedges.

5. Summary

We briefly described the procedures for bothpre-processing and photographic calibrationof the images obtained by the digitizationof the Arcetri Solar Archive. As mentionedbefore more than half of the Arcetri platesinclude wedge calibrations. But this is not theterm of other existing historic solar archives.For example, the plates of the Mt Wilson andthe Kodaikanal archives, as well as those ofthe Sacramento Peak one obtained before1993, do not include wedge calibrations. TheArcetri archive thus allows the optimization ofcalibration methods that are not based on thetransparency values of exposed step-wedges,like those founded on the solar center to limbdarkening or on the transparency distributions.

References

Caccin, B., Ermolli, I., M. Fofi, & Sambuco,A.M. 1997, Solar Physics, 177, 275

Dainty, J.C. and Shaw, R. 1974, ImageScience, Academic (Press)

Godoli, G. and Righini, A. 1950, Mem. Soc.Astron. Ital. 21, 4

Marchei, E., et al. 2005, Mem. Soc. Astron.Ital., in press