Processes and Methods
The methods and processes developed for this project center around three broad goals:
- Utilizing accessible tools and equipment: In practice, this means reasonably priced products that can be easily purchased and used off the shelf, with no customization or specialized skills necessary to use them.
- Producing digital files that meet archival standards, but are also of a quality that they are scientifically useful, particularly for time domain astronomy.
- Accessibility to the materials for both those who care for the objects as well as for those who want to use them.
Perhaps the most important part of digitizing glass plates is making sure they are preserved in such a way that they can be digitized! As a material, glass is physically fragile, and photographic emulsion is also fragile and particularly susceptible to flaking, scratches, and interaction with oils from skin. Following recommendations from the Library Conservation Department as well as the National Archives, we care for the plates by:
- Storing them “upright” – on their edge – rather than flat.
- Handling the plates only while wearing non-vinyl gloves.
- Handling the plates only by two edges.
- Ensuring workspaces are clean of debris, by utilizing a brush, canned air, or a microfiber cloth depending on the surface
- Always placing plates emulsion side up.
- Utilizing paper enclosures, carefully inserting plates with the emulsion facing away from any seams.
Currently, most of the plates at the University are still stored in their original envelopes, or at least envelopes originally used at Yerkes Observatory, and are housed on off-the-shelf metal shelving units in climate controlled environments. In the future, we hope to rehouse the plates stored at the University in more preservation-friendly storage, including buffered four-flap enclosures that will allow access with less sliding, and document boxes with interleaved corrugated boards for support/stability. Frankly, the storage at Yerkes Observatory is less flexible both in terms of space and climate control. These plates are also stored on edge on off the shelf metal shelving in original envelopes, and it is likely that any changes to this storage will happen on a case-by-case basis.
It is important to us to follow digitization standards that meet both archival and scientific standards. This sometimes means that we are “over-digitizing” for one purpose, but aim to never “under-digitize” for either. For archival standards, we follow the Federal Agencies Digital Guidelines Initiative (FADGI), as well as the internal standards set by the Digitization Department at the University of Chicago Library. For scientific purposes, we adjust our specifications based on the size of the plate being digitized.
For example, when working with a 12 x 12 inch plate, we scan the full plate in color at 600 dpi, 48-bit, TIFFs, and then rotated 90º. This full scan not only captures all of the object, by scanning in color we also capture any markings or color on the plate. Then, we also scan the central 6 x 6 inches, straight and rotated 90º, at 1200 dpi, maintaining all of the other specs. At this spatial resolution on a 12 x 12 plate, stars are sampled by about 5px in one dimension, which we have found to be sufficient for astrometric and photometric measurement. Eliminating the edges of the plate also minimizes the optical aberrations seen around the edges of the image. Scanning at higher dpi minimally increased precision in measurement, but significantly increased file size and scan time. We continue to explore scanning specifications for different types of plates. Initial results indicate that 3.25 x 4.25 inch sky survey plates can be scanned at 2400 dpi for scientific use, and student researchers are also working to determine scanning specifications for spectra plates. No matter the plate type, we scan with the emulsion side of the plate up to prevent damage from unintended movement on the scanner platen.
The specifications we use for scanning 12 x 12 inch plates in both Epson Scan 2 and Silverfast software are:
- Mode: Photo mode
- Document source: Transparency Unit
- Document type: Color positive film
- Image type: 48-bit color
- Resolution: 600 dpi or 1200
- Scanning quality: High
- Color Management: None
- Image format: tiff
It is important to note that we also make several other files, including scans of the front and back of the plate envelope and any ephemera found with the plate and a scan of a step wedge for calibration purposes. The envelopes and ephemera are scanned at these specifications:
- Mode: Photo Mode
- Document source: Scanner Glass
- Document type: Reflective
- Image type: 24 color
- Resolution: 300 dpi
- Scanning quality: High
- Color Management: Open and select None from the drop down
- Image format: tiff
We also scan a calibrated Stouffer 21-step density wedge, to calibrate scanner output. For scientific purposes, it calibrates scanner output to transmission via a transformation equation. Research determined that the scanner output is stable enough for us to scan the step wedge at the beginning of a scanning session, and the results from that step wedge can be applied to all scans made during that session. More information about the step wedge and transformation equation can be found in Equipment and Tools. For archival purposes however, the step wedge is used to determine values of grayscale in the image, so we aim to always scan the step wedge in the same image as the full plate scan to make that comparison easier.
File management is an important part of the process. Because we create several different scans of each plate (at least 6: envelope front, envelope back, plate full, plate full rotated, plate center, plate center rotated) and several other derivative files, all files are named with a specific standard: PLATE NUMBER_identifier. So for example, the full scan of plate 10B-161 would be named 10B-161.tif, the full rotated scan of that plate would be 10B-161_R.tif, and the center scan would be 10B-161_cen.tif. While this naming convention works for our team, we recommend others develop a convention that works within your existing data/file management system.
Converting a scan to FITS
Finally, to complete scientific analysis, we must convert the tiff files to astronomy-centered Flexible Image Transport System (FITS) files in order to receive a World Coordinate System (WCS) solution. We utilize the free software ImageJ.
Within ImageJ, we split the color channels, divide them each by 3, and recombine them before saving the resulting image as a FITS file. While color is important for archival purposes, it is less critical for scientific analysis, so this process produces a grayscale FITS file that is ⅓ the size and maintains all of the characteristics of the original tiff.
This FITS file is then uploaded to nova.astrometry.net (hereafter referred to as Astrometry.net), which matches stars identified in the USNO-B catalog with the uploaded file. This orients our image to its actual coordinates in the sky. While it is not required to upload FITS files to Astrometry.net, we do so because it increases the matching speed. It is critical to remember to use Astrometry.net's Advanced settings to select “Invert,” as we are working with negatives. If known, we also set other parameters which can increase the speed of the solution, but this is not required.
After a first pass, we download the “New Image” FITS solution provided, and run it through the system again. This increases the accuracy of both matches and RMS residuals enough that it is worth the effort. We have found that a third pass does not provide either better matches or residuals. The second pass New Image FITS solution is then downloaded and used for all further calculations and extrapolation.
Once these steps are complete, we must transform our file.
Support for this project comes from the National Science Foundation (Grant AST-2101781), University of Chicago College Innovation Fund, John Crerar Foundation, Kathleen and Howard Zar Science Library Fund, Institute on the Formation of Knowledge, and Yerkes Future Foundation.