On the night of Feb 05/06, 2024, under good conditions, I attempted to acquire images of the starfield in which the solar eclipse will occur on Apr 8, 2024. This was my third attempt. At last, I succeeded! Hooray!
During the eclipse, the Sun will be located at (J2000)
RA = 01:10:32 Dec = +07:28:47
= 17.6333 = +7.4797
Tonight, my plan was to use my Nikon Z6II camera together with a different lens: Nikkor 300mm f/4.5 Ai, serial 537255 (circa 1980). This combination yields a field of view around 5 x 3.3 degrees, like this:
I mounted this lens on a dovetail plate on the 14-inch telescope. When I twisted the focus ring to the "infinity" position, stars were indeed in focus. Below is a closeup of one unsaturated star near the center of the frame in a 1-second image, together with a radial profile (units on x-axis are pixels).
Due to the location of the dovetail plate on the telescope, images will always have the long axis in the north-south direction. Below is an example; the peculiar choice of orientation of the image on the left is designed to match the way images appear in my debugging display. The normal "North up, East left" orientation is shown at right.
I took images at ISO 800, with exposure times ranging from 1 to 30 seconds. I converted the 14-bit RAW images to FITS using the nUFRaw software. Each RAW image produced three 16-bit FITS images, one for the red-sensitive pixels, one for the green-sensitive pixels, one for the blue-sensitive pixels. I'll concentrate on images based on the the green-sensitive pixels.
There was quite a bit of vignetting. The image below shows a second-order polynomial fitted to the background light in a 10-second image of the field. The colormap goes from black at 550 ADU to white at 1350 ADU; in other words, the corners record a bit less than half the light of pixels at the center.
Since I am interested in astrometry, not photometry, I performed only rudimentary cleaning to these images:
The result showed some residual background variations with a third-order shape (bright in the corners, bright in the center, faint between), but my star-finding algorithm wasn't bothered. I assume that the effect of these very large-scale gradients on the astrometric measurements of individual stars is negligible.
I used programs in the XVista suite to find stars in each image and measure their positions and instrumental magnitudes. Using a threshold of peaks at least 20 times the stdev above the background, I found 250-300 stars in the 1-second exposures and 500-520 stars in the 10-second exposures.
In order to perform astrometric calibration, I grabbed a set of stars from the Gaia DR3 catalog in Vizier. I then used the match software to match the stars detected in each image against those in the Gaia catalog. It was necessary to choose a plate model including third-order terms in order to achieve a good match; models with first-order only, or first- and second-order only, yielded matches only over portions of the field.
The plate scale from these solutions is 4.004 arcsec per pixel, based on over 200 matched stars in each image. The typical residuals are about 0.5 to 0.6 arcseconds. I have yet to check for positional correlations in the residuals, though I suspect there will be patterns visible on large scales.
I took 3 exposures at 1 second each, and 4 exposures at 10 seconds each. Might it be possible to average the positions from multiple exposures in order to decrease the random errors in each measurement? I think the answer is "yes". When I computed the mean position of each star in each set (1-sec and 10-sec) of images, I found the standard deviation from the mean was smaller for bright stars and a bit larger for faint stars, as one would expect. In the case of 10-second exposures, stars with (8 < mag < 10) had a stdev from mean position of 0.26 arcsec in RA and 0.21 arcsec in Dec.
Since I expect the distortions due to the gravitational influence of the Sun to be around 0.3 to 1.0 arcseconds for a large number of stars (see Gravitational Starlight Deflection Measurements during the 21 August 2017 Total Solar Eclipse ) I am somewhat optimistic that images of this quality may be able to detect it.