Known Issue: Undetected Blends

Synopsis: Plates with low spatial resolution may blend multiple sources into a single image. This can cause seeming brightenings that are not real, and the effect can manifest over both long and short timescales.


This issue is not at all unique to DASCH. However, one must be especially careful about it when working with DASCH data because of the heterogeneity of the DASCH collection: different plates have spatial resolutions that vary by a factor of 30 or more.

Furthermore, thanks to DASCH’s long-term multi-observatory coverage, the prevalence of blends varies systematically with time, coupling into the lightcurves. For instance, in the 1940’s and 1950’s, many very-low-resolution “meteor” plates were obtained as parts of series bi, fa, ka, pz, and others. A naive lightcurve plot for a source with a nearby neighbor may seem to show a spurious brightening or enhanced variability during these decades, due to the increased number of blended measurements.

To assess the significance of this effect for a target of interest, consider plotting a 2D histogram of fitted source positions, as well as brightness against position. If the fitted positions have an asymmetric shape, that may indicate that blends are “pulling” the source position away from its actual location towards the position of the centroid of the blend. Examining deep, high-resolution imagery of the target field may also be helpful, of course.

Depending on your scientific use case, you may wish to simply ignore data from low-resolution plates such as the “meteor” series (plate scales > 900 arcsec/mm) and/or the “patrol” series (plate scales > 400 arcsec/mm).

Back up to the list of known issues.