Synopsis: The DASCH catalogs may contain multiple entries corresponding to a single astronomical source. When the lightcurve processing pipeline matches the detections on a plate against the catalogs, the measurement may get assigned to any one of the catalog entries, causing measurements of the same source to be split between different catalog entries.
Details
The primary DASCH source catalogs, APASS and ATLAS-refcat2, sometimes contain multiple entries corresponding to a single astronomical object. This is something that occurs in astronomical catalogs in general, and then in the particular case of DASCH the relatively poor spatial resolution of many of the photographic plates often results in high levels of blending, exacerbating the problem.
Although the DASCH analysis attempts to account for blending, it is not uncommon for DASCH source searches to turn up multiple catalog entries. For instance, a lightcurve search for the variable star PV Cep (RA 20:45:54, dec +67:57:38) in the APASS catalog returns two entries: N013332320656 at a nominal separation of 8 arcsec, and N01333232066 at a nominal separation of 17 arcsec. (These identifiers correspond to GSC2.2 HTM-6 designations, and differ only in their last digits.) These appear to be two records for the same object (PV Cep is a YSO with an outflow), and the two are certainly indistinguishable as far as most plates in DASCH are concerned.
During photometric processing of each plate, the detections from the mosaic image are matched to the reference catalogs. When there are multiple nearby candidates and the blend processing fails, the measurement will be assigned to one of the candidates in a quasi-random fashion, depending on which seems to be closest. The upshot is that the lightcurve measurements will be distributed unpredictably among the nearby sources.
Mitigation
If a lightcurve search turns up multiple catalog matches within the default 20
arcsec search radius, in many cases it is appropriate to merge the lists of
detections into a unified lightcurve. The daschlab system provides the routine
daschlab.lightcurves.merge() to assist with this; see the associated
HD 5501 tutorial. Keep in mind, however,
that the plates in DASCH vary widely in terms of both spatial resolution and
depth, so that a blend in one plate may be easily resolved into separate sources
on another. As usual, visual inspection of the relevant imagery is strongly
recommended.
As a general rule, the ATLAS-refcat2 catalog is superior to APASS when it comes to this issue, because it is based on Gaia astrometry. For instance, a lightcurve search for PV Cep using ATLAS turns up just one source, ATLAS2_980596492, at a nominal separation of zero arcsec. However, the ATLAS photometry suffers from false long-term features that make the data unsuitable for many science applications.