LSST will have a Science Data Quality Assessment (SDQA) subsystem for the assessment of the data products that will
be produced during the course of a 10 yr survey. The LSST will produce unprecedented volumes of astronomical data as
it surveys the accessible sky every few nights. The SDQA subsystem will enable comparisons of the science data with
expectations from prior experience and models, and with established requirements for the survey. While analogous
systems have been built for previous large astronomical surveys, SDQA for LSST must meet a unique combination of
challenges. Chief among them will be the extraordinary data rate and volume, which restricts the bulk of the quality
computations to the automated processing stages, as revisiting the pixels for a post-facto evaluation is prohibitively
expensive. The identification of appropriate scientific metrics is driven by the breadth of the expected science, the scope
of the time-domain survey, the need to tap the widest possible pool of scientific expertise, and the historical tendency of
new quality metrics to be crafted and refined as experience grows. Prior experience suggests that contemplative, off-line
quality analyses are essential to distilling new automated quality metrics, so the SDQA architecture must support
integrability with a variety of custom and community-based tools, and be flexible to embrace evolving QA demands.
Finally, the time-domain nature of LSST means every exposure may be useful for some scientific purpose, so the model
of quality thresholds must be sufficiently rich to reflect the quality demands of diverse science aims.