Community-level emergency facilities are essential for disaster prevention and response, contributing significantly to urban health equity. This study aims to explore the multi-scale spatial pattern, multi-dimensional availability evaluation, and multi-scenario simulation optimization of nucleic acid testing sites during the later stage of COVID-19. At a macro scale, these sites demonstrated a "multi-core clustering" pattern, while at a micro-scale, they were community-centered, often located in areas with low traffic and better ventilation. Spatial accessibility analysis showed high accessibility in urban centers when only facility capacity was considered, though accessibility dropped relative to suburban areas when community population and time costs were included. Temporally, site availability followed an M-shaped distribution. This study optimized testing site locations, providing insights into equitable distribution and supply–demand balance for emergency facilities.