Recent world-wide quieting of seismic noise caused by COVID-19 lockdown measures has been observed by seismometers. However, current seismic network that has a few seismometers or none in a city-scale area is hard to reveal the spatiotemporal characteristic of seismic noise impacted by COVID-19 measures. Here, we show that a 5-km-long distributed acoustic sensing (DAS) array deployed in State College, PA is able to illuminate seismic noise variation in a broad bandwidth 0.01 - 100 Hz during March - June 2020. The temporal noise variation exhibits a ‘decrease-increase’ trend responding to ‘decrease-increase’ human activities caused by the COVID-19 measures - from stay-at-home to Phase Green. Our results reveal different types of human activities (including footsteps, road traffics, and machines) as noise sources, suggesting that DAS noise recordings using cite widely-installed infrastructure fiber optics could be used for quantifying the impact of COVID-19 measures on human activities in city block dimensions.