Air filtration systems are used in industrial facilities to prevent the release of contaminants and are indicative of ongoing operations. We are investigating the potential of using seismic measurements to identify and characterize the operational state of filtration systems in a nuclear facility. We are using seismic data collected at a nuclear facility in Texas (US) that uses multiple air filtration systems for routine operations. The data were collected at distances of few hundred meters and different azimuths from the units. The seismic data were recorded with threecomponent sensors with a sampling rate of 1000 samples per second. We are evaluating the use of cross-coherence and bi-spectral analysis to identify the spectral regions excited by the operation of the scrubbers. Initial results show a mixture of spectral lines with frequencies below 50 Hz and broadband signals between 200 and 400 Hz. We are also using frequency dependent polarization and cross-bearing analyses to distinguish between multiple machines distributed in the facility and operating asynchronously. The use of specific machines in industrial environments can provide information about the types of operations and level of activity of a facility. Our approach can provide a low-profile and low maintenance tool to identify and characterize operations in facilities with limited access.