Experience with a low cost NO 2 /O 3 sensor network in Zurich, Switzerland Martin Steinbacher, Michael D. Mueller, Lukas Emmenegger, Christoph Hueglin Empa, Laboratory for Air Pollution / Environmental Technology, Duebendorf, Switzerland
Current AQ monitoring in Zurich - NO 2 in Zurich, 18.01.2016 08h Zurich, continuous analyzers 77 µg/m 3 24 µg/m 3 42 µg/m 3 NO 2 passive sampler network (2 week mean) 35 µg/m 3 Low-cost air quality sensors could contribute to: GoogleEarth 63 µg/m 3 more detailed information on air quality complement existing AQM sites replacement of passive samplers? µg/m 3 Input for advanced modelling (e.g. highly resolved statistical models)
Current AQ monitoring in Zurich - NO 2 in Zurich, 18.01.2016 08h Zurich, continuous analyzers 77 µg/m 3 24 µg/m 3 42 µg/m 3 NO 2 passive sampler network (2 week mean) 35 µg/m 3 NO 2 63 µg/m 3 GoogleEarth Mueller et al., Atmos. Env. 2015
NO 2 /O 3 sensor system Aircube (AC) 2x Aeroqual O 3 SM50 3x Alphasense NO 2 B42F temperature relative humidity GSM module for data transmission
Empa-SN + federal/municipal AQM stations Empa sensor network operating since June 2015. Sensor calibration facilities AQM sites provide range of pollutant concentration Test network for development of QA/QC algorithms NO2 passive samplers (two-week averages) GoogleEarth
Sensor locations BUE GES PFI WIN STB ETH
Sensor calibration NO2 raw data (Feb - May 2015) REFERENCE AIRCUBE calibration of individual sensors required NO2 calibrated data (Feb - May 2015) field calibration correlations between observations (e.g. T, O 3, NO 2 ) conditions given by location and time (season) use of a statistical sensor model description of sensor behavior in specific conditions (w.r.t. pollutant concentration, meteorology) calibrated data associated with larger uncertainties in strongly deviating conditions
Sensor calibration O3 raw data (Feb - May 2015) REFERENCE AIRCUBE calibration of individual sensors required O3 calibrated data (Feb - May 2015) field calibration correlations between observations (e.g. T, O 3, NO 2 ) conditions given by location and time (season) use of a statistical sensor model description of sensor behavior in specific conditions (w.r.t. pollutant concentration, meteorology) calibrated data associated with larger uncertainties in strongly deviating conditions
NO2 in Zurich Reference sites Sensor site 001 (Buerkliplatz, Zurich) correction of sensors based on data of reference sites
NO2 comparison: sensor vs. passive sampler Sensor site: Winterthurerstrasse Sensor data : «calibrated» AC NO 2 passive sampler measurements provide an independent check of the sensor readings / data processing strategies. PS 14 days means Sensor data: «calibrated» and corrected based on data of reference sites AC PS 14 days means
NO2 sensor data (one week time series) AQM sites
NO2 sensor data (one week time series) AQM sites [AC ~ 1 year in operation] AC
Diurnal variation of NO2 weekday weekend AQM site (background) weekday weekend AC (background)
Diurnal variation of NO2 weekday weekend AQM site (kerbsite) weekday weekend AC (kerbsite)
Conclusions low-cost sensors can be suitable for air quality (urban) measurements operation of low-cost sensors is demanding sensor model / calibration QA/QC in air quality sensor networks needs to be established experience in long-term sensor operation is still limited achieved precision (~ 5ppb) is still far beyond the GAW quality objectives low-cost sensors could be a feasible alternative to passive samplers in the future