In collaboration with Citi-Sense-MOB, 35 sensor units were tested and deployed in fixed locations in Oslo. The sensors distributed in 17 kindergartens and 7 locations in the city using for instance light poles. The sensor platform has been manufactured by GEOTECH (www.geotech.uk) and measures NO, NO2, O3, CO, Noise, Total Particle Count, Temperature, Relative Humidity and Atmospheric Pressure. The data is provided as hourly averages. The sensor platform has a lithium battery with capacity for approximately 6 months. The independence of power supply facilitates their location in areas where power supply it is not accessible.

Some regular citizen have carried wearable sensors in Oslo. These wearable sensors measure 3 pollutant gases: CO, NO₂ and O₃, as well as temperature and relative humidity. The sensor platform communicates using Bluetooth with an Android phone. The GPS location is obtained from the phone. Other data as accelerometer data is also collected. These data is then sent to the WFS server (main server) when there is network available.

The tracking service in mobile app - Sense City Air, enables the user to check the air quality for the locations following a movement of the user. Using this service, the user will be able to start, pause, re-start and stop the tracking. Once the track is stopped a summery about track will be displayed. In addition included locations within track and their respective calculated air quality levels are represented by the app as a coloured track. The track represents the path which the sensor has moved along, and its colour represents the calculated air quality at these locations. An update to this functionality is currently being looked into, related to some performance issue with the server.

We conducted an exhaustive evaluation of low-cost platforms. The evaluation included laboratory testing against reference instruments in a controlled environment and field testing against reference instruments (co-location) under a range of different environmental conditions (e.g., weather, traffic). The results show clearly that a good performance in the laboratory is not indicative of a good performance in outdoor real-world conditions. The coefficient of determination in laboratory was above r2>0.9 for all the gases, while in field it showed a significant decrease, especially for NO₂ and O₃. This decrease in the performance is most probably due to interferences with temperature and relative humidity. Advances in algorithms to process the data might result in a higher performance of the sensors. Results also showed a high variability in the sensor responses from the same sensor type, making necessary to characterize sensor by sensor, even if by the same producer.

The high variability in individual sensor performance, as well as the variability in the performance depending on weather conditions or changes in emission patterns, etc. makes low-cost platforms difficult to use for applications when high data quality is necessary (i.e. air quality surveillance for regulation purposes). Low-cost sensors is a promising technology, with a rapid evolution in the market and the performance of the sensors is improving. However, the sensors are still in a research phase that requires an exhaustive testing and comprehension of the performance of each individual sensor platform before they can be deployed.

For more information, please contact Dr. Nuria Castell from NILU, Kjeller, Norway, email: ncb@nilu.no