Performance of a multi-sensor system for ground and aerial sampling of pollutants in livestock farms

The dynamic characterization of emissions in terms of horizontal and vertical dispersion of airborne pollutants from open sources poses measurement and computational challenges. This study aimed to conduct the performance assessment of a self-designed multi-sensor system, the “G-eko 1.0”, for fixed and mobile applications such as indoor and outdoor ground measurements, on vehicles or drones. The G-eko 1.0 hosted selected low-cost sensors for measuring air pollutants deriving from livestock farming activities: a non-dispersive infrared sensor for carbon dioxide, an electrochemical sensor for sulphur dioxide, a metal-oxide sensor for natural gases and methane, and an optical sensor for particulate matter (PM2.5, PM10). A Raspberry Pi 3 module controlled data management and storage. The sensors were tested in a laboratory under controlled conditions for their accuracy. Dust was injected in an airtight chamber, and measurements obtained from the optical sensor were compared with those from a DustTrakTM 8533 Aerosol Monitor. For gaseous pollutants, tests were conducted in an airtight chamber with controlled artificial atmosphere at standard temperature and humidity conditions. Measurements from the G-eko 1.0 and the reference instrumentation were compared, and regression equations were estimated to improve the accuracy of the sensors' measurements. The results showed a modest average accuracy of the raw measurements provided by the sensors. Nevertheless, the mean percentage errors of the CO2, SO2, CH4, PM2.5 and PM10 sensor data predicted from the fitted regressions were substantially lower. The study revealed that monitoring selected airborne pollutants using cost-effective commercial sensors is a possible solution. However, acquiring reliable measurements requires calibration under a wide range of controlled environmental conditions.