Volatile organic compounds (VOCs) are emitted both naturally from vegetation, and from human activities. These compounds play an important role in controlling the regional and global concentration of ozone, aerosol and methane, all of which are key players in climate change and air quality degradation. Current detection techniques, such as gas chromatography, provide detailed information of the concentrations of individual VOCs. The existing body of literature suggests that we have a reasonable understanding of the dominant VOCs in different regions of the globe, but representing the spatial and temporal variability of these compounds in models remains challenging. Recent developments in cheap, small, lower power sensors for atmospheric gas detection offer a potential opportunity to address this measurement gap, as such instruments could readily be deployed over long timescales or as a distributed network over large spatial scales. This fellowship used the fellows experience in instrument development combined with expertise in VOC detection, electronic engineering and advanced statistical techniques at the University of York to develop a prototype instrument that uses low cost sensors for the study of VOC variability (in particular those from forested ecosystems). The individual objectives that were identified to achieve this goal were:
• Laboratory characterisation of low cost VOC sensors.
• Develop an observational methodology that could be applied over large spatial and/or temporal scales.
• Develop a data analysis methodology that maximises the potential of low cost sensor technologies.
• Demonstrate the new methodology alongside established techniques.
Through an extensive program of laboratory experiments a range of commercially available low cost VOC sensors were characterized for atmospherically relevant conditions. Sensor signal stability and detection limits where improved through improvements in electronics and signal handling and a sensor was identified that will enable the study of atmospheric VOC variability. During these laboratory experiments it was identified that in order to provide the required information on total VOC variability, it will be necessary to correct for multiple interferences on the sensor signal from other atmospheric constituents. The complexity of these interferences prevent simple corrections, and it was decided that the most practical approach, while maintaining the instruments low cost, was to use signals from a range of different low cost sensors and an advanced regression algorithm to separate the signals from different atmospheric constituents. Although the findings from these laboratory experiments, and the development of new statistical tools, delayed the overall project, the public interest in low cost sensors for atmospheric measurement meant the results proved very timely and resulted in multiple invited presentations and publications from the fellow. Further experiments focused on improving the reliability of data from low cost sensor devices by targeting inter-sensor variability.
