Data Verification Tools for Minimizing Management Costs of Dense Air-Quality Monitoring Networks

Georgia Miskell,Jennifer Salmond,Maryam Alavi-Shoshtari,Mark Bart,Bruce Ainslie,Stuart Grange,Ian G McKendry,Geoff S Henshaw,David E Williams

DOI: https://doi.org/10.1021/acs.est.5b04421

2016-01-19

Abstract:Aiming at minimizing the costs, both of capital expenditure and maintenance, of an extensive air-quality measurement network, we present simple statistical methods that do not require extensive training data sets for automated real-time verification of the reliability of data delivered by a spatially dense hybrid network of both low-cost and reference ozone measurement instruments. Ozone is a pollutant that has a relatively smooth spatial spread over a large scale although there can be significant small-scale variations. We take advantage of these characteristics and demonstrate detection of instrument calibration drift within a few days using a rolling 72 h comparison of hourly averaged data from the test instrument with that from suitably defined proxies. We define the required characteristics of the proxy measurements by working from a definition of the network purpose and specification, in this case reliable determination of the proportion of hourly averaged ozone measurements that are above a threshold in any given day, and detection of calibration drift of greater than ±30% in slope or ±5 parts-per-billion in offset. By analyzing results of a study of an extensive deployment of low-cost instruments in the Lower Fraser Valley, we demonstrate that proxies can be established using land-use criteria and that simple statistical comparisons can identify low-cost instruments that are not stable and therefore need replacing. We propose that a minimal set of compliant reference instruments can be used to verify the reliability of data from a much more extensive network of low-cost devices.