Abstract
ABSTRACT Low cost atmospheric composition sensors are increasingly used in both air quality research and in air pollution monitoring. Those focused on particulates have generally rather rapid (T90) response times due to their reliance on light scattering methods. However, those electrochemical and NDIR sensors targeting trace gases such as, respectively, CO and CO2 typically respond very much slower, with T90 response times that can be further lengthened by the rate at which sample exchange occurs in the measurement cell. Data from these types of low cost sensors are often used to derive emission ratios [ERs] of two simultaneously measured atmospheric species, yet ER derivations made using data from sensors having dissimilar T90 values can be problematic, because any rapid change in the pollutant concentration can mean one sensor responding faster to the changing conditions than the other. Such situations are typical within biomass burning plumes, yet where such ER assessments are routinely required to generate the emissions factors (EFs) needed for biomass burning emissions calculations. Here we confirm that the ERs coming from such analyses can be strongly biased if differential sensor T90 effects remain unadjusted for, and we demonstrate a simple mathematical approach for undertaking this adjustment. The method is able to take data from a sensor taken with a particular T90 value and simulate that which would be collected from a sensor targeting the same species but with a different T90 value, and if the output T90 is selected to match that of the companion sensor measuring the second species then the two measures with the same effective T90 values can be used to generate unbiased emissions ratios and emissions factors. We demonstrate our approach on simulated data with known T90 response times, and then apply it to real data from low-cost sensors collected within biomass burning plumes, including those installed in a sampling system that introduced further – and unknown – additional measurement lag. In the latter case we find that our mathematical approach to the post-measurement T90 adjustment is still able to derive ERs from the low-cost sensor data that are comparable to those derived from data collected with a very high precision laser absorption spectrometer that measures the two species exactly simultaneously. Our methodology is thus well suited to the problem of deriving emissions ratios and emission factors from data collected by low-cost sensing systems deployed in rapidly changing pollutant plumes.
Highlights
Low cost trace gas sensors are increasingly used in both industry and environmental science for assessing the concentrations of target species
The method is able to take data from a sensor taken with a particular T90 value and simulate that which would be collected from a sensor targeting the same species but with a different T90 value, and if the output T90 is selected to match that of the companion sensor measuring the second species the two measures with the same effective T90 values can be used to generate unbiased emissions ratios and emissions factors
We demonstrate our approach on simulated data with known T90 response times, and apply it to real data from low-cost sensors collected within biomass burning plumes, including those installed in a sampling system that introduced further – and unknown – additional measurement lag
Summary
Low cost trace gas sensors are increasingly used in both industry and environmental science for assessing the concentrations of target species Such measurements typically use either electrochemical (e.g., for CO, SO2, HS, NO2; Zampolli et al, 2004; Mead et al, 2013; Piedrahita et al, 2014) or non-dispersive infrared (NDIR) methods (e.g., for CO2; Yasuda et al, 2012; Piedrahita et al, 2014), but in environments characterised by rapidly changing atmospheric concentrations there can be problems introduced when comparing data from different sensors having dissimilar sensor response times. This ISensor(t) is ascertained from the final reading obtained after an exposure time significantly longer than the quoted T90 response time (i.e., typically for many minutes), and depends on the ‘slowness factor’ [F, unitless], the sensor output at the previous timestep [ISensor(t – 1), ppmv], and a sensor baseline [ISensorBaseline(t), ppmv] which is generally pre-set in the lab to be near-zero for the types of electrochemical and NDIR sensors being considered here
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
