AQMesh performance and accuracy
AQMesh is the only small sensor air quality monitoring system that can demonstrate its performance across extensive global co-location comparison trials now spanning more than 6 years. AQMesh pods have been operating out in the field since 2013 next to certified reference equipment, in real applications, in real conditions, across all environments, throughout all seasons, for months rather than weeks or days. The data from the reference station and AQMesh pods are compared to provide a correlation coefficient (R2). A higher correlation coefficient between AQMesh and the reference data shows a close match in readings from the two types of equipment (an R2 value of 1.0 is a perfect relationship, and is unlikely to be achieved even by two co-located reference instruments; whereas an R2 of 0.0 would indicate no relationship at all between the instruments).
AQMesh has been in development and available for sale since January 2013. During that time, improvements in electrochemical sensors and the unique, carefully developed, fully traceable AQMesh processing algorithms have given a dramatic improvement in performance and accuracy with AQMesh, initially seeing low R2 values for NO2 in the early days of 2013, right up to R2 values >0.8 and >0.9 today.
The timeline below shows that the R2 value achieved for NO2 with AQMesh has shown a significant improvement through development since AQMesh was first launched in 2013.
In addition, please browse the wealth of co-location study results demonstrating the high R2 values achieved across all parameters in a wide variety of environments.
|Version||Production date||NO2 sensor||Characterisation||Online processing||Typical R2*||Typical MAE**|
|V3.0||Jan 2013 – Dec 2014||Significant O3 cross-gas effect||Manufacturer’s data||Correction for cross-gas effects and environmental factors||0-0.03||Requires scaling|
|V3.5||Jan 2015 – Oct 2015||O3-filtered||As V3.0||Correction for cross-gas effects and environmental factors||0.1-0.7||Requires scaling|
|V4.0||Jan 2015 – Oct 2016||As V3.5||Manufacturer’s data and characterisation at factory||Correction for cross-gas effects and environmental factors||0.5-0.8||Requires scaling|
|V4.1||Limited release||As V4.0||Factory characterisation and quality check||More sophisticated corrected for cross-gas effects and environmental factors||0.7-.095||Dependent on variance between sensors|
|V4.2||Oct 2016 – Dec 2018||O3-filtered and improved response at higher ambient temperatures||Factory characterisation, quality check, accuracy and precision tests||As V4.1, but with optimised electronic setup and performance enhancements below the limit of detection||>0.8||<5ppb|
|V5.0||Jan 2019 – Jul 2019||O3-filtered with new design and improved response to a wider range of conditions||As V4.2||As V4.2, but optimised performance across a wider range of temperatures||>0.8||<5ppb|
|V5.1||Jul 2019 – current||As V5.0||As V5.0||As V5.0, but optimised further below the limit of confidence||>0.85||<5ppb|
R2 of >0.6 for NO2 is generally considered to be a strong enough performance.
*Readings at 15 minute average intervals.
**Typical prescaled variance.
A number of independent comparison studies have been published online (please note any date / AQMesh version against the table above to establish whether the study relates to the latest version).
Latest AQMesh news
AQMesh pods are being co-located with reference stations globally and further performance data will be published in the news section and co-location comparison trials as it becomes available.