27/06/21
Air pollution (AQ) has a significant impact on human health and is of global concern. Current AQ monitoring in cities has generally two accepted methods; one of which uses diffusion tubes. These are deployed for a set amount of time (usually a month) and then collected and analysed in a laboratory. These give average concentration over the period of deployment with an agreed accuracy of +/- 25%.
The alternative accepted method uses accurate and expensive instruments, using reference methods specified by government (DEFRA) deployed in a roadside cabinet. The equipment in the cabinet must be routinely serviced and maintained and can give data frequencies in the range of 15 min-hourly time steps. The main issue with these methods are either low temporal resolution with tubes or low spatial coverage with roadside monitors.
IoT based AQ monitoring provides the capability to gather AQ metrics at high temporal and spatial resolution, for a fraction of the cost of roadside monitors and provide the capability to capture dynamic changes in air quality. This project will utilise the urban observatory sensor network and how these sensors compare against traditional methods and build a standard procedure to grow confidence in the data.
This project focused on addressing the barriers associated with technology maturity and trust in data accuracy. The immaturity of the technology has created a mistrust of the data by local authorities due to uncertainty of electro-chemical behaviour over time and therefore sensor maintenance requirements.
The below aims and activities that address those barriers are:
Review and report from scientific and grey literature on electro-chemical sensors.
Experiment to assess stability and/or deterioration of sensors over time and also reproducibility of data from co-located sensors.
Comparison of co-located sensors with currently accepted standard methods.
The intended beneficiaries are the various levels of regional and local government interested in public health and air quality, and citizens who are interested in air quality in their neighbourhoods. Others that potentially benefit from this work are local, regional and national government responsible for clean air zones and evidence on impacts of measures; sensor suppliers, traffic management who are keen to look at how changes in traffic management can lead to reductions in pollution in locations.
Multiple strands of work were undertaken, which created the following individual components for the project:
Review of scientific and grey literature.
Co-location of three air quality sensors inside the cage at Tyne Bridge roadside precision air quality monitoring station. These sensors were modified to contain both brand new, and 12 month old NO2 sensors allowing an analysis of correlation with the reference station over a 10 month period.
Data analysis on the performance of sensors were compared against each other to look at reproducibility and stability/deterioration. This analysis would inform the appropriate maintenance interval required to ensure good data.
Data analysis on the performance of the roadside station was compared to performance of sensors. This analysis would inform the design of an appropriate standard procedure (SOP) to give confidence in the data to the users.
1. The literature review undertaken further emphasised that despite there being a growing body of evidence that the emergent electrochemical sensor based IOT AQ monitoring offering can produce reliable and meaningful insight into the pollution trends in an area (eg Breathe London), wide uptake of the technology is being held up by an absence of established certification and standard operating methodologies, leading to a general lack of trust in the data.
2. Over 90 Envirowatch E-MOTE devices (over 12 months of operation) were recovered from their deployment locations within the Newcastle Urban Observatory network and co-located with a series of brand-new E-MOTE devices for a period of two weeks.
A regression analysis of the data revealed that, with the exception of a very isolated number of sensor failures, all devices showed very good correlation (mostly over 0.85 R2) with the new sensors, demonstrating that the electrochemical sensors are now stable over much longer timescales than in previous iterations and longer deployment timescales are possible without manual maintenance visits.
3. The Tyne Bridge colocation exercise demonstrated that the Envirowatch E-MOTE data, when averaged down to 15 minute values to match the reference station produced a correlation of up to 0.88 R2 over three month analysis periods for NO2, with no manual data correction. The levels of accuracy demonstrated are comfortably within the allowable tolerances for diffusion tubes.
These results further demonstrate that IOT type pervasive sensors using electrochemical sensors produce meaningful, reliable and insightful gas levels and therefore can be used for ambient air quality monitoring, long term evaluation of mitigation measures and potentially can be used for real time, on-the-fly traffic management.
Publication and case study. Data analysis on sensor performance and reproducibility between sensors and comparison with roadside sensor.
Recommendation on sensor maintenance interval to ensure data quality. This is based on comparison of sensor stability between old and new sensors over a 12 month period.
SOP/method to validate sensors in a network to further improve ‘trust’ in data. This includes information to highlight the different operating details of roadside stations and IoT sensors and recommends an appropriate method to allow comparison of data that takes into account their differences.
Knowledge of sensor stability and reproducibility will allow users to have confidence in the data and allow the data to be confidently used in spatial interpolation.
Future impact is reasonably expected to allow LAs to utilise IoT sensing in air quality campaigns or in long term monitoring to gather evidence on performance of clean air zones and other measures at a denser spatial resolution.
If the project influences wider uptake of pervasive IOT type sensors then this will lead to further job security or job creation in the sector.
The SOP, recommendations on maintenance produced in this project are to be utilised in subsequent maintenance regimes in Urban Observatory’s 200+ IoT air quality network.
These outputs will also be shared with other UKCRIC Urban Observatories and our wider community (including local authority public health and air quality teams) and be open to stakeholders on the Urban Observatory data portal as meta data.
As E-MOTE data does not require post processing then real time traffic management could be investigated using IoT sensors to inform on-the-fly traffic light signal regimes based on air quality information.
Investigation of the use of IoT sensors as a checking mechanism for precision systems – that is, testing whether spikes in data are actually representative of a pollution incident rather than instrument error. This would lead to more reliable precision station data and assessment of population exposure to atypical, short term pollution incident.
Despite Covid, regular update meetings have taken place via zoom and progress/outcomes discussed Envirowatch gained insight into the operating procedure for the precision type equipment, allowing us to better understand the data they produce and how this relates to the data produced by the E-MOTE and how our data is useful in the full context of air quality monitoring
Uncertainty over comparing to single precision station – would have been beneficial to have co-located multiple precision stations.
Covid impacted the ability for us to co-locate devices at different locations and collaborate on using different pervasive monitoring systems in different settings.
More input from precision device manufacturers/suppliers/calibrators.
From Envirowatch’ s point of view – a better understanding of the role of IOT type AQ sensors in future monitoring procedures. Gained insight into how the reference stations operate and how data is processed to produce the published results which will influence how we present data to our customers in the future.
As a result of the study we have even more confidence in the lifespan of the electrochemical sensors and sensitivity stability over time.
From Urban Observatory’s point of view – we gained insight into how the reference station operates and issues with some data that we were previously unaware of. Envirowatch analysis and their expertise helped us understand this. The project used additional data sources around the location (CCTV data, weather station data) to understand anomalies in the data from the reference station as events or errors.
We also feel confident in using the data for spatial analysis and enthusiastic about opportunities that can arise in future work in the smart city arena.
Professor Phil James – Newcastle University