Machine learning techniques to improve the field performance of low-cost air quality sensors
<p>Low-cost air quality sensors offer significant potential for enhancing urban air quality networks by providing higher-spatiotemporal-resolution data needed, for example, for evaluation of air quality interventions. However, these sensors present methodological and deployment challenges whic...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2022-06-01
|
| Series: | Atmospheric Measurement Techniques |
| Online Access: | https://amt.copernicus.org/articles/15/3261/2022/amt-15-3261-2022.pdf |
| _version_ | 1817988850536939520 |
|---|---|
| author | T. Bush T. Bush N. Papaioannou F. Leach F. D. Pope A. Singh G. N. Thomas B. Stacey S. Bartington |
| author_facet | T. Bush T. Bush N. Papaioannou F. Leach F. D. Pope A. Singh G. N. Thomas B. Stacey S. Bartington |
| author_sort | T. Bush |
| collection | DOAJ |
| description | <p>Low-cost air quality sensors offer significant potential for
enhancing urban air quality networks by providing higher-spatiotemporal-resolution data needed, for example, for evaluation of air quality
interventions. However, these sensors present methodological and deployment
challenges which have historically limited operational ability. These
include variability in performance characteristics and sensitivity to
environmental conditions. In this work, we investigate field “baselining”
and interference correction using random forest regression methods for
low-cost sensing of NO<span class="inline-formula"><sub>2</sub></span>, PM<span class="inline-formula"><sub>10</sub></span> (particulate matter) and PM<span class="inline-formula"><sub>2.5</sub></span>. Model performance
is explored using data obtained over a 7-month period by real-world field
sensor deployment alongside reference method instrumentation. Workflows and
processes developed are shown to be effective in normalising variable sensor
baseline offsets and reducing uncertainty in sensor response arising from
environmental interferences. We demonstrate improvements of between 37 %
and 94 % in the mean absolute error term of fully corrected sensor
datasets; this is equivalent to performance within <span class="inline-formula">±2.6</span> ppb of the reference
method for NO<span class="inline-formula"><sub>2</sub></span>, <span class="inline-formula">±4.4</span> <span class="inline-formula">µ</span>g m<span class="inline-formula"><sup>−3</sup></span> for PM<span class="inline-formula"><sub>10</sub></span> and <span class="inline-formula">±2.7</span> <span class="inline-formula">µ</span>g m<span class="inline-formula"><sup>−3</sup></span> for PM<span class="inline-formula"><sub>2.5</sub></span>. Expanded-uncertainty estimates for PM<span class="inline-formula"><sub>10</sub></span>
and PM<span class="inline-formula"><sub>2.5</sub></span> correction models are shown to meet performance criteria
recommended by European air quality legislation, whilst that of the NO<span class="inline-formula"><sub>2</sub></span>
correction model was found to be narrowly (<span class="inline-formula">∼5</span> %) outside of
its acceptance envelope. Expanded-uncertainty estimates for corrected sensor
datasets not used in model training were 29 %, 21 % and 27 % for
NO<span class="inline-formula"><sub>2</sub></span>, PM<span class="inline-formula"><sub>10</sub></span> and PM<span class="inline-formula"><sub>2.5</sub></span> respectively.</p> |
| first_indexed | 2024-04-14T00:39:33Z |
| format | Article |
| id | doaj.art-67116e12fa2f428a80fac7f111f0abd8 |
| institution | Directory Open Access Journal |
| issn | 1867-1381 1867-8548 |
| language | English |
| last_indexed | 2024-04-14T00:39:33Z |
| publishDate | 2022-06-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | Atmospheric Measurement Techniques |
| spelling | doaj.art-67116e12fa2f428a80fac7f111f0abd82022-12-22T02:22:14ZengCopernicus PublicationsAtmospheric Measurement Techniques1867-13811867-85482022-06-01153261327810.5194/amt-15-3261-2022Machine learning techniques to improve the field performance of low-cost air quality sensorsT. Bush0T. Bush1N. Papaioannou2F. Leach3F. D. Pope4A. Singh5G. N. Thomas6B. Stacey7S. Bartington8Department of Engineering Science, University of Oxford, Parks Road, Oxford, OX1 3PJ, UKApertum Consulting, Harwell, Oxfordshire, UKDepartment of Engineering Science, University of Oxford, Parks Road, Oxford, OX1 3PJ, UKDepartment of Engineering Science, University of Oxford, Parks Road, Oxford, OX1 3PJ, UKSchool of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UKSchool of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UKInstitute of Applied Health Research, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UKRicardo Energy & Environment, The Gemini Building, Fermi Avenue, Harwell, Didcot, OX11 0QR, UKInstitute of Applied Health Research, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK<p>Low-cost air quality sensors offer significant potential for enhancing urban air quality networks by providing higher-spatiotemporal-resolution data needed, for example, for evaluation of air quality interventions. However, these sensors present methodological and deployment challenges which have historically limited operational ability. These include variability in performance characteristics and sensitivity to environmental conditions. In this work, we investigate field “baselining” and interference correction using random forest regression methods for low-cost sensing of NO<span class="inline-formula"><sub>2</sub></span>, PM<span class="inline-formula"><sub>10</sub></span> (particulate matter) and PM<span class="inline-formula"><sub>2.5</sub></span>. Model performance is explored using data obtained over a 7-month period by real-world field sensor deployment alongside reference method instrumentation. Workflows and processes developed are shown to be effective in normalising variable sensor baseline offsets and reducing uncertainty in sensor response arising from environmental interferences. We demonstrate improvements of between 37 % and 94 % in the mean absolute error term of fully corrected sensor datasets; this is equivalent to performance within <span class="inline-formula">±2.6</span> ppb of the reference method for NO<span class="inline-formula"><sub>2</sub></span>, <span class="inline-formula">±4.4</span> <span class="inline-formula">µ</span>g m<span class="inline-formula"><sup>−3</sup></span> for PM<span class="inline-formula"><sub>10</sub></span> and <span class="inline-formula">±2.7</span> <span class="inline-formula">µ</span>g m<span class="inline-formula"><sup>−3</sup></span> for PM<span class="inline-formula"><sub>2.5</sub></span>. Expanded-uncertainty estimates for PM<span class="inline-formula"><sub>10</sub></span> and PM<span class="inline-formula"><sub>2.5</sub></span> correction models are shown to meet performance criteria recommended by European air quality legislation, whilst that of the NO<span class="inline-formula"><sub>2</sub></span> correction model was found to be narrowly (<span class="inline-formula">∼5</span> %) outside of its acceptance envelope. Expanded-uncertainty estimates for corrected sensor datasets not used in model training were 29 %, 21 % and 27 % for NO<span class="inline-formula"><sub>2</sub></span>, PM<span class="inline-formula"><sub>10</sub></span> and PM<span class="inline-formula"><sub>2.5</sub></span> respectively.</p>https://amt.copernicus.org/articles/15/3261/2022/amt-15-3261-2022.pdf |
| spellingShingle | T. Bush T. Bush N. Papaioannou F. Leach F. D. Pope A. Singh G. N. Thomas B. Stacey S. Bartington Machine learning techniques to improve the field performance of low-cost air quality sensors Atmospheric Measurement Techniques |
| title | Machine learning techniques to improve the field performance of low-cost air quality sensors |
| title_full | Machine learning techniques to improve the field performance of low-cost air quality sensors |
| title_fullStr | Machine learning techniques to improve the field performance of low-cost air quality sensors |
| title_full_unstemmed | Machine learning techniques to improve the field performance of low-cost air quality sensors |
| title_short | Machine learning techniques to improve the field performance of low-cost air quality sensors |
| title_sort | machine learning techniques to improve the field performance of low cost air quality sensors |
| url | https://amt.copernicus.org/articles/15/3261/2022/amt-15-3261-2022.pdf |
| work_keys_str_mv | AT tbush machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT tbush machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT npapaioannou machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT fleach machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT fdpope machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT asingh machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT gnthomas machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT bstacey machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors AT sbartington machinelearningtechniquestoimprovethefieldperformanceoflowcostairqualitysensors |