Element 22.9
Choosing and calibrating sensors
- Sheet
- 22.9
- State
- Drafted
- Release
- 16 November 2026
- Editor
- Contributed by Praecise as open research
- Reviewers
- To be named at review
- Safety
- S1, as the chapter
- Licence
- CC BY-SA 4.0
- Resources
- 10
The element
- Requirement
Every sensor the community relies on is chosen against a stated question, has published evidence of how it performs, and is checked against a reference before its readings are used to decide anything. Its readings carry a statement of their uncertainty and of the conditions in which they hold.
- How to tell
- The question is written down first: what is measured, where, how often, and what decision the reading will inform. The Enhanced Air Sensor Guidebook sets out this workflow, from question to evaluation (The Enhanced Air Sensor Guidebook).
- An independent evaluation of the model exists and has been read. AQ-SPEC, run by the South Coast Air Quality Management District since 2014, tests commercial low-cost air sensors in the field beside reference monitors and in the laboratory (Air Quality Sensor Performance Evaluation Center (AQ-SPEC)). The US EPA publishes voluntary testing protocols and target values for sensors measuring ozone and PM2.5 (2021) and PM10, NO2, CO and SO2 (2024), with field collocation as base testing and laboratory work as enhanced testing (Air Sensor Performance Targets and Testing Protocols).
- At least one unit has been run beside a reference instrument in local conditions and a correction derived. A US-wide study found raw readings from one widely used low-cost particle sensor overestimated PM2.5 by about 40 percent; a correction with a humidity term cut the error from about 8 to 3 micrograms per cubic metre (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor).
- The maker's specification is recorded with its conditions. For example, a photoacoustic CO2 sensor is specified over 400 to 5,000 ppm and -10 to 60 degrees C (Sensirion SCD41 CO2 sensor); a motion sensor over -40 to 85 degrees C (Bosch Sensortec BMI270 inertial measurement unit).
- Uncertainty is stated using the international method (JCGM publications: Guide to the expression of uncertainty in measurement (GUM) and International Vocabulary of Metrology (VIM)).
- Options
- Reference or regulatory-grade instruments: highest confidence and cost; fit a single fixed site that anchors others.
- Low-cost sensors corrected against a reference: fit dense networks. The WMO states that the trade-offs that make them cheap usually limit data quality, selectivity, sensitivity at low levels, robustness at high levels and lifetime, and recommends combining them with reference monitors, models and satellites (Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293)).
- Open community kits with shared data, such as GPS-logging radiation monitors that record every 5 seconds and upload to an open map (Safecast bGeigie Nano).
- Reading existing open data before buying anything (OpenAQ).
- Settled, open, contested
Settled: low-cost sensor readings need correction against a reference, and humidity is a main source of error for optical particle sensors (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor, Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293)). Open: how corrections derived in one climate transfer to another. The US study notes few rural sites and possible errors in Alaska's winter and in dust events (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor). Contested: whether corrected low-cost data should sit beside regulatory data. EPA frames its targets as voluntary and for non-regulatory monitoring (Air Sensor Performance Targets and Testing Protocols); the WMO sees them as filling gaps where reference monitors are missing (Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293)).
The four stages
- Learn: The Enhanced Air Sensor Guidebook; Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293); the measurement vocabulary and uncertainty guide in JCGM publications: Guide to the expression of uncertainty in measurement (GUM) and International Vocabulary of Metrology (VIM).
- Design: choose models from published evaluations (Air Quality Sensor Performance Evaluation Center (AQ-SPEC), Air Sensor Performance Targets and Testing Protocols); record the maker's ranges (Sensirion SCD41 CO2 sensor, Bosch Sensortec BMI270 inertial measurement unit).
- Build: collocate units with a reference and derive a correction, as in Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor.
- Deploy: publish corrected data with its method; compare with open data (OpenAQ); recheck collocation at intervals.
Resources at each stage
1 Learn
Understand
- Air Quality Sensor Performance Evaluation Center (AQ-SPEC) Case study, S0
- JCGM publications: Guide to the expression of uncertainty in measurement (GUM) and International Vocabulary of Metrology (VIM) Standard, S0
- The Enhanced Air Sensor Guidebook Guide, S0
- Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293) Guide, S0
2 Design
Size and choose
- Air Quality Sensor Performance Evaluation Center (AQ-SPEC) Case study, S0
- Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor Paper, S0
- JCGM publications: Guide to the expression of uncertainty in measurement (GUM) and International Vocabulary of Metrology (VIM) Standard, S0
- Bosch Sensortec BMI270 inertial measurement unit Reference design, S0
- Air Sensor Performance Targets and Testing Protocols Standard, S0
- The Enhanced Air Sensor Guidebook Guide, S0
- OpenAQ Dataset, S0
- Sensirion SCD41 CO2 sensor Reference design, S0
- Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293) Guide, S0
3 Build
Build or obtain
- Bosch Sensortec BMI270 inertial measurement unit Reference design, S0
- Safecast bGeigie Nano Reference design, S0
- Sensirion SCD41 CO2 sensor Reference design, S0
4 Deploy
Install, run and measure
- Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor Paper, S0
- Air Sensor Performance Targets and Testing Protocols Standard, S0
- The Enhanced Air Sensor Guidebook Guide, S0
- OpenAQ Dataset, S0
- Safecast bGeigie Nano Reference design, S0
What changes between 30 adults and 300 people
At 30 adults, a handful of sensors, one collocation period beside the nearest public monitor and a written correction are enough. At 300 people, a named keeper holds the reference or transfer instrument, rotates sensors through collocation, and keeps a log of drift and replacement; the data are good enough to share with an authority only when the method is published.
Reference designs
-
Reference design
Bosch Sensortec BMI270 inertial measurement unit
Maker's page for a low-power six-axis IMU: accelerometer ranges of 2 to 16 g, gyroscope ranges of 125 to 2,000 degrees per second, 685 microamps at full data rate, -40 to 85 degrees C, SPI and I2C, with on-chip motion, step and gesture recognition.
External resource. Not made or verified by Public ProtocolMaturity: EstablishedSafety S0CommercialListed 16 November 2026 -
Reference design
Mobile, GPS-logging radiation sensor kit from the volunteer Safecast project, built to ride in a car window or on a bicycle and record a reading every 5 seconds with time and place. Logs are uploaded to the Safecast API and map; board files, cut templates and firmware are on GitHub, and a successor kit (bGeigieZen) is offered.
External resource. Not made or verified by Public ProtocolMaturity: ReplicatedSafety S0Open, with paid products or servicesListed 16 November 2026 -
Reference design
Maker's product page for a miniature photoacoustic NDIR carbon dioxide sensor with on-board temperature and humidity compensation: specified range 400 to 5,000 ppm, accuracy given as plus or minus 50 ppm plus a percentage of the reading, 2.4 to 5.5 V supply, 15 mA average current and -10 to 60 degrees C operation.
External resource. Not made or verified by Public ProtocolMaturity: EstablishedSafety S0CommercialListed 16 November 2026
Parts list
Every resource that supports this element. Those marked External carry the label “External resource. Not made or verified by Public Protocol”.
| Resource | Type | Origin | Maturity | Safety | Licence, as stated at its source |
|---|---|---|---|---|---|
| Air Quality Sensor Performance Evaluation Center (AQ-SPEC) | Case study | External | Established | S0 | Not stated on the page opened |
| Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor | Paper | External | Replicated | S0 | CC BY 4.0 |
| JCGM publications: Guide to the expression of uncertainty in measurement (GUM) and International Vocabulary of Metrology (VIM) | Standard | External | Established | S0 | Copyright BIPM, all rights reserved (page footer); no open licence stated |
| Bosch Sensortec BMI270 inertial measurement unit | Reference design | External | Established | S0 | Not stated on the page opened |
| Air Sensor Performance Targets and Testing Protocols | Standard | External | Established | S0 | Not stated on the page opened |
| The Enhanced Air Sensor Guidebook | Guide | External | Established | S0 | Not stated on the page opened |
| OpenAQ | Dataset | External | Established | S0 | Not stated on the page opened |
| Safecast bGeigie Nano | Reference design | External | Replicated | S0 | Website CC BY-NC 4.0; licence of the hardware files not stated on the page opened |
| Sensirion SCD41 CO2 sensor | Reference design | External | Established | S0 | Not stated on the page opened |
| Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293) | Guide | External | Established | S0 | Not stated on the page opened |
Depends on, and completes
It is one of 12 elements that complete Delivery, operations and measurement.
The standing questions
Who pays for the reference instrument or the time beside one. Who keeps the calibration record. Who decides when a reading is good enough to act on.
Environments
no source found on salt exposure for the sensors listed.
the US particle correction may underestimate in Alaska's winter (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor); check each sensor's rated temperature, for example -10 degrees C lower limit for one CO2 sensor (Sensirion SCD41 CO2 sensor).
dust events are named as a condition where the particle correction may fail (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor).
humidity is a main error term for optical particle sensors and must be in the correction (Development and application of a United States-wide correction for PM2.5 data collected with the PurpleAir sensor).
Temperate forest, high altitude, flood plains
no source found specific to these settings.
Places without reference monitors
the WMO names low- and middle-income countries as where sensor networks fill the largest gaps (Integrating Low-Cost Sensor Systems and Networks to Enhance Air Quality Applications (GAW Report No. 293)).
Environment profiles for these settings: Small islands and coasts; Arctic, subarctic and permafrost, including sea ice; Extreme cold; Hot deserts; Extreme heat; Tropical rainforest and humid tropics; High humidity; Temperate and boreal forests; Mountains and high altitude; River deltas and flood plains. All environments
Safety
State and evidence
Drafted
An editor has written the element's text from the resources listed under it, citing them. It is checked when it is reviewed.
Evidence from a real place or system is recorded here, with what did not work. What the states mean
History
- Drafted in the release of 16 November 2026.