1. Knowledge Base
  2. Collocating your devices and calibrating your network

What is calibration and why is calibration important?

Calibration is necessary in order to provide accurate and reliable data, especially during major air quality events.

The calibration process adjusts the air quality data from the Clarity device  based on a relationship determined by data comparisons of the Clarity device with a reference monitor (an air monitor you trust) data from a collocation study. A collocation study is when you place a Clarity device physically next to the trusted reference monitor so both are sampling comparable air pollution. All monitors, including reference monitors, need calibration before and periodically during use to provide accurate and reliable data. 

In some case, relative values (trends in data rather than the actual numbers i.e., data are increasing or decreasing) are useful enough to provide insightful information but in most cases, having reliable absolute concentrations are more useful. Therefore, the indicative monitors must be calibrated against a reliable monitor. The calibration process will reduce biases (over- or under-estimation of concentrations) and improve the performance of the indicative sensor for a specific location by “tuning” it for the local meteorological and PM pollutant profiles.

As these metrological and pollutant profiles change over time (due to seasons, changes in sources), this calibration may need to be updated to continue providing accurate data.

Therefore, the calibration process is an on-going process and important step of the QA/QC of the data. How the calibration is conducted is in part based on the degree of access to reference monitors and their data. This access determines the robustness of the calibration and ability to comment on reliability of the data over long periods of time.

An example

Calibration adjusts readings from indicative sensors to account for changes in environmental conditions and in the case of PM2.5, particle composition.

For the 2020 wildfires in California, we used a 10-fold cross-validated multiple linear regression model using data from collocated Clarity devices. 

As outlined in our blog on wildfire calibrations in California, the following calibration is applied to sensors located in areas with wildfires:  

[calibrated PM2.5 mass concentration] = 0.596 * [raw PM2.5 mass concentration] - 0.045*Relative Humidity + 3.819 ug/m³

Calibration makes it possible for indicative sensors to accurately fill in the geographical gaps between air quality reference stations and create a more widespread network of sensors, with readings adjusted depending on localized air pollution events. 

If you are interested in applying a calibration to your network, please refer to this article: Collocating your Clarity device with a reference monitor.