calibration factors derived in the lab will not apply accurately when the
sensor is deployed to the field under different conditions.
Calibration can often be done using data from a regulatory
monitoring station located near the sensor’s deployment site. Using your
local environmental protection agency’s website (e.g., airnow.gov in the
US), locate an air quality monitoring station near you, and deploy your
sensors nearby (<1 km in all cases, but < 50 m is best). Depending on the
agency, it may be possible to contact your local branch or representative
and inquire about collocation opportunities. Whenever possible,
calibration should be carried out against Federal Reference Methods
(FRM) or Federal Equivalent Methods (FEM) that apply in the deployment
jurisdiction. This ensures that the ObservAir’s gas data can be
meaningfully compared to regulatory data collected elsewhere.
Collocation should be long enough to ensure that a representative
sample of gas concentration measurements is collected – typically at least
1 week is required. Compare the data collected by the ObservAir to that
collected simultaneously by the reference instrument and derive a linear
calibration model, as shown below. Usually, hourly-average concentration
data is used to derive the linear regression factors.
C
ref
(t) = Reference gas concentration measurement at time ‘t’ (ppm)
C
OA
(t)
= ObservAir gas concentration measurement at time ‘t’ (ppm)
m = Linear regression slope
b = Linear regression intercept (ppm)
Using these factors, the ObservAir data can then be corrected using the
equation below. Span and zero factors are uploaded to the instrument over
the Cloud Dashboard or serial connection.
C
cal
(t) = Calibrated gas concentration measurement at time ‘t’ (ppm)
C
raw
(t)
= Raw gas concentration measurement at time ‘t’ (ppm)
span = m = Derived linear slope
zero = b = Derived linear intercept (ppm)
Note: The same collocated data collected for gas calibration can be used
for environmental compensation. If the sensors are connected to the Cloud
Dashboard, simply send the reference data to DSTech – our staff will
derive compensation factors for each sensor (using machine learning) and
upload them remotely. If the sensors are not connected to the Cloud
To Next Page
Loading...