|
|
|
Here are a few examples showing the raw unscaled data (15-minutes) and then the final
processed concentrations (hourly). |
|
Does your final data look this good? I can help you clean up your data. |
|
Calculating the
concentrations from the raw data and eliminating bad data is the very essence
of air quality data processing. Making the final dataset look good can become
difficult if you do not understand how instruments and calibrations can go
wrong. Applying the correct sensitivity, ramping between calibrations,
correcting for NOx converter inefficiencies and using stable certified gas
standards is another topic all together. |
|
NO Large NO baseline offset, a calibration peak in
June and a step change in August.
|
|
NO, NOx and NO2 NO and NOx instrument channels show regular
auto-calibrations peaks and then baseline instability. The NO2 instrument channel should always be
ignored!
The after plots show the cleaned up NO and NO2 ppb
concentrations.
|
|
Ozone Regular ozone auto-calibration spans showing in the
data.
Internal sampling between LSO visits
|
|
SO2 Large baseline offset, zero spikes and roller
coaster baseline.
|
|
PM10 TEOM instruments can show random down spikes and
long periods of symmetrical noise.
|
|
Contact me
Geoff.Broughton@aqdm.co.uk |
