Reports

Contents

Title: PAR comparison 09-2018
Date:2018-09-28
Data File: PARcompare_20180928.csv
PARcompare_20180928_10min.csv
PARcompare_20180928_1min.csv
Refers to:161073,140455,040412,010159

We set up 3 PAR sensors from the lab (test sensors) next to each other for a comparison with an Ameriflux-calibrated (reference) PAR sensor from 2018-09-28 to 2018-10-03. The 4 sensors were set up about 1m outside of a south-facing lab window on the 2nd floor. There are some instances of hysteresis, but in general the 4 sensors are pretty well matched. They were connected to a CR6, which recorded data every 10 seconds. The new calibration coefficients are below.

The original experiment ran for 5 days. Using only the first 2 days of data yielded calibration coefficients that were less than 0.2% (0.002) different.

Sensor model Serial number Factory calibration [µV/(µmol/m2·s)] New calibration [µV/(µmol/m2·s)] Recommendation
Kipp & Zonen PQS1 sn 161073 4.78 n/a (Calibrated by Ameriflux 08/2017) n/a
Kipp & Zonen PQS1 sn 140455 8.15 8.47 Apply new correction
Kipp & Zonen PAR-LITE sn 040412 5.58 5.74 Apply new correction
Kipp & Zonen PAR-LITE sn 010159 5.16 5.59 Apply new correction

 

Figure 1. Time series of PAR data.

Figure 4. Example of hysteresis between the 4 PAR sensors.

Regression Data

Residuals

Figure 2. Scatter plot of test sensors plotted against the Ameriflux sensor. All R2 values are >0.99. Only values >0.2 umol/m2/s are included to prevent to much weight from night-time data.

Regression Data

Residuals

Figure 3. Scatter plot of test sensor mV readings against the Ameriflux sensor umol/m2/s readings. Again, only values >0.2 umol/m2*s are included to prevent to much weight from night-time data.  I used the slope of the linear regression lines to calculate the new calibration coefficients.

 

Averaging the data from 10s to 1-min or 10-min averages did not alter the calibration factors very much, so in the future 1-min or 10-min data could be used to calculate calibration factors if 10s data was not available.

  New calibration factors [µV/(µmol/m2·s)]
Serial number 10s data 1 min data

10 min data
sn 140455 8.47

8.46

 8.46
sn 040412 5.74 5.76 5.76
sn 010159 5.59 5.60 5.60

Figure 6. Time series of PAR data averaged over 1 minute. Some hysteresis remains.

Regression Data

Residuals

Figure 7. Scatter plot of 1-minute data plotted against the Ameriflux sensor umol/m2/s readings.

Figure 5. Time series of PAR data averaged over 10 minutes. Hysteresis is gone.

Figure 8. Scatter plot of 10-minute data plotted against the Ameriflux sensor umol/m2/s readings.