Reports

Contents

Title: Eosense CO2 Probe Compare
Date:2018-04-23
Data File: EosGP_Compare_20180423_A.csv
EosGP_Compare_20180423_B.csv
Refers to:GP20150101,GP20160096,GP20140084,GP20160127,GP20140088,GP20150109,GP20150111,GP20160127

On April 23 and 30, 2018 an inter-comparison of Eosense CO2 probes left in the lab was performed.  A follow up inter-comparison was done on 2018-09-01 and can be found here: Eosense CO2 Probe Compare.  The probes were compared in two batches against a new ("master") probe.  The probes were put into a plastic box and the CO2 level was raised to 2%-3% and allowed to dissipate back to near ambient levels.  The data was recorded at 10sec intervals by a CR1000.  The analog range of all the probes is their range plus 20%

CO2 (ppm) = mV / 5000 * (1.2 * range)

Temp (C) = mV / 5000 * 70 - 20

BatchSerial No.RangeNotes
A GP20150101 2% Okay when removed from SW for higher range
GP20160096 1% From LBNL - New
GP20140084 2% Removed from TA soil profile for funky readings
GP20160127 1% Master Rep - From LBNL - New
B GP20140088 8% No history - Looks relatively clean but has been deployed
GP20150109 8% Removed from WP for funky readings
GP20150111 2% Okay when removed from MB
GP20160127 1% Master Rep - From LBNL - New

Figure 1. Time series for Batch A.  All probes were maxed out with only sn GP20150101 not following the others.

Regression Data

Residuals

Figure 2. Regressions for the Batch A probes against the Master sn GP20160127.  The relationships seem robust but sn GP20150101 seems to have a slope offset of about 18%.

Figure 3. Time series for Batch B.  Two of the probes have a much higher range (8%) and one, sn GP20140088 is very noisy and offset.

Regression Data

Residuals

Figure 4. Regressions for Batch B probes against the Master sn GP20160127.  As expected sn GP20140088 looks bad but the other two seem okay.

Both of these runs were done using my breath to raise the CO2 concentration in the plastic box.  I tried to use a 1% CO2 cal gas but could only get the concentration in the box up to about 0.675% - maybe need a higher flow rate or seal the box better.

 Conclusions:

Serial No.RangeSlopeOffsetr^2CountNotes
GP20150101 2% 1.1791 259.77 0.9991 195 Okay but needs this correction applied
GP20160096 1% 1.0042 -226.56 0.9989 195 Good to go
GP20140084 2% 1.044 -302.71 0.9995 195 Okay but consider applying this correction
GP20140088 8% 0.69373 -1630.9 0.8448 415 Don't use - noisy and offset
GP20150109 8% 1.0993 152.4 0.997 415 Okay but consider applying this correction
GP20150111 2% 1.0661 77.622 0.9975 415 Okay but consider applying this correction