Reports

Contents

Title: Tonzi Tower CO2 Profile
Date:2018-03-11
Data File: TT_co2profile_20180311.csv
TT_co2profile_20180311cal.csv
Refers to:TZ_co2prof

From 2018-03-06 to 2018-03-11 Two CO2 profile systems were run together on the Tonzi Tower.  The old system used a LI-820 CO2 analyzer and solenoid system to pump air from the four inlet locations in series, which results in about 12 samples per half hour.  The new System uses four Vaisala GMP343 analyzers sampling very 30sec and storing 30min averages on a CR6 data logger.  The Vaisala are protected from the sun inside 4in PVC drain pipe with the probes exposed on the north facing side.  Their CO2 measurements are corrected for temperature and pressure base on data from a HMP60 and Vaisala PTB at about 6m above the ground.

Figure 1.  30min averages from the Old LI-820 system.  The data is kind of noisy but notice that all four locations read the same during the day when the atmosphere is well mixed.

Figure 2. 30min averages from the New Vaisala system.  This data looks cleaner but the readings don't match during the well mixed periods.

Also at the site are two LI-7500As one at the top of the Tower and the other at the Floor site.  The readings from these sensors also match during well mixed periods.

Figure 3. Data from the LI-7500As.

Additionally the Old LI-820 system runs calibration gasses twice a day at 0300 and 1500.  For this sampling period there is about a +-30ppm range in the span and a +-15ppm range in the zero, however the overall regression only as about a 5% difference in the span and an offset of 1.2ppm.

Regression Data

Residuals

Figure 4. Calibration runs of the Old LI-820 system with a linear regression: y = 1.053*x + 1.173.

Another way to calibrate the profile systems might be to use the well mixed periods where the Tower 7500A matches the Floor 7500A.  At this time all levels of the profile should be reading nearly the same concentrations.  We could look for times when the Tower 7500 and Floor 7500 are within a ppm or two of each other, but this might happen coincidentally when the air column is not well mixed.  A better method may be to use the Floor wind speed to determine well mixed periods.  If we plot the Floor wind speed versus the difference in the Tower and Floor 7500 CO2 concentrations, we can see that above 0.71m/s the two concentrations agree very closely.

Regression Data

Residuals

Figure 5. Floor wind speed versus the difference of the Floor and Tower CO2 concentrations.

Plotting these points on the time series.

Figure 6.  Time series of Tower and Floor CO2 concentration with periods of well mixed conditions determined by the Floor wind speed.

Now if we take these well mixed periods and use them as the standard to calibrate the Vaisala GMT profile, the following regressions can be determined.

Regression Data

Residuals

Figure 7.  Linear regressions between the four Vaisala GMT343 sample locations and the average of the Tower and Floor 7500 data when Floor wind speed is greater than 0.7

These regressions seem pretty good but should improve using longer sampling periods of several weeks.  Applying these corrections to the Vaisala GMT343 profile data, all four levels collapse to the same concentration during well mixed periods and the ranges more closely match the Old LI-820 ranges.

Figure 8.  Time series of corrected Vaisala GMT343 CO2 profile data.

The GTM343 CO2 probes also report temperatures.  It might be possible to relate these to the Tower and Floor HMP45 air temperatures.

Figure 9.  Time series of the Vaisala GMT343 probe temperatures.