A cross comparison of RH-T sensors was carried out on Tonzi Tower from 2021-01-21 to 2021-03-02.  A new HMP155 and a new METER ATMOS14 were installed on the top of the tower in ventilated solar shields next to the existing HMP45.  Other RH-T sensors in the vicinity include and HMP60 for the Tower CO2 profile and the Floor HMP45.  The ATMOS14 also include pressure so it was compared with the LI7500 and the Floor PTB.  The following table has the sensor details:

The HMP155 is the newest and highest quality sensor and will be taken as the standard for this inter-comparison.  From 2021-01-21 to 2021-02-10 the CR1000 B data logger ran at 10sec intervals and stored 30min averages and it had three par sensors described elsewhere.  From 2021-02-03 the CR1000 B data logger ran at 1sec intervals storing 30min averages and had a sonic described elsewhere.  During the first period the HMP155 was turned on, measured and turned off each measurement cycle.  This seemed to cause bad measurements whether from a programming or wiring bug or some other cause it is hard to say.  This is typically done to save power but is insignificant at our sites so I recommend to leave the sensor powered.  Additionally during the first period the ATMOS14 pressure was logged in low resolution (FP2) which was fixed (IEEE) during the second period.  The data from the first period is retained here to show the noise in the HMP155 and for further inter-comparison of the other sensors if desired. 

Figure 1. Noise in the HMP155 data due to power cycling maybe.  You can see downward excursions in the HMP155 air temperature.  And you will see that these do not exist in Figure 2 below.

There are three goals here: 1) compare the RH and T from the existing Tonzi sensors with the standard HMP155.  2) Evaluate the performance of the METER ATMOS 14 sensor.  3) Look at the atmospheric pressure measurements.

Spoilers: The Tower HMP45 seems to be in fine condition.  The ATMOS14 sensor seems to be just a good as the HMP155 and its pressure is usable but its long term stability is unknown.  The Floor HMP45 RH is cutting off at 93% and its air temperature needs a closer comparison - it probably should be replaced .  The co2prof HMP60 seems okay but could benefit from a close comparison too.  The co2prof PTB air pressure has issues that might be partly caused by a programming bug but needs attention.

Part I RH and T inter-compare.

Figure 2. Time series of air temperature measurements.  Here you can see that the three sensors at the top of the tower follow each other tightly and that the two lower sensors (Floor and co2prof) diverge from the tower top sensors but follow each other well.  For this reason we will not be comparing the lower sensors with the upper sensors but will instead inter-compare the sensors according to their heights.

Figure 3. Scatter plot or air temperatures.  Both the Tower HMP45 and the ATMOS14 agree with the HMP155 to better than 1C.  The Floor HMP45 and co2prof HMP60 have more scatter and further from a 1:1 slope.  This can be partly due to their different height and exposures (the Floor HMP45 is at the edge of the canopy and partly shaded whereas the co2prof is in full sun for most of the day).  However I believe that the Floor HMP45 may also be drifting - see its RH values.

Figure 4. Time series of RH measurements.  Again the upper and lower sensors are clustering together.  The main discrepancy seems to be that the Floor HMP45 is maxing out at about 93%.

Figure 5. Scatter plot of RH measurements.  The ATMOS14 and the HMP155 are very tight.  The Tower HMP45 also has little scatter but there is a little wiggle to the regression that I think makes the slope and offset look worse than they really are.  There is lots of scatter between the co2prof and the Floor HMP45 again some of which I think is due to location difference but the regression is not all that bad.

Part II Evaluation of ATMOS14

As seen above the ATMOS14 compares very well with the HMP155.  Below is the Pressure comparison also looks promising.  The only reluctance I have on relying on the ATMOS14 is whether it will remain stable over many years.  Hard to know.

Part III Air pressure inter-comparison

Figure 6. Time series of Pressure measurements.  The Floor P and LI-7550 P are virtually identical.  The ATMOS14 is slightly higher.  The co2prof is much higher and might be due to a programing error, but there also seems to be some noise the co2prof data.

Figure 7. Scatter plot of pressure measurements.  The Floor PTB and ATMOS14 pressure compare well against the 7500 P, but it is interesting that the ATMOS14 which is at the same height as the 7500 P is more different than the Floor PTB which is 20m lower.  The co2prof PTB is all over the place and far off from the others.  I'll need to look at its code but I think has more problems.