Reports

Contents

Title: SW AirT/RH comparison 2020
Date:2020-09-10 - 2020-11-12
Data File: GoldenSet_SW_20200910.csv
Refers to:sn W0730036, sn R0360206

In fall 2020, we set up some comparison sensors on the SW tower for about 2 months.

  • HMP155 sn R0360206 (reference sensor, brand new)
  • 2 multi-globe thermometers (link to paper) -These sensors are from Joe's friend Atsushi Maruyama in Japan. The globe thermometers are three thermocouples in different size stainless steel spheres (~1mm – 7mm). Combining the three measurements should give actual air temperature without the need of a solar shield or ventilation.

All these sensors were connected to a CR1000 sn 4190. Data was measured every 10 seconds and 30-minute averages were recorded.

We compared these data to the tower AirT/RH data. The tower sensors were connected to the tower met datalogger. Data was measured every 10 seconds and 30-minute averages were recorded.

https://nature.berkeley.edu/biometlab/bmetdata/_photodb/2020/20200910SW_GlobeThermometers.jpg

On the left in the green lab clamps are 2 globe thermometers. On right right is a new HMP155 sensor to act as our reference sensor.

 

Sensor model Serial Number

Air T

Field calibration

Air T

Factory spec

RH

Field calibration

RH

Factory spec

 Recommendation

HMP 155

  sn R0360206 (reference sensor) +/- 0.2°C for 0-40°C (reference sensor) +/-1% for 0-90% RH

n/a
HMP45 sn W0730036 y=0.99*x-0.23, R2=0.99 +/- 0.3°C for 0-40°C y=0.97*x+3.3,R2=0.99 +/-2% for 0-90% RH Use field calibration for RH. No calibration needed (w/in spec) for Air T.

Figure 1. Time-series of air temperature. Fit looks good except HMP155 has slightly lower daily minimums.

Regression Data

Residuals

Figure 2. Linear regression of air temperature. Tight fit without much noise.

Figure 3. Time-series of RH. Fit looks good except HMP155 has slightly higher daily  minimums.

Regression Data

Residuals

Figure 4. Linear regression of RH. There is a slightly different slope when RH<20%, but that doesn't happen frequently so I feel ok ignoring it.

Figure 5. Time-series of calibrated RH. Seems like a reasonable calibration.