Description: | Beginning early in June 1995, the relative humidity probe output began to drift upwards.
A check of the probe on May 19, 1995 by site operations personnel indicated that it was
reporting 3% higher than an aspirated psychron (an ANL instrument that indicates dry and
wet bulb temperatures within 0.1 degree of each other when both are dry). 3% is within the
combined accuracies of the sensors (3% for the RH probe, 2% for the psychron).
Comparison of the RH probe output, SMOS output, and Sonde outputs indicates that the RH
probe was outputting approximately 5% to 10% high by June 30. By August 15 it appeared to
be near 10% to 15% high. On September 19, 1995 Chris Martin (of site operations) and I
compared the RH probe with the HMI31 at 3 meters (tower carriages lowered). The RH probe
indicated 100.9% whereas the HMI31 indicated 83%. Comparison of the HMI31 and the
psychron indicated that the HMI31 reads 2% to 3% lower than the psychron. So the difference
between the RH probe and the psychron would have been approximately 15.5%.
Capacitive RH elements typically drift upwards at the rate of 1% to 2% each year. The
difference between the psychron and the RH probe at the May 19 check was reasonable
considering that it had been in use for more than two years. Previous checks (except for at the
time of installation, when the two agreed within 1%) have indicated a somewhat larger
difference (closer to 5%). The rapid drift of the tower RH probe is likely not a result of
this aging process, but a more rapid deterioration that is attributed to an electronic
malfunction, as has been seen on a couple of occasions in SMOS RH probes.
At 1407 GMT on September 19 the tower carriages were lowered and checks and calibrations
of the RH probe and PRTD were made. The RH probe was checked (as indicated above) and
removed (S/N 231). A replacement probe (S/N 109) was installed and checked. The new probe
indicated the same RH as the HMI31 and just less than 2% less than the psychron did. The
tower carriages were returned to their proper positions at 1715 GMT. Comparison of the
60 meter RH (89%) and the surface RH (81% from the psychron) at that time was reasonable
for the temperature difference of almost 2 degrees C. Tower vapor pressure (the correct
temperature sensor calibration had been entered by that time) was 0.06 kPa less than that
for the SMOS. This was reasonable for a cloudy, windy day. Correct RH data began after
1715 GMT. All 60 meter tower data is incorrect between 1407 GMT and 1715 GMT on September 19.
Furthermore, all vapor pressure data between 1828 and 1845 GMT on September 18 is
incorrect as a consequence of working on the 21X datalogger to find the cause of the 60 meter
temperature problem.
Because the tower temperature has been recording too low and the RH has been indicating
too high, tower vapor pressures have not been greatly different than the actual ambient
vapor pressure. Examination of the SMOS, tower, and Sonde data indicate that the tower
vapor pressure was approximately 0.2 kPa too large in mid-June and 0.7 kPa too large by
September 18.
Correction of the tower RH record for the period of May 19 through September 18 may be
difficult. It may be possible by careful comparison with the Sonde and SMOS records, but
this would be time-consuming. Since the vapor pressure information from the tower is
apparently within about 0.5 kPa of the correct values for most of the period, I indicated that
the data is questionable. Following the information above, one can extrapolate to
obtain an estimate of the 60 meter RH. |