Kestrels are extremely accurate instruments if used correctly. We use them at our CWO for backup if ASOS goes down.
For my station I use an SHT31 in the FARS for temperature and a SHT75 in a passive shield for humidity. This is exactly like ASOS is setup except its humidity sensor can be heated when air nears saturation. Ive had this 31 for nearly a year now without a filter and it still reaches 100% easily in fog. The 75 tops out at 98% so I use a slight slope offset so it reports 100%. The 75 really outperforms the 31 in the mid and lower ranges though as it is usually is between 4-8 percent lower.
My experience with the SHT75 and the Davis SHT31 is exactly the same, except for the easily hitting 100%. The SHT75 is demonstrably more accurate in the mid and low ranges of humidity, even if it has its own wet bias. What I find frustrating and what I have been trying to understand is why this is the case. I have been recently following data from a few new Acurite Atlas weather stations, and despite the issue related to that weather station, I have seen from their forum posts (
http://www.wxforum.net/index.php?topic=35220.0) that it also uses the SHT31, but using the I2C protocol. There is an Atlas in Eastern Washington state (
https://www.wunderground.com/personal-weather-station/dashboard?ID=KWAANATO5&cm_ven=localwx_pwsdash) that has reported sustained 100% humidity at 20F, which is almost unimaginable on a VP2, aside from the occasional 100% spike. I've also seen an Acurite SHT31 perform better at the midranges as I've posted before in this forum.
I know absolutely nothing about I2C and Sbus or the 'analog' version of the SHT31 that I have learned from this forum that Davis uses. What I have noticed looking at the Sensirion datasheets between their I2C and 'analog' versions of the SHT31 is that the analog version requires a conversion of signal output to achieve a temperature and humidity readout and Sensirion publishes on their datasheets the equations to do so. The I2C version does not require such a conversion. Looking back to the SHT11, which Davis originally used, these sensors also required conversions of the signal output and those equations are also published. I presume the VP2 must make these conversions. Is that correct? And if it does, I presume it does these conversions at the ISS, so that a firmware update to the console would not be helpful in updating the new conversion equations for the SHT31. Where I am going with all this and what I'd be interested to hear about from people with bigger brains that understand this stuff is this:
Is the VP2 SHT31 poor humidity performance possibly due to the ISS using a signal output conversion for the old SHT11 and not the appropriate equations for the new SHT31 because there is no way for Davis to retroactively update them in the ISS? Is the use of these old conversions also possibly why they direct putting in a -0.9F temperature correction in an "old" transmitter to approximate the difference in conversion of signal output between the SHT11 and SHT31? Is it also a possible explanation as to why the humidity performance of the SHT75 is superior because it is using a conversion more accurate for the SHT75's signal output?
Just putting it out there because I'd love to get something that at least approximates the stated accuracy specifications of the SHT31. Assuming I'm understanding this correctly (someone who gets this stuff please correct me if I am wrong), I really wish Davis would move to I2C with their Sensirion sensors.