Doug's Surface Weather Observations
I have Paroscientific's MET3A mounted about 9.1m under DWH1's ARP. It is on the same mast, and the mounting mechanism places it a few inches north. The input for the temperature sensor is about 1.25m above the ground, and the location is on a north wall of my house (NE of Seattle WA).
I record the pressure, temperature, and relative humidity readings every minute and currently integrate pressure over about 13 seconds (which results in .1ppm). Pressure accuracy and resolution is .1 hPa (hPa=mbar), temperature accuracy and resolution is .1°C, while relative humidity accuracy is 2%. The humidity sensor seems somewhat noisy at the one minute rate, so users presumably should smooth using nearby minutes. I've also added a simple rain gauge, and sometimes generate daily summaries for viewing graphically here. It seems that typically the daily high temperature occurs ~3:10pm (low relative humidity follows ~3:27pm), and low temperature occurs around 4:09am to 4:40am
After the end of each UTC day (around 4:30pm Pacific time), I post the daily
log file at ftp://dsl.niceties.com/pub
in directories/files of style YYDOY/dwh1DOY0.YYm, where DOY is the
numbered day of year (starting at 001) and YY is the two-digit year (such as
01); someday I might use a file compression tool which would result in a suffix
The log is in the RINEX MET standard format used by the GPS-style geodetic industry.
The important things to know when reading the file are that the "columns" of the data lines are as follows (timestamp is GPS time which is a small number of seconds advanced to UTC):
Year Month Day Hour Minute Second Pressure (hPa) Temperature (C) Relative Humidity (%)
and the sensor position gives the height of pressure sensor in meters above WGS84 ellipsoid. Since I'm about 123.5 m above MSL, I could multiply my station pressure value by about 1.01496 to estimate Sea-Level Pressure, but that could sometimes be off by around .5 hPa so I estimate using methods similar to the following. If you you want high accuracy, you try using calculators to estimate Virtual Temperature (perhaps you'll come up with ~11.0C for my average temperature of ~9.5C and average RH~82%, remembering that the mean temperature between me and sea-level is probably about .4°C higher than mine assuming standard 6.45K/Km lapse rate), and then input that result along with my estimated Gravity (9.8073 or 9.8074 m/s) into a calculator for SLP.
If you have a MET3A (or MET3) connected to a Win32-based PC (Win9x/Me or WinNT/2000/XP) and want to produce similar output, feel free to ask for my source code or executable.
Sometimes I try comparison measurements:
Met3A 994.3hPa 9.1C 98%RH; wet/dry 9.2C 9.2C (~100%RH), 748.1mm-2.13mm+.15mm=994.7(+/-.3)hPa
Met3A 989.9hPa 12.1C 87%RH; wet/dry 11.1C 12.8C (~83%RH), 744.7mm-2.12mm+.15mm=990.2(+/-.3)hPa
My temperature page describes methods I use to achieve accurate temperature measurement.
In mid-October'01 I carefully verified that my Met3A was reading about .794C high (at ~10C). I won't adjust archived data, but at ~2130Z 18 Oct I changed Z1 from 0 to -.794 and will try to study again later at some lower/higher temperatures. Sorry about such ~1.43F error; Note that the RH was presumably low due to the higher temperature.
Z1 was set to -1.1 sometime around summer'02. November 10'02 I changed Z2 from 0 to 7.5 to raise RH readings by 7.5% since they haven't hit 100% for years (a few checks from 80 to 95% showed the Met3A had read ~9% low). Later the same day I added another percent (Z2=8.5) which probably brings it within a couple percent (verified from around 75-95%). I also revised my weather program to ensure rounding RINEX pressure/temperature, round sea-level pressure and degrees F, and display another digit of resolution for degrees C if available.
Around August'06 around 90F I logged that Met3A was reading only a few tenths of degree higher than Hart thermistor, and in late November'06 around 18F I logged that Met3A was reading around 1.5 degree lower than Hart thermistor (I did not make adjustment since those are extremes and still reasonable).
The following chart shows various items of interest. The right axis is Relative Humidity with the dark blue line being the noisy Met3A and the purple line near the end being the Vaisala; I believe the actual RH was somewhere between those two. The left axis is temperature, where the Hart (yellow line) was in free air outside the Met3A's temperature block for the first 14 hours (the Met3A block seems to read ~.8C warmer). The expected thermal lag can also be seen. Then the Hart was inserted into the block to show agreement at the .1C level. Near the end, the Met3A is also warmer than the Vaisala (light blue line). The increased temperature and RH event near the fifth hour is due to me running a cloths dryer which occurs 1-2 times per week for about an hour - the vent outlet is a few meters away. The Hart in free air quickly warms up, then cools for the short time I interrupted the dryer to remove some items, then quickly warms back up again. The Met3A's thermal block seems to cap the overall effect to about a quarter-degree C, while the RH effect might be a couple percent.
Up to Doug Hogarth's GPS