Tomasz Lewicki <[email protected]> writes:
> thank you both for thoughts and hints. Indeed, when I wrote "When
> calculated by Weewx itself ("software" option enabled in weewx.conf), my
> weather station shows 1001 hPa - still too low", I meant what Weewx
> calculates and show on my webpage, not on console's display - display is
> completely wrong in my opinion.
Thanks for clarifying. This is tricky so we need to be 2x clearer than
we think we need to be :-)
> According to design of PWS, I own Ambient Weather WS-1001-WiFi clone. I use
> HP-1000 driver. I don't know if it is designed wrong or not, but I know
> that many people still use it.
True that people use it but that doesn't mean it really works, and the
path to getting useful data out is to understand how it is wrong so that
can be worked around.
> Manual for my station say "To determine the relative pressure for your
> location, locate an official reporting station near you (the internet is
> the best source for real time barometer conditions, such as Weather.com or
> Wunderground.com), and set your weather station to match the official
> reporting station." I understand it, I should calibrate PWS entering proper
> value from METAR for example.
I assume -- but you didn't say -- that "set your weather station to
match" means "do some UI adjustment with up/down so that the thing the
weather station displays as 'relative' is the same value as the
barometric pressure from a nearby official station". That's what I
would expect.
What is going on then is:
there is an absolute pressure measurement device in the console, which
is
a) probably not super well calibrated (might be within 2-3 hPa,
might be a bit better - we simply don't know so far)
b) at some unknown altitude (even if you know it, the console
doesn't), so the console has no way to do the proper reduction to
barometric pressure.
console takes that value and adds a constant offset. You can change
that in the UI process above. This is a hack to get something that is
more or less good enough for stations in this price/performance class,
approximating the barometric pressure formula by "add K".
What I still don't know is what value is sent to the computer. The
uncalibrated station pressure? The 'relative' pressure, aka
"station+K, for the entered K"? Both? I think you don't know this
either, and we can't figure out what's going on and what to do until
we understand it.
> But I still don't understand why Weewx, using "software" option for
> calculating the sea-level (relative) pressure, shows too low value. As I
Of course we don't understand, because we don't know what is being
reported and how the console adjustment procedure affected it. We also
don't know what kind of errors the sensor has.
Sea-level pressure is not "relative" pressure. Sea-level pressure
normally means "barometric pressure", which is station pressure reduced
with temperature measurments. (There is also "altimeter pressure",
which is station pressure reduced with a standard atmosphere profile,
but that's close to barometric and you are a long way from worrying
about that difference.)
There is no such thing as "relative pressure", only "fine offset
mislabeled as relative but is actually station pressure + K". As I said
before, you should understand the proper terms and use them carefully,
if you want to get correct answers.
> said, my station location AMSL in weewx.conf is "altitude = 550, meter"
> (and for me it is neglible if it is 552 or 560; it is fraction of 1 hPa).
> So the only way to show "real" value is to calibrate the relative pressure
> (barometer in Weewx's terms) on the station's display and set
> "prefer_hardware" in weewx.conf? I'm really confused.
You are confused because you don't know what's coming over the wire from
the station, and you don't know if the station pressure is
miscalibrated.
> Or maybe I have some units in weewx.conf messed up? I use metric system so
> I entered meters when it was possible.
Maybe, but I don't think we can conclude that.
First, read the driver "fousb.py" in the sources. Even if you don't
know python, look for "pressure calculations" in the code and read the
comment.
Looking at the driver, it looks like "abs_pressure" is the only thing
actually read from the station.
I would suggest:
look at your database and the values for station pressure ("pressure")
and barometric pressure. Maybe change your skin to also graph station
pressure.
run for an hour with barometer set to prefer_hardware, and for an hour
with it set to prefer_software. Or longer, doesn't matter. See how
pressure/barometer relate and if it is any different. I am just not
seeing a hardware path for barometer.
probably conclude that the behavior is the same, but that
prefer_software is cleaner because it documents what happens.
run for an hour with your newly-concluded-as-right prefer_software
setting and the console as is.
go into cal mode on the console and set the pressure 5 hPa higher than
it is, 50 clicks up assuming 0.1 hPa. Yes I know this is wrong but
it's an experiment.
run for another hour
50 clicks back down in cal setting, and a third hour
now look at the dB and examine (graph even better) station pressure
and barometer. I more or less expect either:
- station pressure was steady, and barometer also
- station pressure changed 5 hPa up and back down, and barometer
also roughly 5
This tells you if the console calibration is just changing the console
display, or if it affects the station pressure reported to the
computer.
Probably, the path to getting this right is:
if station pressure as reported to weewx is unchanged by console
calibration, calibrate console to match barometer from nearby official
station, then ignore console relative to weewx and consider it just
for local humans. Then, look at station pressure in weewx and compare
to your friend's calibrated baromoeter and figure out the offset. It
should be only a few hPa. Put this in a StdCalibrate section, which
is correcting the hardware error in your pressure sensor. If it's
more than a few hPa, post actual data.
If station pressure as reported to weewx is modified as you change
console calibration, then have your friend with a calibrated barometer
(which can measure station pressure) visit and adjust the console
calibration to match that reading. This will abuse the console's
mechanism (intended to badly calculate barometric pressure) instead as
a mechanism to calibrate out the sensor error. Weewx will get correct
station pressure, and the console will show station pressure, not
"barometric pressure".
Now, weewx's station pressure from the hw is correct, and if your
elevation is right the barometric pressure should be close to the
nearby official station. Examine the barometer vs the official
station and see how it tracks. Please report back!
I will speculate wildly a few ways
1) The pressure reported by the console to weewx is changed by the
calibration process, and the sensor itself is ok
You say 550m, and for 1013.25 hPa at 0 m elevation and 20 C,
station pressure should be around 950 hPa. If the calibration
causes the device to report 1013.25, and then weewx calculates, it
should get about 1080 (which is impossibly high, but is what
barometric pressure would be if station pressure at your altitude
was 1013.25 hPa.
2) The pressure reported is not changed, but the sensor reads low.
weewx is getting a wrong station pressure.
Next message, please give actual pressure values, rather than "too
high", as this is too hard to follow. Include all of, from the same
time:
if you have any stdcalibrate lines
elevation of console (550m so far)
'relative' display on the console
station pressure from another device physically at your console
barometric pressure at official station
station pressure at official station
official station's altitude
"pressure" and "barometer" from the weewx database
Greg
--
You received this message because you are subscribed to the Google Groups
"weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/weewx-user/rmi4jkffpvl.fsf%40s1.lexort.com.
