Huge jumps in rain fall

[Mark Fraser]

I have an Aercus Instruments WS3085 that I'm using with Weewx-SDR and
occasionally I get huge jumps (100mm) in the recorded rainfall. Like
yesterday when it recorded 100.58mm within 5 minutes.

It can't be the WS3085 as the base station didn't record this, unless it
is programmed to ignore huge jumps like this.

Is it possible to correct the database and/or is there some way for it
to ignore these huge jumps?

[Tom Keffer]

For correcting the database, see the wiki article Cleaning up old 'bad' data.

For preventing future events, see the section [StdQC] in the User's Guide.

-tk

--
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 weewx-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/21634571-6325-8c0f-b831-67d221e61ef3%40gmail.com.

[Mark Fraser]

On 21/12/2022 12:38, Tom Keffer wrote:
> For correcting the database, see the wiki article /Cleaning up old 'bad'
> data <https://github.com/weewx/weewx/wiki/Cleaning-up-old-'bad'-data>/.

>
> For preventing future events, see the section [StdQC]

> <http://www.weewx.com/docs/usersguide.htm#StdQC> in the User's Guide.

>
> -tk
>
> On Wed, Dec 21, 2022 at 12:19 AM Mark Fraser <mfr...@gmail.com
> <mailto:mfr...@gmail.com>> wrote:
>
> I have an Aercus Instruments WS3085 that I'm using with Weewx-SDR and
> occasionally I get huge jumps (100mm) in the recorded rainfall. Like
> yesterday when it recorded 100.58mm within 5 minutes.
>
> It can't be the WS3085 as the base station didn't record this,
> unless it
> is programmed to ignore huge jumps like this.
>
> Is it possible to correct the database and/or is there some way for it
> to ignore these huge jumps?

I already have
rain = 0, 10, inch
in [StdQC]. Does that refer to total rainfall or rain rate?

[Greg Troxel]

> I have an Aercus Instruments WS3085 that I'm using with Weewx-SDR and
> occasionally I get huge jumps (100mm) in the recorded rainfall. Like
> yesterday when it recorded 100.58mm within 5 minutes.
>
> It can't be the WS3085 as the base station didn't record this, unless
> it is programmed to ignore huge jumps like this.

(I don't use weewx-sdr but I do receive sensor data with rtl_433 and I
have seen bad data from rtl_433.)

The real question is what went specifically went wrong. I would guess
(and I do mean guess) that the gauge itself and the transmitted signal
were ok, and that the decoded signal from the sdr program was wrong.
With enough noise/errors on RF, it is not possible to have zero chance
of a decoding error, but I suspect it is possible to do better. It may
be that there is a CRC that is more robust than is being checked. It
may also be that the same report is sent 3 times spaced out a bit (in an
attempt to get through even if there is a noise burst) and that the sdr
received 3 and 1 was in error.

Looking quickly, I see that Acurite Tower and 6045 sends 3 copies,
Acurite 986 fridge/freezer 2, TPMS sensor 4, EcoWitt moisture 3,

rtl_433 intentionally just outputs each decode and does not attempt any
kind of cross-decode error processing. It also does not attempt to
print out (and I am 99% does not directly measure) the symbol errors
which would enable a confidence output, or cross-burst fusion in a
smarter way.

So, one path to improvement is to dig into the above issues. The
easiest is probably to see if there are multiple packet and do some kind
of combining of them, perhaps combined with a priori plausibility
checks.


The other things to do are

- get a better sdr dongle, if you don't
already have a good one. My recommendation is the Blog v3 with the
dipole (get a real Blog v3, not counterfeit). The NooElec Smart v4/v5
are also great, but the included mini-magmounts work less well than a
dipole. So that and a dipole will be good too. This will get you better
quality signals and make lots of things work better, and it's easy, even
if it is $45. The older dongles (no metal case, no txco) just
don't work as well. (Yes, I did a comparison, and it's in the
rtl_433 archives.)

- Use a USB extension cable and get the dongle away from the
computer. Try to have the antenna away from metal.

[Tom Keffer]

I already have
rain = 0, 10, inch
in [StdQC]. Does that refer to total rainfall or rain rate?

It refers to the amount of rain recorded in a LOOP packet or an archive record. So, you are saying no more than 10 inches of rain in an archive interval (typically 5 minutes or so). Clearly, that's way too high.