Station hardware upgrade introduces rainfall error

Discussion of the Meteohub software package

Moderator: Mattk

Post Reply
kerberos
Fresh Boarder
Fresh Boarder
Posts: 8
Joined: Sat Sep 13, 2014 11:06 am
Location: Melbourne, Australia

Station hardware upgrade introduces rainfall error

Post by kerberos » Wed Aug 24, 2016 3:33 pm

I have been running two stations concurrently: an Oregon WMR100, connected to Meteohub and a Davis VP2, running standalone. As the WMR100 was increasingly unreliable I planned to retire it and connect the VP2 to Meteohub.

On Monday I took delivery of a logger for the VP2. On Tuesday the WMR100 lost contact with its sensors. Since it wasn't raining, this seemed like a good time to swap stations.

I deleted the WMR100 and disconnected it from Meteohub. I plugged in the cable to the VP2, defined a new station and adjusted the sensors. Unfortunately changing the stations created a significant discontinuity in the raw rainfall data.

As you probably know, WMR100 consoles stop transmitting data over their USB interface after seemingly random amounts of time. This is one of reasons that drove me to replace it.

The only way to re-establish communication is to power-cycle the console. Doing this zeros the rainfall accumulator. Consequently, the WMR100's total rainfall was 32.8mm (which is the amount of rain that had fallen since the last console reset).

The VP2 is much more robust and doesn't require frequent resetting. It also preserves its data when the power is off. Consequently its total (yearly) rainfall is 341.6mm. This is the correct year-to-date amount of rain at my station's location.

The two raw rain records either side of the changeover are:

Code: Select all

20160823085619 rain0 0 0 328
20160823093200 rain0 0 0 328
20160823095441 rain0 0 0 3416
20160823095543 rain0 0 0 3416
Meteohub has interpreted this to mean that 308.8mm of rain fell between 0932 and 0954.

I welcome any suggestions as to how I might eliminate the 308.8mm glitch from Meteohub while maintaining the correct yearly rainfall in the VP2 console.
Cameron

kerberos
Fresh Boarder
Fresh Boarder
Posts: 8
Joined: Sat Sep 13, 2014 11:06 am
Location: Melbourne, Australia

Re: Station hardware upgrade introduces rainfall error

Post by kerberos » Sun Sep 25, 2016 5:13 am

I've let this go for a month because I don't want to be unreasonably demanding. I think I have a genuine problem that is not addressed in the Meteohub documentation. I can find no mention of a similar issue by searching and reading this forum. If it's covered somewhere that I've missed, maybe some kind soul could give me a pointer.

It's possible that my problem has been ignored because I've described it inadequately. I welcome any suggestions as to how I might describe it more fully or clearly.

Cameron

Edit: correct a typo
Cameron

kerberos
Fresh Boarder
Fresh Boarder
Posts: 8
Joined: Sat Sep 13, 2014 11:06 am
Location: Melbourne, Australia

Re: Station hardware upgrade introduces rainfall error

Post by kerberos » Tue Jun 13, 2017 12:51 pm

I finally resolved this issue. Yesterday I concluded migrating my station's hardware and data to Weewx after experimenting with it since January this year. Meteohub has been permanently retired.

The new software has range filters that are applied to raw observation data. They are designed to prevent this kind of error from occurring. Cleansing erroneous input that passes through the filter is trivial. My 3-year summary dataset can be rebuilt in 20 minutes, a process that takes Meteohub almost 2 days. It now seems possible to migrate the previous 7 years of observations onto a single system.

I've enjoyed the time I spent tinkering with Meteohub. It was a vast improvement over VWS. It seems that the next chapter of my obscession with weather observation has started with a huge power-up.
Cameron

Post Reply