Negative spikes in Sense Hat pressures

[TideMan]

I'm monitoring the pressure from my sense hat every 10 minutes, and almost (but not quite) every day there is a negative spike of 10 or so hPa at 00:10 UTC. Local time is UTC + 12, so it's happening at 12:10 local time.
Anyone else seen this?
It's on an RPi2 running Jessie.

[buja]

I guess the question is: what happens at 12.10 local time (almost) every day?

My idea is that the pressure sensor picks up on an external event like a door or window opening or closing, an airco fan switching on or off, maybe the sensor is suddenly in the shadow by the movement of the sun. Many possibilities here, assuming the Sense Hat is working correctly.

[TideMan]

A drop of 10 hPa in atmospheric pressure is much larger than would be expected from local changes in the atmosphere, e.g. a door opening and closing, or the sun suddenly striking the sensor. Such changes occur, but over days or weeks, not minutes.
But 10 hPa is within the expected range, so it does not seem large enough to be an electronic/transmission fault, or a crontab problem.
And why does it only occur some days?

I can handle the spikes using a simple despiking algorithm, but the cause is intriguing...................

[DougieLawson]

Do you have the right version of RTIMUlib installed?
Is a task getting killed by the OOM killer?

10 hPa is worth about 350 metres of elevation change (close to sea level) so it won't be due to some ambient change.

[buja]

A drop of 10 hPa in atmospheric pressure is much larger than would be expected from local changes in the atmosphere, e.g. a door opening and closing, or the sun suddenly striking the sensor. Such changes occur, but over days or weeks, not minutes.
But 10 hPa is within the expected range, so it does not seem large enough to be an electronic/transmission fault, or a crontab problem.
And why does it only occur some days?

I can handle the spikes using a simple despiking algorithm, but the cause is intriguing...................

I was not aware that a 10 hPa pressure drop is significant, but I now I have a Pi monitoring a weather station with Weewx and I see that it's quite a lot in a meteorological sense.

Some suggestions for things you could try to find out what is happening:

• Increase the measuring frequency from once every 10 minutes to 5 minutes or even 1 minute. Maybe the spike does happen every day, but you just don't see it because it's between measurements.
• Change the time on the Pi (play with Daylight Saving Time or choose another timezone). Does the spike move with the wrong time, or does it keep true local time?

[TideMan]

Yes, good suggestions.
I'll record continuously over noon and see what happens..............

[buja]

Any luck finding out what is happening?

[TideMan]

It hasn't happened since my last post (18-Nov).
Perhaps that fixed it?