I've started looking at creating an average of the diurnal cycle of the Earth's magnetic field as detected by my FGM-3h fluxgate magnetometer linked up to the Arduino mega in my Earth Magnetic field Observatory.

There are a great deal of problems with this due to different errors in the readings, so to start with I've simply taken the average reading on each day. I've used this to calculate the variance from the average at each 5 minute interval during each day, and then averaged all these together.

This doesn't get a nice result of the diurnal magnetometer reading cycle - so I've excluded everything outside one standard deviation to remove the outliers and end up with a pretty smooth curve showing the daily cycle where the magnetic field reading from the FGM-3h dips during the morning and spikes during the afternoon.

I haven't yet decided if the magnitude of the cycle varies with the seasons - but I'm working on that. Ultimately the goal is to subtract the diurnal variation from the current output to see just the spikes caused by unusual variations in the magnetic field.

See here for the live output.

fgm-3h daily cycle