Since radmon.org is logging all values, it would be straight-forward to set warning- and alert-levels automatically depending on the recorded range of values from a certain station.
During a calibration phase the average value and standard deviation could be calculated.
Warning-Level could then be set for example to 4-sigma (0.00317% of all values, equivalent to one false warning every 22 days when reporting once per minute)
Alert-Level could be for example 4.5-sigma (0.000345% of all values, or statistically occurring once every 201 days) or maybe even 5 sigma which would be equal to one false alert every 6.3 years.
With 132 active stations the entire network would produce statistically one false alert every 17 days.
I think that would make alerts more meaningful and make the levels easier to compare and observe.
What do you think of this idea?