[rv-de]

Precisely speaking "dryness" cannot be normally distributed to begin with as it is a value on a finite interval.

[hwillis]

Normal distribution of logarithmic dryness, then.

[rv-de]

Sure, as an approximation. But using a more appropriate distribution matching the statistical model would be preferable.

[hwillis]

No, it's literally done with a log-normal distribution: https://www.hydrol-earth-syst-sci.net/12/1339/2008/hess-12-1...

Log-normal is extremely common in hydrology. Turns out an anomaly in this case is defined as outside one standard deviation[1], so in a perfectly normal year you would expect ~15.9% of soil to be dryer than normal.

[1]: http://edo.jrc.ec.europa.eu/documents/factsheets/factsheet_s...