I agree with rjalex comment. I’ve had to modify the OS code down to 2% for each degree F, and prefer that. My last smart commercial irrigation controller used the predicted high temp and 1% for each degree F. The Zimmerman algorithm uses the MEAN/Average (not MAX) temperature and 4%. It seems to work pretty well with 2% on the mean temperature. I do believe it would be great if we could plug in scale factors for each weather parameter to adjust watering. The OS does get a few of the parameters, then you can code your own based on those parameters, but again it would be nice to have more flexibility on the OS to allow the user to configure it.
The zimmerman algorithm takes into account only one days data point for temp and humidity even though most people don’t water every day (i.e. water every 2nd or 3rd day). Uses average humidity, temp from previous day and precip from yesterday and today. Highly recommend we take into account (or have the option of) humidity, precip. and temp for yesterday, today and tomorrow with a percent chance of rain % threshold to not water for today and tomorrow (forecast).
So having each parameter roll over 3 days for example, with a baseline for each (like 80F or 70F) and a multiplier like 4% or in our case 2% for each degree F would support more customization and more accurate & water savings.
FYI wunderground kept returning summary data for MEAN temp incorrectly that OS was using (it appeared to return MIN temperature instead of MEAN temperature). I contacted wunderground via email and a few days later it seems to have stopped doing that. Not sure if anyone else observed this a few weeks ago, but having the MIN temperature for the MEAN, and having 4% scaling factor, it would get to 80F outside and decide almost not water because the low was 55 at night.