Gotcha. It’s definitely a good way to turn things on (which I’m already doing with my light sensors), but I was hoping it could also provide some intelligence to turn things off. In retrospect, that would be too much to hope for. Still, a really interesting approach.
You could use a weather service to modify Astro’s (maximum) prediction, to improve the guesswork. I’ve no idea how much you take off for “overcast” or “rain”, but a few days experimentation might give a start.
I was thinking that, too. The ultimate solution would be to monitor a radar map, but that is waaaaaayyyy beyond my skill set.
What I do works fairly well. It’s not perfect but serviceable. I simply use OWM’s cloudiness channel which is a percent. If it’s over 60% I turn the lights on. Even here where weather is so localized that one could be in full sun in a downpour it always turnes the lights on when they are needed and infrequently turns them on when they are not needed.
That’s a good idea! I’ll try that combined with diffused radiation value when I come back and see if that provides any better results. Thanks for the tip