According to the Zimmerman algorithm:
70F is used as the reference temperature (i.e. temperature above that will result in more watering and temperature below that will result in less watering). Because the algorithm calculates a watering percentage, it will go smoothly to 0 as temperature lowers.
I am trying to think how to balance customization need vs. simplicity. As is, you can always customize the weather script (which is written in Python) and host it on your own server. This way you have complete control over how you’d like to implement the rules. However, this will be a technical barrier for average users. I think it should be possible to support a per-user, customizable script stored on the cloud server. Users can modify the script according to their own need (and we can provide a few default ones). This will provide the most flexibility.