Check weather data before watering to save water.

[Jan-Frase]

Basic idea:
The user can already enter the location of their FarmBot as coordinates.

Based on these coordinates we could look up the weather/rain forecast.
This data could then be used to decide if watering should proceed or be canceled for now.

For example:
Depending on your location, especially during the hot summer months, water can be quite scarce.
A user might have a regimen set up that waters the garden every day. If it happens to rain a lot 10 minutes after watering, we wasted water.

By checking the rain forecast before watering and using it as a condition, this scenario could be avoided.

---

I would potentially be interested in implementing this feature myself.
Is a feature like this of interest to the FarmBot team?

[roryaronson]

Thanks for opening this issue :)

First, I'd like to point out that we already have a Lua example to Water Based on the Weather. I just updated this example today so that it caches the forecast as a custom setting to avoid overusing a free API. This might be the solution you're looking for or can branch from.

That said, the Lua example is really only useful for those willing to write custom code for their machines, and not suitable for more casual FarmBot users. Making the use of weather forecasts a 1st-class part of FarmBot's feature set is definitely something that would add a lot of value, but it will be a full-stack effort across the API, FarmBot OS, documentation, etc. Here's roughly what it would take, if you are interested in helping out for some (or all!):

• On each hour, the app's backend would need to get a weather forecast for all bots that, based on their timezone, are in the 3am to 4am hour (a good time to get the freshest data before any likely watering activities begin for the day). We would likely use a Google API for this, as we already use some of their cloud services.
• The relevant forecast values (precipitation, temperatures, frost, ??) would be saved with the datetime, most likely to a new data model, and then get synced to the FarmBot.
• The new values would need to be made available across our APIs (REST API, farmbot-js, Lua, farmbot-py).
• The docs would need to be updated.
• We would want to add some new helper functions such as get_forecast() for quickly retrieving key values in an ergonomic way.
• Relevant helpers, featured sequences, and documentation examples would need to be updated to make use of the new functionality. For example, maybe water() or get_curve() should support an argument like weather_aware=true|false.
• Add some display to the frontend (probably the time travel popup) to show the latest forecast data to users.

Other considerations:

• Should we save weather data over time for historical purposes?
• Should we get a farther out forecast than just the next 24 hours?