The other idea I've toyed with is an automatic plant watering system, in service of my aspiration to start growing my own hot peppers (ghost peppers, habaneros, carolina reapers, etc.)
so I have a lot to learn and figure out in processing the raw data.
If you're interested in weather / environmental "stuff" in a broader sense, you might also be interested in a lot of the other data the USG (and others) make available. NOAA, the USGS, etc. put out all sorts of neat datasets. You can get all the obvious stuff (temp, barometric pressure, etc.) as well as water levels in various creeks and lakes and rivers, discharge volume for same, ground conductivity, stuff about lightning, severe storms (tornados, hurricanes, etc.), wildfires, etc., etc., etc. I'm pretty sure somebody creative could find some really interesting applications for all of that stuff.
As far as processing the raw data, I'm more interested right now in deciding if I should average which two of the thermometer or barometer readings I've got, and also deciding what to use as a high/low temp for the day - I've had some data dropouts and server crashes where I missed the noon/1pm high temp.
I did get GOES satellite data with a RTL-SDR once a couple of years ago. The antenna is more important than they'd lead you to believe, as is where the antenna is - you pick up a lot of fuzz up to some degrees above the horizon if you place the antenna in an urban setting.
Bonus points, I can tinker with some probably over-complicated timeseries python classes for sqlite.
Time to first graph was quickest with influxdb for me.
Zerotier and nats.io connect the things.
I want to sprinkle some pico w's around for door/window position monitoring.
https://github.com/bediger4000/station
I'm hoping to detect the "tide" in the atmosphere with barometric pressure readings, but I haven't done anything towards that.
I'd also like to have an automated way to pick out high/low temperature for the day. I've got 2 thermometers taking temperatures. There's data loss caused by both sensors quitting and operational issues, so I have a lot to learn and figure out in processing the raw data.