Weather station "predictions" automation / blueprint?

Hi everybody,

I have a GW1100A weather station that connects to Home Assistant via Ecowitt integration. Currently, it provides me with these measurements

- entity: sensor.gw1100a_v2_2_3_absolute_pressure
  name: Absolute Pressure
- entity: sensor.gw1100a_v2_2_3_daily_rain_rate
  name: Daily Rain Rate
- entity: sensor.gw1100a_v2_2_3_dewpoint
  name: Dewpoint
- entity: sensor.gw1100a_v2_2_3_event_rain_rate
  name: Event Rain Rate
- entity: sensor.gw1100a_v2_2_3_feels_like_temperature
  name: Feels like Temperature
- entity: sensor.gw1100a_v2_2_3_hourly_rain_rate
  name: Hourly Rain Rate
- entity: sensor.gw1100a_v2_2_3_humidity
  name: Humidity
- entity: sensor.gw1100a_v2_2_3_indoor_dewpoint
  name: Indoor Dewpoint
- entity: sensor.gw1100a_v2_2_3_indoor_humidity
  name: Indoor Humidity
- entity: sensor.gw1100a_v2_2_3_indoor_temperature
  name: Indoor Temperature
- entity: sensor.gw1100a_v2_2_3_max_daily_gust
  name: Max Daily Gust
- entity: sensor.gw1100a_v2_2_3_monthly_rain_rate
  name: Monthly Rain Rate
- entity: sensor.gw1100a_v2_2_3_outdoor_temperature
  name: Outdoor Temperature
- entity: sensor.gw1100a_v2_2_3_rain_rate
  name: Rain Rate
- entity: sensor.gw1100a_v2_2_3_relative_pressure
  name: Relative Pressure
- entity: sensor.gw1100a_v2_2_3_solar_lux
  name: Solar Lux
- entity: sensor.gw1100a_v2_2_3_solar_radiation
  name: Solar Radiation
- entity: sensor.gw1100a_v2_2_3_total_rain
  name: Total Rain
- entity: sensor.gw1100a_v2_2_3_uv_index
  name: UV Index
- entity: sensor.gw1100a_v2_2_3_weekly_rain_rate
  name: Weekly Rain Rate
- entity: binary_sensor.gw1100a_v2_2_3_wh65_battery
  name: WH65 Battery
- entity: sensor.gw1100a_v2_2_3_wind_direction
  name: Wind Direction
- entity: sensor.gw1100a_v2_2_3_wind_gust
  name: Wind Gust
- entity: sensor.gw1100a_v2_2_3_wind_speed
  name: Wind Speed
- entity: sensor.gw1100a_v2_2_3_windchill
  name: Windchill
- entity: sensor.gw1100a_v2_2_3_yearly_rain_rate
  name: Yearly Rain Rate

Yesterday we had a storm and lots of rain. Right before, absolute pressure, relative pressure, humidity, solar lux changed significantly.



(no screenshot for relative pressure as it seems to be almost, if not exactly, the same as absolute pressure.

Other values changed as well (less UV index, more on all rain entities, higher wind speed, etc.).

Do any of you that use weather station have some kind of “reliable prediction” routine to determine whether or not there might be rain / thunderstorm / “bad weather” coming very soon? I mean, yeah, there are weather entities that already tell me that it might rain today. But these are for the general area. This morning, my parents living ~7km / ~4.3 miles away had rain, nothing at my place. Now it has shifted to here, while they don’t have any rain at the moment. So this general area warning is too general to actually determine my exact locations weather conditions.

A use case might be to get a warning “if there is laundry drying outside, collect it NOW” because it might start raining very soon. Or (if the prediction can be made for a point in time not that close) “don’t take the dogs for a walk now; it might look sunny, but will rain in x minutes”. Just some ideas.

Currently, pressure is still up. Then again, the weather is still kinda bad. I vaguely remember something about being able to determine thunderstorms if x (pressure drops, then rises??? I cannot remember). While I could just research this, I am sure it is more complicated than that. So I cannot just use two input_booleans (one if pressure drops by x percent, and one if after this it rises by y percent); especially now: pressure is still up, I believe it is a little bit higher than usually (but cannot say for sure as I haven’t had this weather station for long).

And yes, the device will tell me current rain status as well - but by the time rain has already starting falling, it is too late. By this point, I would know because I can hear or see the rain. The warning about getting the laundry inside would need to be sent before it actually has started to rain.

So I thought, perhaps somebody here actually knows these physical rules that one could utilize for an automation or script… if this is the case, would you be willing to share yours?

Thank you in advance for your input :slight_smile:

You might try using the Bayesian integration to work out the probability of rain based on your observations.

It can be a bit brain-numbing to set up, but there’s an excellent tutorial here which includes an interactive spreadsheet to work out the observation probabilities for you.

Once you get it right, I’ve found it to be really accurate.

Edit: To be clear, I’ve never tried weather…

@prankousky your topic is very interesting, did you find a solution with Bayesian sensor? i’m studying it but is a bit complicated for me