From “Meh” Sensors to a Meteorologist in Your Pocket

It all started with an impulse buy on Amazon (as most dangerous hobbies do).
I grabbed an Ecowitt Wittboy Weather Station GW2001 — a beast of a device packed with every sensor imaginable: temperature, humidity, pressure, wind, rain, solar radiation, UV… basically the Swiss Army knife of weather gadgets.

And honestly? It’s a great device. But after the honeymoon phase, I hit a wall: all these raw numbers were cool, but not very actionable. I wanted a simple way to combine them into a smart weather component that could supercharge my Home Assistant automations.

Because raw sensor data is… dumb.

• Temperature says it’s 75°F — but is that under blazing sun or dense fog?
• A lux sensor tells you it’s dark — cool, but is that because of clouds, nightfall, or a biblical swarm of locusts?
• Rain sensors? They’re great at saying “yep, water just hit me”, but terrible at hinting that a storm is about to throw your patio furniture into the neighbor’s pool.

In short: sensors ≠ intelligence.

That’s where Micro Weather Station comes in.

---

Why I Built This Thing 🤓

I wanted my automations to be smarter than “turn on sprinkler at 6 AM no matter what.”
Dream goals:

• Skip watering if it’s actually raining, not just “maybe, possibly raining” according to some API three cities over.
• Adjust indoor lighting when the sky goes full doom and gloom.
• Batten down the hatches before the wind eats my BBQ grill.
• Make my HVAC react to real microclimate conditions instead of generic weather app vibes.

Problem: no integration gave me the conditions I needed. So I built one.

---

Meet Micro Weather Station: Your Hyperlocal Weather Brain 🧠

Instead of dumb numbers, it translates your sensors into actual weather conditions like sunny, cloudy, rainy, stormy, foggy. It’s like turning your backyard into a tiny NOAA office, minus the bureaucracy.

Why It’s Better Than Random Weather APIs

• 🎯 Accuracy that actually matters: My backyard ≠ airport weather station 20 miles away.
• 🌧️ Real-time rain: If a drop hits your sensor, you’ll know immediately.
• 💨 True wind detection: Terrain and houses mess with airflow — external stations don’t see that.
• ☀️ Cloud cover analysis: Uses solar radiation + UV, not guesswork.
• 🏡 Personalized microclimate: Tailored to your property, not some stranger’s.

---

The Secret Sauce: Science (But Fun)

Here’s the logic stack that Micro Weather Station uses to decide what’s happening outside:

Condition	What It Looks For	Priority
⛈️ Stormy	Rain + wind over 25 km/h	1
🌧️ Rainy	Precipitation active	2
❄️ Snowy	Rain + temps below 2°C	3
🌫️ Foggy	Humidity >98% + dewpoint + low sun	4
☀️ Sunny	Solar radiation >400 W/m² or high UV	5
⛅ Partly Cloudy	Radiation 100–400 W/m²	6
☁️ Cloudy	Low radiation (<100 W/m²)	7

Translation: it doesn’t just guess; it triangulates multiple sensors, checks dewpoints, and generally acts like that nerdy friend who always says “technically…”

---

Real-Life Automations That Don’t Suck

• Smart irrigation: No sprinklers while it’s raining. Your water bill and tomatoes will thank you.
• Storm prep: Phone buzzes “Storm incoming — tie down the trampoline.”
• Adaptive lighting: Cloud rolls in → lights gently brighten. Chef’s kiss.

It even has YAML examples you can drop right into Home Assistant.

---

Geek Corner 🛠️ The Math Behind the Magic

The heart of Micro Weather Station is its detection algorithm: a set of rules + calculations that take raw sensor inputs and turn them into meaningful weather states (rainy, sunny, foggy, etc.). Here are the key pieces and how they work together:

Input Variables

• Temperature (T)
• Relative Humidity (RH)
• Atmospheric Pressure (P)
• Rainfall / Precipitation Flag
• Wind Speed (W)
• Solar Radiation (S_rad)
• UV Index (UV)
• Dew point (T_dew) — either measured or calculated

Derived Metrics

1. Dew Point (T_dew)
   Calculated from temperature and RH using the Magnus formula. When ambient temperature is close to T_dew, air is saturated → fog or dew likely.
2. Solar & UV Thresholds
   High solar radiation (e.g., >400 W/m²) → sunny; medium → partly cloudy; low → cloudy.
3. Wind Speed Thresholds
   Wind >25 km/h combined with rain = stormy. Otherwise, used for nuance in forecasting and comfort detection.
4. Humidity & “Fog Logic”
   RH >98% + dew point close to current temperature + low solar = fog detected.

Priority-Based Decision Tree

The algorithm checks for severe states first, then moves down:

• Rain + high wind → Stormy
• Rain → Rainy
• Rain + near-freezing → Snowy
• High humidity + dewpoint match + low sun → Foggy
• Strong sun radiation/UV → Sunny
• Medium radiation → Partly Cloudy
• Low radiation → Cloudy (default)

This ordering ensures that urgent states like storm override milder ones like cloudy.

Why It Works

• Redundancy: Multiple sensors cross-check each other (no false “sunny” while it’s pouring).
• Prioritization: Severe conditions are detected first.
• Context Sensitivity: Uses combinations (temp + humidity, radiation + UV, etc.) rather than single raw values, giving accurate microclimate results.

---

What’s Next 🚀

The roadmap is spicy:

• Machine learning for even sharper predictions
• Longer-range forecasting
• Emergency weather alerts
• Energy optimization that saves $$ based on real conditions

---

Open Source, Baby 🎉

Micro Weather Station is MIT-licensed, open source, and waiting for you to install, fork, or star.

👉 Check it out here: caplaz/micro-weather-station

---

So, what about you? Have your weather-based automations ever betrayed you? Tell me in the comments — bonus points if your sprinklers once ran during a thunderstorm.