About This Weather Station

How the readings get here

Every number on this site that describes the weather right here is measured in the yard, not downloaded from anywhere. An Acurite Atlas mounted at the chicken coop broadcasts a packet over 433 MHz every few seconds. A Raspberry Pi 4 in the house listens for those packets with a software defined radio dongle and a 433 MHz antenna, decodes them with rtl_433, and hands them to WeeWX, which records a row in its database every five minutes and publishes each live reading to the website over MQTT. That is why the figures on the home page move without the page reloading.

In the case that the Acurite station goes down, a custom driver on the Pi takes over. If no Atlas packet arrives for five minutes it falls back to an Ecowitt WS90 read over Bluetooth through Home Assistant, and switches back on its own the moment the Atlas returns. Rainfall is the exception: it is only ever taken from the Atlas, because two rain counters telescoping into each other produced some memorably fictional storm totals. The status table above says which station is reporting at this moment.

What measures what

Temperature, humidity, wind speed and direction, rainfall, UV, light	Acurite Atlas, 433 MHz radio
Lightning strike count and distance	Acurite Atlas lightning detector
Barometric pressure and indoor humidity	BME280 sensor wired to the Pi
Air quality	PurpleAir PA-II on the local network
Backup for everything but rain	Ecowitt WS90, over Bluetooth through Home Assistant
Feels like, dewpoint, wind chill, heat index, evapotranspiration	Calculated by WeeWX from the above

A note on wind height

The Atlas sits about ten feet above the ground. The standard height for measuring wind is ten meters, roughly thirty three feet, which is where airports and official stations put their anemometers. Air is slowed by friction with the ground and whatever stands on it, so a sensor at this height reads lower than one at the standard, commonly by fifteen to twenty five percent, and more where trees or buildings break the flow. Gusts are blunted the same way. Read the wind here as an honest record of what the yard is doing, not as a figure to set beside an airport's.

A note on the air quality number

The air quality index shown on the home page is computed here, from the PurpleAir sensor in the yard, and it will usually read lower than the number on the sensor's own page on the PurpleAir map. Two things account for that. The raw particulate reading is corrected with the EPA adjustment for this hardware, which reads high in humid air, and the corrected value is then NowCast weighted across recent hours rather than taken as an instant snapshot. The AQI scale is defined on a daily average, so a two minute sample pushed through it would overstate a passing puff of smoke. The method here is the one AirNow uses to publish a number from sensors like this one. When the air is changing quickly, about half the value comes from the last hour.

What comes from elsewhere

• Forecast, watches and warnings come from the National Weather Service office in Indianapolis through its public API. No account, no vendor.
• Sky conditions and visibility in the current conditions line come from the automated station at Batesville airport, about fourteen miles away, falling back to Columbus if Batesville goes quiet. The Atlas cannot see cloud cover or fog, so that one line is borrowed.
• The hourly forecast page is drawn here from the Weather Service gridded forecast, refreshed every ten minutes.
• The rain comparison table reports this station's daily manual gauge reading to CoCoRaHS as station IN-DC-7, alongside neighboring observers.
• The live photo is a camera in the yard, refreshed every minute.

Where this station's data goes

Observations are uploaded continuously to several public networks, so the data outlives this website:

• KINMILLH2 on Weather Underground
• FW6569 on CWOP, which feeds the NOAA MADIS network used in forecast models
• KINMILLH2 on PWSWeather
• Millhousen on AWEKAS
• Listed in the WeeWX station registry

Software

• WeeWX 5.1.0, recording and uploading, on Raspberry Pi OS
• The Belchertown skin 3.0.0 by Pat O'Brien, which builds these pages
• rtl_433 for decoding the radio packets
• Highcharts for the charts
• MQTT for pushing live readings to your browser, and Caddy for serving the site
• Local Python for the radio and Bluetooth failover, the air quality calculation, the hourly forecast page and the CoCoRaHS comparison

Credits

• Pat O'Brien, for the skin this site is built on
• Tom Keffer, Matthew Wall and the WeeWX contributors
• The CoCoRaHS observers nearby, whose gauges keep this one honest