IoT Automated Weather Station network
Eight field-deployed AWS nodes across the Mumbai and Pune corridor report temperature, relative humidity and wind every thirty seconds, giving the platform the ground truth it needs to check a forecast before anyone acts on it.
| Node ID | Deployment site | Temp | RH | Wind (km/h) | Battery | Status | Last seen |
|---|
Forecast validation against ground observation
Model output for each station location compared with the co-located AWS reading
Node hardware
Each station pairs a shielded temperature and humidity sensor with an anemometer and a barometric sensor, read by a low-power microcontroller and published over MQTT. A solar panel and a lithium-iron cell carry the node through cloudy days and overnight.
- SHT31 temperature and relative-humidity sensor in a radiation shield
- BMP280 barometric pressure and altitude sensor
- Cup anemometer with reed-switch pulse counting
- ESP32 controller with Wi-Fi and NB-IoT fallback
- 10 W solar panel with a 6600 mAh LiFePO4 pack
Telemetry payload
Every reading carries the node identifier, a timestamp, the coordinates and a battery figure, so a suspect series can be traced back to a specific device and power state during quality control.
{
"node_id" : "AWS-PNQ-02",
"ts" : "2026-09-02T11:45:00+05:30",
"lat" : 18.508,
"lon" : 73.926,
"temp_c" : 41.2,
"rh_pct" : 27,
"wind_kmph" : 12.1,
"pressure" : 948.6,
"battery" : 79,
"rssi" : -68
}Readings that fail a range or rate-of-change check are flagged and withheld from validation rather than silently dropped.