jotone/pcfvg-weatherstations

FVG Protezione Civile — Weather & Hydro Monitoring Network

Near-real-time observations from the environmental monitoring networks operated by the Protezione Civile della Regione Autonoma Friuli Venezia Giulia (north-eastern Italy): temperature, humidity, precipitation, wind, pressure, global radiation, soil temperature/moisture, snow, river level/discharge, and sea/wave variables, at ~15-minute resolution across the regional station network.

This dataset is a tidy, archived snapshot of the public API at https://monitor.protezionecivile.fvg.it/, packaged as Parquet for analysis and ML.

Attribution

Data source: Protezione Civile della Regione Autonoma Friuli Venezia Giulia — Via Natisone 43, 33057 Palmanova (UD), Italy. https://monitor.protezionecivile.fvg.it/

License: Creative Commons Attribution 4.0 International (CC BY 4.0) — https://creativecommons.org/licenses/by/4.0/

The data belongs to the source organization; this repository only collects, formats, and redistributes it under the same CC BY 4.0 license declared by the source API. If you use this dataset, you must credit the Protezione Civile FVG as above and indicate if changes were made.

License note. The source API explicitly declares its data as CC BY 4.0 (in its OpenAPI specification, info.license). The organization's general website legal notice (https://www.protezionecivile.fvg.it/note-legali) is a broader, site-wide copyright statement. This dataset is published under the specific CC BY 4.0 license that the API attaches to the data itself.

Dataset structure

Config	File(s)	Description
measurements (default)	data/{station_id}/measures.parquet	One row per sensor reading; one consolidated file per station.
stations	stations.parquet	One row per station (metadata).
sensors	sensors.parquet	One row per sensor (maps sensor_id → physical quantity/unit).

measurements columns

Column	Type	Description
station_id	int64	Station identifier (join key to stations.id).
sensor_id	int64	Sensor identifier (join key to sensors.id).
dt	timestamp[us, UTC]	Observation time, UTC.
value	float64	Measured value (unit given by the sensor's unit). May be null ("no reading").
lat	float64	Station latitude (EPSG:4326).
lon	float64	Station longitude (EPSG:4326).

stations columns

id, code, name, istat, lat, lon, alt, status

sensors columns

id, station_id, code, name, quantity (physical quantity measured), unit, status

To interpret a measurement, join measurements.sensor_id → sensors.id for the quantity/unit, and measurements.station_id → stations.id for station name/location.

Usage

from datasets import load_dataset

# Measurements (default config)
measures = load_dataset("jotone/pcfvg-weatherstations", split="train")

# Metadata
stations = load_dataset("jotone/pcfvg-weatherstations", "stations", split="train")
sensors  = load_dataset("jotone/pcfvg-weatherstations", "sensors", split="train")

Or read the Parquet directly with polars / pandas / duckdb.

Coverage & provenance

• Region: Friuli Venezia Giulia, Italy.
• Network size: ~365 stations / ~150 sensor types (varies over time).
• Cadence: ~15-minute observations; near-real-time.
• History: the source API exposes a rolling window (~14 months); this archive grows as it is polled over time, so coverage depends on when collection started.
• Times are UTC. The dt column is UTC; derive a calendar day from it if you need to partition.

Limitations

• Values are as published by the source (not independently quality-controlled here); some series contain nulls or sensor-offline gaps (status).
• The source API offers only a rolling history, so periods before this archive began are not recoverable from it.

Citation

@misc{fvg_protezionecivile_weather,
  title  = {FVG Protezione Civile — Weather & Hydro Monitoring Network},
  author = {Protezione Civile della Regione Autonoma Friuli Venezia Giulia},
  url    = {https://monitor.protezionecivile.fvg.it/},
  note   = {Redistributed under CC BY 4.0},
}