Platform
Sensors
The index is assembled from public networks that publish real measurements — national meteorological services, hydrological agencies, transport authorities, research mesonets and community projects. Nothing is simulated and nothing is inferred.
What counts as a sensor
The bar for inclusion is deliberately concrete: an instrument that returns a numeric value with a timestamp. That test excludes a lot of things that superficially look like sensor data — station registries with no readings, dataset catalogue rows, survey records, administrative boundaries.
It is not a cadence test. A profiling float reporting every ten days and a snow gauge reporting daily both qualify; what disqualifies a record is having no instrument behind it.
Categories in the index
- Weather and climate — temperature, humidity, pressure, wind, precipitation.
- Air quality — particulates, nitrogen dioxide, ozone and related pollutants.
- Water — river level and discharge, groundwater, tide gauges, water quality.
- Marine — buoys, sea surface temperature, wave and ocean profiles.
- Traffic and mobility — vehicle counts, road sensors, bike share, parking occupancy, EV charging.
- Earth and environment — seismic motion, radiation, soil, GNSS ground displacement.
Normalisation is most of the work
Every network names things differently. One publishes air temperature as "TA", another as "air_temp", a third as a BUFR code in Kelvin. Units differ, timestamp formats differ, and identifiers collide between networks.
Readings are mapped onto a common set of measurand names with consistent units, while the original values stay traceable. This is unglamorous and it is the entire reason cross-network search is possible.
History
Readings are retained per sensor across five windows — one hour, one day, one week, one month and one year — so a sensor can be read as a trend rather than a single instantaneous value.