info@agrizapp.it
AgriZapp — Project sheets Blooms & pollinators · 1/3

FLORAPIS

Bloom forecasting and hive monitoring, in service of precision beekeeping.

Knowing when and where melliferous plants will bloom is the difference between a good harvest and a lost season. FLORAPIS is the platform AgriZapp built for beekeepers: it cross-references in-hive sensor data with regional weather stations to anticipate the flowering windows of six nectar-bearing species, closing the loop with beekeepers' own field observations.

What we built

Geolocated apiaries & hives — GPS position history for nomadic beekeeping, moving hives to follow the blooms
In-hive sensor sync — weight, humidity and temperature collected from 3bee control units
Automated weather pipeline — daily data from 92 ARPAT/SIR Toscana stations plus 16-day forecasts
Phenology engine — growing degree-day thresholds per species, computed per weather cell and year
Sightings & alerts — beekeepers log observed blooms, the system notifies predicted windows
Six species tracked — clover, sulla, sainfoin, black locust, linden, chestnut

One hive, one whole winter

Weight (kg) Internal humidity (%)
34.6 kg 30.0 kg 100% 70% Dec 20 Jan 1 Feb 1 Mar 1 Mar 17
Hive "Florapis" (apiary "Casa") — weight and internal humidity logged roughly every 2 hours by the 3bee sensor, Dec 20, 2024 – Mar 17, 2025 (88 days, 1,169 readings). The slow weight decline is the colony's natural consumption of winter stores: the same kind of signal the model cross-references with weather forecasts to recognise the onset of a nectar flow.

How it works

FIELD CALIBRATION 1 Sensors Hives (weight, humidity, temp.) + 92 ARPAT stations 2 Phenology model Growing degree-day thresholds per species/cell 3 Bloom forecast Predicted window for each apiary 4 Alerts & sightings Notifies the beekeeper, collects the real observation
From raw data to a decision at the apiary: sensors feed the phenology model, which estimates the flowering window for each apiary; alerts notify the beekeeper, and their field observations flow back into the model as calibration data.