StarJelly prediction service
StarJelly service offers short-term prediction, almost like a remote detection of current jellyfish occurrence during the same day of the prediction.
The StarJelly data sources include:
- Input data: Earth Observation, oceanographic models, meteorological data
- Jellyfish sightings (presence and absence) as ground truth data
How does it work?
Variables such as water currents, chlorophyll, temperature and water mass characteristics exert an effect on jellyfish accumulation on coastal areas, under specific spatial and temporal windows. The StarJelly service is based on a computation model that follows a non-linear pattern recognition approach aimed at capturing such physical-biological interactions.
When you contract StarJelly for your area, we first conduct a pilot exercise in order train our system for the specific area. This trial requires regular retrieval of ground truth, i.e. in-situ observations by an agent of Jellyfish occurrence and associated conditions.
Once the pilot exercise is finished we provide a ready to use operational forecast service, together with associated expected service performance indicators.
Preliminary pilot exercises shall be conducted for an optimal selection of prediction mode for area and needs. Modelling and validation of StarJelly require jellyfish sightings. For areas where these are not available, it might be possible to execute StarJelly based on environmental cues only, and incorporate ground truth data when this becomes available. In any case, Star2Earth is committed to provide validation indicators of all measures for the user to evaluate the goodness of the current prediction.
- 2008-2010 EO JELLY, Forecasting of coastal jellyfish appearance using Earth Observation data, http://eojelly.starlab.org.
- 2009-2011 JELLYFOR: Forecasting of jellyfish occurrence given salinity, temperature and chlorophyll-a conditions. See the Projects section.