Chapter Citizen science and machine learning forecasting Ostreopsis cf ovata blooms

Abstract

The toxic benthic dinoflagellate Ostreopsis cf. ovata causes harmful algal blooms in Apulia region of Southern Italy. In this study the volunteers of a citizens’ observatory engaged with public research centres to apply a machine learning approach and develop a predictive modelling tool able to forecast O. ovata blooms. We applied the Quantile Regression Forest to draw up two models named Model4Cities and Citizens’Model. Model4Cities was trained with data of cell abundance detected by the Regional Agency of Environmental Protection of Apulia from 2010 to 2022 in the cities of Bisceglie, Molfetta, Giovinazzo and Bari where the microalgae proliferate at high rates. Citizen’sModel was trained with data of cell abundance detected by citizens in two sites within the coastline of Molfetta from 2016 to 2022. Both models show a good capacity to forecast O. ovata concentrations as function of meteorological open data with 81% and 89% of prediction accuracy.