Harmful algal blooms have been increasing in frequency in recent years, and attention has shifted from describing to modeling and trying to predict these phenomena, since in many cases they pose a risk to human health and coastal activities. Predicting ecological phenomena is often time and resource consuming, since a large number of field collected data are required. We propose a novel approach that involves the use of modeled meteorological data as input features to predict the concentration of the toxic benthic dinoflagellate Ostreopsis cf. ovata in seawater. Ten meteorological features were used to train a Quantile Random Forests model, which was then validated using field collected concentration data over the course of a summer sampling season. The proposed model was able to accurately describe Ostreopsis abundance in the water column in response to meteorological variables. Furthermore, the predictive power of this model appears good, as indicated by the validation results, especially when the quantile for predictions is tuned to match management requirements. The Quantile Random Forests method was selected, as it allows for greater flexibility in the generated predictions, thus making this model suitable as a tool for coastal management. The application of this approach is novel, as no other models or tools that are adaptable to this degree are currently available. The model presented here was developed for a single species over a limited geographical extension, but its methodological basis appears flexible enough to be applied to the prediction of HABs in general and it could also be extended to the case of other ecological phenomena that are strongly dependent on meteorological drivers, that can be independently modeled and potentially globally available.

A novel application of and an adaptable modeling approach to the management of toxic microalgal bloom events in coastal areas.

ASNAGHI, VALENTINA;PECORINO, DANILO;CHIANTORE, MARIACHIARA
2017-01-01

Abstract

Harmful algal blooms have been increasing in frequency in recent years, and attention has shifted from describing to modeling and trying to predict these phenomena, since in many cases they pose a risk to human health and coastal activities. Predicting ecological phenomena is often time and resource consuming, since a large number of field collected data are required. We propose a novel approach that involves the use of modeled meteorological data as input features to predict the concentration of the toxic benthic dinoflagellate Ostreopsis cf. ovata in seawater. Ten meteorological features were used to train a Quantile Random Forests model, which was then validated using field collected concentration data over the course of a summer sampling season. The proposed model was able to accurately describe Ostreopsis abundance in the water column in response to meteorological variables. Furthermore, the predictive power of this model appears good, as indicated by the validation results, especially when the quantile for predictions is tuned to match management requirements. The Quantile Random Forests method was selected, as it allows for greater flexibility in the generated predictions, thus making this model suitable as a tool for coastal management. The application of this approach is novel, as no other models or tools that are adaptable to this degree are currently available. The model presented here was developed for a single species over a limited geographical extension, but its methodological basis appears flexible enough to be applied to the prediction of HABs in general and it could also be extended to the case of other ecological phenomena that are strongly dependent on meteorological drivers, that can be independently modeled and potentially globally available.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11567/865223
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 22
  • ???jsp.display-item.citation.isi??? 18
social impact