Descrizione del progetto
Svelare i misteri del nostro passato preistorico con l’intelligenza artificiale
La storia climatica della Terra può essere studiata analizzando le carote di ghiaccio, cilindri di ghiaccio estratti dalle lastre di ghiaccio della Groenlandia e dell’Antartide o dei ghiacciai alpini. Tra le impurità che si trovano nelle carote di ghiaccio ci sono particelle insolubili, come le particelle di vetro vulcanico o le particelle di origine biologica, come il polline e le alghe. Il rilevamento di queste particelle è fondamentale per comprendere le condizioni passate e le interazioni tra i componenti del sistema climatico. Il progetto ICELEARNING, finanziato dall’UE, svilupperà una tecnica per rilevare automaticamente le particelle insolubili nelle carote di ghiaccio utilizzando tecniche di riconoscimento dei modelli di intelligenza artificiale. Questi rivoluzionari metodi automatici e non distruttivi possono rivelare ulteriori informazioni sui cambiamenti climatici e ambientali avvenuti in Antartide negli ultimi 1,5 milioni di anni.
Obiettivo
The detection of insoluble particles trapped in ice or sediment cores, like pollen grains, foraminiferal and diatom assemblages, volcanic and dust particles represents the basis for paleoresearch on the biosphere, volcanism and oceanic and atmospheric realms. To date, except for ice core dust, this analytical goal is achieved during years of particle observations by manual microscopy. Artificial Intelligence predictive models are already applied to several research fields within geoscience, but up to date its implementation to paleoclimate is missing. With ICELEARNING, I aim to develop a two-phase routine for the automatic quantification of insoluble particles trapped in ice cores. The routine is based on a commercial Flow Imaging Microscope producing particle images from within melted ice samples. The images are then analyzed by Pattern Recognition algorithms which will be developed for automatic particle classification and counting. The routine will be specifically developed in order to be implemented in Continuous Flow Analysis (CFA) systems, therefore surpassing the traditional methods by providing continuous particle records from ice cores. ICELEARNING methodology is suitable to any diluted sample, thus representing a ground-breaking analytical advancement from ice core science to marine geology. This innovative routine is automatic and non-destructive, imperative prerequisites for the future Antarctic ice core project analytical measurements, aiming to retrieve a continuous climatic and environmental record covering the last 1.5 Myr. ICELERNING will be developed at Ca’ Foscari University of Venice with Prof. Carlo Barbante, leading expert in trace and ultra-trace level impurity detections in ice cores and with the University of Bergen, a top institution in marine geology and paleoceanography. This unique synergy, in addition to the proposer’s knowledge of CFA systems and machine learning techniques will provide the best preconditions for the project success.
Campo scientifico
- natural sciencesearth and related environmental sciencespalaeontologypaleoclimatology
- natural sciencesearth and related environmental sciencesgeologyvolcanology
- natural sciencesphysical sciencesopticsmicroscopy
- natural sciencesearth and related environmental sciencesatmospheric sciencesmeteorologybiosphera
- natural sciencescomputer and information sciencesartificial intelligencepattern recognition
Programma(i)
Argomento(i)
Meccanismo di finanziamento
MSCA-IF-EF-ST - Standard EFCoordinatore
30123 Venezia
Italia