CMCC Seminar
28 February 2024, 10.00 am CEST
Title
Advances in machine learning techniques can assist across a variety of stages in sea ice applications
Advances in machine learning techniques can assist across a variety of stages in sea ice applications
Speaker
CMCC Bologna , ESYDA division
Abstract
Only five years ago, there was very little mention of the use of AI in sea ice modeling research. The development of machine learning (ML) techniques in recent years is progressing faster than previously supposed by our community and these methods show potential as tools across the different stages of the sea ice modeling workflow. AI approaches could be used for data pre- and post-processing in support of traditional numerical simulations, to replace subgrid-scale parameterizations, and to directly produce sea ice forecasts. The employment of AI to a broad range of sea ice applications will be showcased and discussed at the presentation, focusing in particular on the ongoing activities and future plans of the sea ice research group at CMCC. I will showcase how ML can be effectively used to correct atmospheric reanalyses, which are boundary forcings for our sea ice models, discuss its employment in designing parameterizations for sea ice thermodynamics based on in-situ Arctic observations, and finally highlight the skill of an AI Antarctic sea ice prediction based on a convolutional long short-term memory network.
Only five years ago, there was very little mention of the use of AI in sea ice modeling research. The development of machine learning (ML) techniques in recent years is progressing faster than previously supposed by our community and these methods show potential as tools across the different stages of the sea ice modeling workflow. AI approaches could be used for data pre- and post-processing in support of traditional numerical simulations, to replace subgrid-scale parameterizations, and to directly produce sea ice forecasts. The employment of AI to a broad range of sea ice applications will be showcased and discussed at the presentation, focusing in particular on the ongoing activities and future plans of the sea ice research group at CMCC. I will showcase how ML can be effectively used to correct atmospheric reanalyses, which are boundary forcings for our sea ice models, discuss its employment in designing parameterizations for sea ice thermodynamics based on in-situ Arctic observations, and finally highlight the skill of an AI Antarctic sea ice prediction based on a convolutional long short-term memory network.
The seminar will be held in the CMCC-Bo meeting room (second floor)