Go to content

Year: 2020
Project manager: Andrea Marinoni
Project code: 502 932019

AMUSIC is a three-year project to address new methods to enhance sea ice characterization
from multi-sensor remote sensing data. By leveraging different physical principles of
acquisition, various sensors (e.g., satellite-based synthetic aperture radar (SAR) and altimeters,
passive microwave radiometers, multispectral, airborne laser scans) can grasp different aspects
of the sea ice medium. Integrating this information hence enables a new way of understanding
sea ice characteristics and dynamics. AMUSIC will produce a framework to automatically
extract reliable information about sea ice from multiple remote sensing datasets. Results will
be validated using complementary data (ship radar, Ice Watch) and in-situ data collected in the
field. AMUSIC is divided into four interrelated work packages. WP1 will generate reliable
labeled datasets to be used for training and validating models. WP2 will design an automatic
platform for multi-sensor remote sensing data fusion based on of deep learning (DL) protocols.
WP3 will focus on DL design optimization to guarantee accurate and useful characterization
and physical interpretation of data generated features and results. The framework will be
designed to support near real time operational use with low computational costs. WP4 will
include fieldwork to collect validation information. AMUSIC will take advantage of the results
obtained in the Fram Centre-funded ALSIM project and field data measured during e.g.
MOSAiC and Nansen Legacy field cruises. The project has high relevance to ongoing work at
all partner institutions. Results will be disseminated in scientific publications, outreach and a
capstone workshop.

Frost proof websites from: Gnist