Assessment of Machine Learning-Driven Retrievals of Arctic Sea Ice Thickness from L-Band Radiometry Remote Sensing

Special Issue Machine Learning and Statistical Learning with Applications 2025.-- 18 pages, 5 figures, 5 tables.-- The data supporting the conclusions of this article will be made available by the authors on request. These data were derived from the following resources available in the public domain: The production of SMOS sea ice thickness data was funded by the ESA project SMOS & CryoSat-2 Sea Ice Data Product Processing and Dissemination Service, and data from 15 October 2010 to 15 April 2021 were obtained from AWI. The BGEP data used in the validation were collected and made available by the Beaufort Gyre Exploration Program based at the Woods Hole Oceanographic Institution (https://www2.whoi.edu/site/beaufortgyre, accessed on 24 July 2025) in collaboration with researchers from Fisheries and Oceans Canada at the Institute of Ocean Sciences. Remote sensing data processing was executed at the Barcelona Expert Center on Remote Sensing (BEC-RS, https://bec.icm.csic.es, accessed on 24 July 2025) of the Institut de Ciències del Mar ICM-CSIC. The source codes are available for downloading at the following link: https://github.com/ferranhema/smos-mlsit (accessed on 24 July 2025)

Data availability: Hernández-Macià, Ferran; Gabarró, Carolina; 2025; BEC SMOS Sea Ice Thickness [Dataset]; DIGITAL.CSIC; https://doi.org/10.20350/digitalCSIC/17749; http://hdl.handle.net/10261/407464

This study evaluates machine learning-based methods for retrieving thin Arctic sea ice thickness (SIT) from L-band radiometry, using data from the European Space Agency’s (ESA) Soil Moisture and Ocean Salinity (SMOS) satellite. In addition to the operational ESA product, three alternative approaches are assessed: a Random Forest (RF) algorithm, a Convolutional Neural Network (CNN) that incorporates spatial coherence, and a Long Short-Term Memory (LSTM) neural network designed to capture temporal coherence. Validation against in situ data from the Beaufort Gyre Exploration Project (BGEP) moorings and the ESA SMOSice campaign demonstrates that the RF algorithm achieves robust performance comparable to the ESA product, despite its simplicity and lack of explicit spatial or temporal modeling. The CNN exhibits a tendency to overestimate SIT and shows higher dispersion, suggesting limited added value when spatial coherence is already present in the input data. The LSTM approach does not improve retrieval accuracy, likely due to the mismatch between satellite resolution and the temporal variability of sea ice conditions. These results highlight the importance of L-band sea ice emission modeling over increasing algorithm complexity and suggest that simpler, adaptable methods such as RF offer a promising foundation for future SIT retrieval efforts. The findings are relevant for refining current methods used with SMOS and for developing upcoming satellite missions, such as ESA’s Copernicus Imaging Microwave Radiometer (CIMR)

This project is funded from the AEI with the ARCTIC-MON project (PID2021-125324OB-I00), and from a Doctorat Industrial (AGAUR), with expedient number 2023 DI 0007.

This work is supported by the Spanish government through the “Severo Ochoa Centre of Excellence” accreditation (Grant CEX2024-001494-S funded by AEI 10.13039/501100011033)

Peer reviewed