A neural network algorithm for sea ice edge classification

Alhumaidi, Sami M.; Jones, W. Linwood; Park, Jun-Dong; Ferguson, S. M.

doi:10.1109/36.602524

Cited by 10 publications

(1 citation statement)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, a Bayesian approach is selected. Sea‐ice detection and characterization is particularly suited to machine‐learning based algorithms (Alhumaidi et al., 1997; Zhang et al., 2019), but this approach, although quite efficient, has the disadvantage of remaining a black box. More classical approaches are also usual, especially Bayesian ones (Belmonte Rivas et al., 2012; Belmonte Rivas & Stoffelen, 2011; Breivik et al., 2012; Lindell & Long, 2016a, 2016b; Meier & Stroeve, 2008; Otosaka et al., 2018).…”

Section: Introductionmentioning

confidence: 99%

Sea‐Ice Detection From Near‐Nadir Ku‐Band Echoes From CFOSAT/SWIM Scatterometer

Peureux

Longépé

Mouche

et al. 2022

Earth and Space Science

View full text Add to dashboard Cite

SWIM on board CFOSAT is the first spaceborne, low incidence, rotating scatterometer, aiming at measuring in near‐real time ocean waves spectra. With five off‐nadir beams at incidences between 2° and 10° plus one nadir beam, it covers the Earth in 13 days, including polar regions thanks to its polar orbit. This work aims at exploiting SWIM data over ice regions with two objectives. An off‐nadir data‐based sea‐ice flag is here proposed that allows, first, to eliminate sea‐ice polluted echoes for improving the wave spectrum retrieval, and second, to open perspectives for application of sea‐ice monitoring with near‐nadir Ku‐band active sensors. To this end, the signature of both open water and sea‐ice radar backscatter is parameterized into Geophysical Model Functions. Then, comparisons with observed profiles through a Bayesian scheme provide a probability of sea‐ice presence. After comparison with both model (ECMWF‐IFS) and radiometer (SSMI) derived reference data sets, the proposed flag is found to be ready for operational use. At latitudes greater than 40° in absolute value, the proposed flagging algorithm exhibits accuracies of approximately 98% for all beams compared to SSMI data. Beam to beam performances are characterized and show potential for the characterization of sea‐ice at Ku‐band.

show abstract