W. Aldenhoff, C. Heuzé, and L.E.B. Eriksson (2018), Comparison of ice/water classification in Fram Strait from C- and L-band SAR imagery, Annals of Glaciology, vol 59, pp 1-12, doi: 10.1017/aog.2018.7.
Obs: W. Aldenhoff was then a PhD student under my supervision (graduated 2020).
Ice charts for navigation are created daily by manual interpretation of SAR C-band data mostly. We here trained a neural network to create ice charts automatically and made its classification more accurate (as visually checked) by including not only C- but also L-band data. L-band has a longer wavelength, and hence is able to better detect areas with thin ice in the ice pack.
The main limitation is that the two datasets are not acquired by the same satellite. For our application, that means that there was a varying time difference between the images. And the larger the time difference, the more the ice had drifted, and the more the two results differed. In practice however, this time difference can become an advantage: an algorithm using either dataset whenever data is received will create ice charts more often than is currently possible by using only C-band, and thus provide more up-to-date information to those who are navigating in ice-infested waters.