Mei, M. Jeffrey
The snow depth on Antarctic sea ice is critical to estimating the sea ice thickness distribution from laser altimetry data, such as from Operation IceBridge or ICESat-2. Snow redistributed by wind collects around areas of deformed ice and forms a wide variety of features on sea ice; the morphology of these features may provide some indication of the mean snow depth. Here, we apply a textural segmentation algorithm to classify and group similar textures to infer the distribution of snow using snow surface freeboard measurements from Operation IceBridge campaigns over the Weddell Sea. We find that texturally-similar regions have similar snow/ice ratios, even when they have different absolute snow depth measurements. This allows for the extrapolation of nadir-looking snow radar data using two-dimensional surface altimetry scans, providing a two-dimensional estimate of the snow depth with ∼22% error. We show that at the floe scale (∼180 m), snow depth can be directly estimated from the snow surface with ∼20% error using deep learning techniques, and that the learned filters are comparable to standard textural analysis techniques. This error drops to ∼14% when averaged over 1.5 km scales. These results suggest that surface morphological information can improve remotely-sensed estimates of snow depth, and hence sea ice thickness, as compared to current methods. Such methods may be useful for reducing uncertainty in Antarctic sea ice thickness estimates from ICESat-2.
Mei, M. Jeffrey
Sea ice thickness has long been an under-measured quantity, even in the satellite era. The snow surface elevation, which is far easier to measure, cannot be directly converted into sea ice thickness estimates without knowledge or assumption of what
proportion of the snow surface consists of snow and ice. We do not fully understand how snow is distributed upon sea ice, in particular around areas with surface deformation. Here, we show that deep learning methods can be used to directly predict snow
depth, as well as sea ice thickness, from measurements of surface topography obtained from laser altimetry. We also show that snow surfaces can be texturally distinguished, and that texturally-similar segments have similar snow depths. This can be used to
predict snow depth at both local (sub-kilometer) and satellite (25 km) scales with much lower error and bias, and with greater ability to distinguish inter-annual and regional variability than current methods using linear regressions. We find that sea ice thickness can be estimated to ∼20% error at the kilometer scale. The success of deep learning methods to predict snow depth and sea ice thickness suggests that such methods may be also applied to temporally/spatially larger datasets like ICESat-2.