Mapping Arctic Sea-Ice Surface Roughness with Multi-Angle Imaging SpectroRadiometer

Sea-ice surface roughness (SIR) is a crucial parameter in climate and oceanographic studies, constraining momentum transfer between the atmosphere ocean, providing preconditioning for summer-melt pond extent, being related to ice age thickness. High-resolution estimates from airborne laser measurements are limited spatial temporal coverage while pan-Arctic satellite does not extend over multi-decadal timescales. Launched on Terra 1999, NASA Multi-angle Imaging SpectroRadiometer (MISR) instrument acquires optical imagery nine near-simultaneous camera view zenith angles. Extending previous work model specular anisotropy, training dataset of cloud-free angular reflectance signatures roughness, defined as standard deviation within-pixel lidar elevations, near-coincident operation IceBridge (OIB) data generated modelled using support vector regression (SVR) with radial basis function (RBF) kernel selected. Blocked k-fold cross-validation implemented tune hyperparameters grid optimisation assess performance, an R2 (coefficient determination) 0.43 MAE (mean absolute error) 0.041 m. Product performance assessed through independent validation by comparison unseen similarly surface-roughness characterisations pre-IceBridge missions (Pearson’s r averaged six scenes, = 0.58, p < 0.005), AWI CS2-SMOS sea-ice thickness (Spearman’s rank, rs 0.66, 0.001), known proxy. We present derived product at 1.1 km resolution (2000–2020) seasonal period OIB corresponding time-series analysis. Both our instantaneous swaths monthly mosaics show considerable potential detecting surface-ice characteristics such deformed rough ice, thin refrozen leads, polynyas.