Microwave emissivity of freshwater ice, Part II: Modelling the Great Bear and Great Slave Lakes

TL;DR

This study models microwave emissivity of freshwater ice on Great Bear and Great Slave Lakes, demonstrating high accuracy outside melt seasons and highlighting the need to address biases during melt periods.

Contribution

It introduces a combined modeling approach using CLIMo and radiative transfer to predict ice emissivity, improving understanding of microwave signatures of lake ice.

Findings

Strong correlation (0.926) at 18 GHz during non-melt seasons

Model residual of 0.78 Kelvin at 18 GHz

Biases observed during melt season due to snow dirt content

Abstract

Lake ice within three Advanced Microwave Scanning Radiometer on EOS (AMSR-E) pixels over the Great Bear and Great Slave Lakes have been simulated with the Canadian Lake Ice Model (CLIMo). The resulting thicknesses and temperatures were fed to a radiative transfer-based ice emissivity model and compared to the satellite measurements at three frequencies---6.925 GHz, 10.65 GHz and 18.7 GHz. Excluding the melt season, the model was found to have strong predictive power, returning a correlation of 0.926 and a residual of 0.78 Kelvin at 18 GHz, vertical polarization. Discrepencies at melt season are thought to be caused by the presence of dirt in the snow cover which makes the microwave signature more like soil rather than ice. Except at 18 GHz, all results showed significant bias compared to measured values. Further work needs to be done to determine the source of this bias.