Initial Investigations into the Application of SMRT for Lake Ice Forward Modelling

Abstract:

The formation of lake ice is an important process in the northern hemisphere that is dependent on meteorological patterns while influencing the economy, culture, and recreation of local communities. Despite the value of lake ice, in situ measurements have declined in recent years with remote sensing becoming a crucial tool for its monitoring. Active microwave observations, specifically synthetic aperture radar (SAR), is the most common type of data for studying lake ice due to its ability to “see” in any weather condition or time of day. Recent technological advancements provides new understanding into the primary scattering mechanisms for microwaves interacting with lake ice. However, a comprehensive evaluation of the effect that changes in ice properties have on SAR backscatter is required, which forward modelling using radiative transfer models can provide. For example, the Snow Microwave Radiative Transfer (SMRT) allows for the simulation of ice columns with multilayer snow and ice mediums. This research makes use of the SMRT model to study the impact of various ice properties (thickness, bubble radius, porosity, and ice-water interface roughness) on backscatter throughout the ice season. Sensitivity testing for incidence angles between 0 and 70° with HH, VV, and VH polarizations reveal that ice-water interface roughness (co-polarized) and bubble radius (cross-polarized) have the largest impact on backscatter. RADARSAT-2 Quad-pol data for Noell Lake near Inuvik, Northwest Territories, Canada, the Canadian Lake Ice Model (CLIMo), and SMRT were used to calibrate ice-water interface roughness for an entire ice season and simulate backscatter from a C-band SAR sensor. Correlation between satellite measured and modeled backscatter (RADARSAT-2 and Sentinel-1 data) ranged from -0.45 to 0.91 and RMSE values from 1.12 to 5.68 dB, respectively. Discrepancies were identified during initial ice formation and ice decay due to the presence of cracks/deformations and roughness of the ice surface. This method was also used to estimate backscatter for areas where field measurements on Noell Lake were made in March 2019. The results of these initial model explorations provide the framework for future inversion experiments to retrieve lake ice parameters (i.e. ice thickness and ice-water interface roughness) using SMRT.