Impact of small-scale microstructure variations on passive microwave brightness temperature

Passive microwave observations have been used for over 30 years to gain information on snow mass on a global scale. Original algorithms to retrieve snow mass were necessarily simple because of the computational limitations of the era, but the simplifying assumptions of constant and uniform snow properties have been shown to result in significant errors in the retrieved snow mass, even after spatial averaging of snowpack heterogeneity to the 25km satellite footprint scale. There is a clear need for a new approach to retrieving snow mass from microwave remote sensing measurements. One of the key questions that will govern the design of a new retrieval system is how well the snowpack needs to be quantified in order to interpret the remote sensing measurements. Here, detailed information about the snow microstructure was used to drive a snow microwave emission model and assess how the microwave brightness temperature is affected by variations in snowpack layering. Measurements used to drive the microwave emission model were taken during a fieldwork campaign in Churchill, Manitoba throughout the winter of 2009-2010. One site was selected for this study, which had minimal vegetation. Despite only minor changes in the height of the snowpack surface, significant variation was observed in the snow water equivalent, layering and number and location of ice lenses. Measurements from this site provide a good test of the sensitivity of the snow microwave emission model to changes in snowpack microstructure. Ground-based passive microwave measurements were made along a transect. Immediately after these measurements were made, a 4m long trench was excavated in the microwave footprint. Near-infrared photography was used to image the near-vertical wall of the trench and images were digitally processed to georeference snowpack layer boundaries. After the images had been taken, snowpack properties such as temperature, density and grain size were measured throughout the trench. Vertical profiles of snowpack information were used to drive a snow microwave emission model at 1cm intervals along the snow trench. The microwave emission model chosen was the multiple-layer version of the HUT model. Small-scale variations in the snow microstructure were related to changes in the simulated brightness temperature. Significant changes in brightness temperature were simulated when ice lenses were present, and could not be reproduced when the mass of the ice lenses was accounted for as snow within other existing layers. This indicates that information on ice lenses may need to be known for interpretation of remote sensing measurements. Finally, simulated microwave brightness temperatures were compared to the passive microwave observations as a test of the multi-layer HUT model. Outcomes from this study will be used to inform the next generation of snow models and snow mass retrieval systems from remote sensing.