Multi-scale characterization of sub-pixel effects for microwave remote sensing of snow

Current methods to retrieve snow water equivalent (SWE) using passive microwave (PM) remote sensing measurements are often characterized by large uncertainties which result from the natural heterogeneity of snowpack and vegetation within the microwave footprint. Snowpack heterogeneities include snow grain size, snow depth, and layering of snowpack. Vegetation height, needle or leaf density, and species also vary within microwave footprints. It is not currently understood the extent to which the passive microwave measurement is sensitive (or insensitive) to different variables listed above, as a function of the scale of the microwave measurement. Our study characterizes these individual sensitivities in an attempt to improve the algorithms and methods that are currently used for the retrieval of snow properties. In our investigation, we utilized the multi-scale Cold Land Processes Experiment dataset for in situ and microwave datasets. In order to better characterize the effect that variability in the snowpack and vegetative states has on the microwave observation as function of scale, we conduct analysis in three different ways. 1) First, we estimate what we expect the observed radiances(brightness temperature) to be at specific locations, based on in situ and LiDAR derived snow data gathered from those same locations, using a radiative transfer model (RTM). We then compare our estimated radiances to the actual observed radiances, and characterize the errors, as a function of the variability of the different snowpack and vegetative properties. 2) Based on the "real" variability of snow and vegetative states, we will create synthetic observations of microwave brightness temperatures, and perturb the different snowpack and vegetative properties in order to examine the extent to which the synthetic microwave signal is affected by each individual property, at differing scales. Based on the synthetic analysis described above, there still exists sensitivity to the mean snow depth within microwave footprints, even with highly heterogeneous alpine snowpack. Furthermore, we find that the microwave signal is sensitive to the mean snow depth at differing spatial scales, which lends credence to the goal of using currently available passive microwave measurements, as well as the proposed ESA and NASA missions dedicated to remote sensing of snow properties from space. The primary research advance that we expect from our work is a fundamental understanding of the sensitivity of the passive microwave observation (or lack thereof) to the different distributions of sub-pixel phenomena that are characteristic within each footprint of the microwave observation, as a function of scale of the footprint. This analysis is synergistic with several current missions which utilize spaceborne passive microwave sensors, specifically NASA's AMSR-E instrument, and has application in past, present, and future passive microwave datasets.