Implication of an improved field based snow particle size retrieval method for remote sensing of snow

Snow particle size is an important parameter when studying snowpack properties as the size of the snow particles affects the snow density and the snow pack energy balance by changing albedo and backscattering properties. In this study we present an improved method that objectively provides detailed information on the size and size distribution of snow particles. The goal of this study was to first develop an efficient field sampling method that provides a quantitatively accurate snow particle size distribution and secondly to estimate how the various snow particle size distributions affect the reflectivity and backscatter of remotely sensed imagery. Our method is based on a two step approach of (i) the rectification of digital images of snow and (ii) the object oriented segmentation and classification of the photographed snow particles. We have tested the method at different spatial and temporal scales ranging from study sites on the East Antarctic ice sheet, sites with snow on sea ice within the Ross Sea to two field sites in Northern Sweden. In addition, we have compared the method to existing methods in terms of visual analysis of snow. The advantages of our method are that it is observer-independent and that it allows the determination of both the snow particle size and shape distribution from just one snow sample. Application of our method on the East Antarctic ice sheet showed a decreasing snow grain size towards the center of Antarctica and larger grains in the coastal areas. The data demonstrate further a regional snow particle size variability from 0.63 to 0.91 mm between the plateau and the coast. The local variability ranged between 0.23-1.03 mm within 10 by 10 square meter grid at the polar plateau. A comparison of the Antarctic dataset with AMSR-E, MOA, MODIS and MERIS satellite imagery showed a strong correlation with the 89GHz AMSR-E H-pol data at r2 = 0.68-0.73. The results indicate that our method provides a quick but objective field approach to retrieve accurate ground-truth information for remote sensing products across vast spatial areas. Thus, we believe our method will help to narrow the gap between the field and remotely sensed characterization of snow particles and snow physical properties.