CLASSIFICATION OF DIFFERENT VARIETIES OF SNOW BASED ON SPECTRAL REFLECTANCE USING ASD SPECTRORADIOMETER FROM THE HIMALAYAS

摘要

The monitoring of snow cover in the Indian Himalaya is important for climate research and for operational activities i.e. hydrology and weather forecasting. For this reliable and accessible snow data are required with spatial variability and long-term trends. Hyperspectral imagery requires ground truth data for better image interpretation and analysis. Spectroscopy is used to quantify snow optical properties, based on reflectance measurements in the visible (VIS), near-infrared (MR) and short wave infrared (SWTR) regions for the classification of different varieties of snow. Conventional crystal gauge with a magnifying glass was used in the field for observing snow properties. The spectral signature at Dhundi - Solang (snowbound) and Patio (glaciated), Himachal Pradesh, India during the successive years of 2017 and 2018. A total of 663 spectral signatures from different locations of the study areas were collected by measuring and analyzing their reflectance curves using Analytical Spectral Device (ASD) FieldSpec 4 in the range of 350-2500 nm. These spectra provide an integrative technique that measures the fundamental characteristics and composition of the snow and grain size. Clean, fine-grained snow is scattering-dominated and reflects the majority (97-99%) of incident sunlight in ultra-violate to visible wavelengths (350-750 nm), and is more absorbing in NIR/SWIR (750-2500 nm). As the snow ages its grain size increases and the reflectance decreases 350 -2500 nm. however, maximum changes are recorded in the visible & NIR region. Reflectance differs for snow and ice depending on the actual composition of the material including impurities and grain size. For glacier bound snow area the debris being dominant contaminant plays an important role in the reflectance of different wavelength regions. A drastic change in reflectance for fresh and moderately dirty snow is noted in the visible portion. In general, snow has higher spectral contrast in visible and SWIR region, however, reflection in SWIR related to its microscopic liquid water content. Variation in reflectance for moderate to highly dirty snow is attributed to vehicular emission and construction upstream. Thus, the study proved its importance in improving the accuracy of snow and ice parameter retrieval from field-based spectroradiometer in turn useful climatic studies.