Item: Classification of snow anisotropic surfaces for mountainous terrain with mixed vegetative cover using optical and thermal spectra from Landsat-8.
-
-
Title: Classification of snow anisotropic surfaces for mountainous terrain with mixed vegetative cover using optical and thermal spectra from Landsat-8.
Proceedings: International Snow Science Workshop Proceedings 2018, Innsbruck, Austria
Authors:
- Santiago Rodriguez [ CryoGARS group, Geosciences Department - Boise State University, Boise, ID, USA ]
- Pedro Rodriguez [ CryoGARS group, Geosciences Department - Boise State University, Boise, ID, USA ]
Date: 2018-10-07
Abstract: New remote sensing platforms under development will improve availability of optical and thermal spectral data for mountainous regions on a more regular basis, possibly on a daily basis at targeted areas. Newly available spectral remote sensing data will allow to characterize the evolution of the snow surfaces, particularly during the winter months when little is known with respect to large spatial scale development of snow surfaces. Understanding of snow surface development during winter has been hampered due to limited remote sensing data due to a combination of inclement weather, bimonthly satellite passes, and the difficulty of producing training and validation data for the snow surfaces. Landsat-8 optical and thermal spectral data from January 16-2017 for Central Idaho was used to demonstrate that snow surface with significant anisotropic development can be classified. During ski touring label data was generated to identifying snow surfaces. This GPS referenced data was used to generate the training and validation data sets for the same Landsat pass period. Spectra Optical and thermal spectra were processed using machine learning classification techniques. This study suggest that winter snowpack surfaces can be tracked with optical and thermal spectral sensors. Tracking of the snow surface temporal development is not only important for the forecasting of avalanches but it also valuable in complementing snowpack development models, as well as identifying snow layered structures that will impact snow melt dynamics.
Language of Article: English
Presenters:
Keywords: snow anisotropy, neural networks classification, support vector machines classification.
Page Number(s): 364-368
-
Digital Abstract Not Available
-