Bibliographic Metadata

Title
Monitoring of Alpine Snow Conditions Using C-Band SAR / von Claudio Navacchi
AuthorNavacchi, Claudio
Thesis advisorWagner, Wolfgang
PublishedWien, 2018
Description94 Seiten
Institutional NoteTechnische Universität Wien, Diplomarbeit, 2018
Annotation
Arbeit an der Bibliothek noch nicht eingelangt - Daten nicht geprueft
LanguageEnglish
Document typeThesis (Diplom)
Keywords (EN)Remote sensing
URNurn:nbn:at:at-ubtuw:1-119859 Persistent Identifier (URN)
Restriction-Information
 The work is publicly available
Files
Monitoring of Alpine Snow Conditions Using C-Band SAR [15.6 mb]
Links
Reference
Classification
Abstract (English)

The cryosphere is an essential part of the Earth's climate system, even more in the light of climate change actively impacting the extent of polar ice caps, glaciers and snow. Regions in interaction with these parts of the cryosphere must adapt to challenging conditions, as slight changes in temperature can have irreversible consequences. Informations about the state of a snow pack and ongoing processes within can be a valuable aid, e.g. for avalanche risk management, hydrological run-off models and tourism development. This thesis focuses on revealing connections between various snow parameters, e.g. grain size, snow height or snow wetness, and high-resolution C-band SAR backscatter from ESA's Sentinel missions. Water has a very significant effect in the C-band for different states of aggregation (e.g., solid, liquid) offering a profound physical basis for investigating these relationships in alpine areas with a vast variety of snow conditions. For this case study, an alpine region covering parts of North and South Tyrol, was chosen. Data was acquired for a timespan of over two years, from summer 2015 to autumn 2017. Well-known states of C-band backscatter like sigma naught, which can be related to backscatter from a unit area on ground, hinder a comparison with in-situ snow data due to the inuence of different observation geometries. To overcome this, alternative representations of backscatter, like normalised backscatter, either being normalised by incidence angle or by performing a radiometric (terrain attening) normalisation, are presented in this work. In the former case, linear regression and a novel approach, the piecewise linear percentile slope method, which takes the backscatter distribution of each orbit into account, were used. C-band backscatter was not only analysed as a single band, but also by including cross-polarisation ratios and change detection benefiting from a new method for an automatic, pixel-based reference image selection. Overall, normalised backscatter by means of linear regression and VH polarisation appeared as the best setup, when correlating these data with in-situ snow measurements. Results were enhanced by spatial and temporal Filtering of backscatter data leading to a partial increase in correlation by nearly 0.2. The most meaningful and consistent correlation of -0.64 was found with respect to maximum snow wetness, followed by air temperature (-0.59). Snow height was characterised by the highest correlation (0.67), but its significance is questionable. Concerning snow wetness, change detection performed best, when taking pixels at coldest conditions as reference values into account. Derived maps indicating wet and dry snow could offer useful information for run-off models and for determining fragile snow packs.

Stats
The PDF-Document has been downloaded 9 times.