The use of Small Baseline (SBAS) interferometric stacking for high precision calculation of displacement using Synthetic Aperture Radar (SAR) data has been proven a reliable tool in a wide range of applications. The Arctic, with its high seasonal variability in scattering behaviour and low density of persistent scatterers presents an interesting but complicated environment for this type of analysis. In an effort to improve the understanding of SBAS analysis performance in the Arctic, this study aimed to determine the impact of snow cover on the SBAS method applied in Arctic environments using Sentinel-1 C-band SAR.

The study areas chosen for this project was the Stordalen permafrost area as well as several other locales (urban and sub-urban) in and around Abisko, Sweden. The data used was Sentinel-1A/B Single Look Complex (SLC) Interferometric Wide Swath (IW) co-polarized captures on a 6-day revisit time from mid-June 2017 to mid-June 2018.

To evaluate method behaviour, specifically the effects of emerging and receding snow cover, a moving stack was used. The moving stack experiment was performed by keeping the stack length at a predetermined optimal of 22 images while iteratively removing the two earliest acquisitions and adding two to the end of the stack thus creating a comparable but displaced temporal coverage.

Earlier experiments have shown that when snow cover is introduced at the extremes of a stack, displacement calculations become erratic for the images in question, but as more images with snow cover are introduced, this erratic behaviour fades showing a method sensitivity to poorly-distributed bi-modal image stacks. Running the moving stack experiment over an entire snow season yields insight on the direct impacts of emerging, full and receding snow cover, allowing for a better understanding of method reliability in arctic environments. This furthermore grants the opportunity to better approximate the noise-fraction in the data, thus increasing precision of snow property retrieval.

In conclusion, full-season SBAS moving stack analysis clearly highlights variations in method behaviour connected to the seasonal variability of the Arctic environment, specifically with the introduction of snow cover. Transition periods have proven to cause erratic behaviour in displacement calculation results, with increased reliability allowing for further analysis during periods with mono-modal distributions.

Area D - Board 488

Remote sensing of avalanche debris in mountain areas offers new opportunities to improve our understanding of avalanche activity and to evaluate the physical models of avalanche hazard forecasts. The location of avalanche debris and the estimation of their sizes are of great interest for studies dealing with the stability of the snowpack and also for studying the variability of natural avalanche activity. In addition, time series of avalanche events, with relevant time and space resolutions, would be highly relevant to better identify avalanche risk zones and periods. Such time series would complement some other existing database mostly based on visual observations.
Sentinel-1 satellites offer a unique tool to monitor some properties of the snowpack using a C-band Synthetic Aperture Radar (SAR) with a high spatial resolution (20m) and with a revisit frequency of 6 days over the French mountain massifs. We use a change detection algorithm to isolate avalanche debris-like features based on the backscatter contrast between avalanche debris and the surrounding undisturbed snowpack. The debris detection is based on major changes in the backscatter coefficients due to changes in snow properties following the avalanche event (height, density, roughness, ...), with the medium around the avalanche remaining almost unchanged. Based on avalanche debris signatures in SAR images, new avalanche activity indicators are then derived from Sentinel-1 data over Alpine regions and are evaluated against in-situ measurements. An overview of the method will be given as well as avalanche activity indicator trends over the French Alps during the 2016/2019 seasons. Some statistics about the method performances will also be presented.

Area D - Board 489

Seasonal snow cover is an important component in the Earth’s climate system. The detailed information about the snow cover and its characteristics with high spatial and temporal scales are essential for snow-melt runoff modeling, snow avalanche risk assessment, climatology, hydro-power industry. Synthetic Aperture Radar (SAR) data with high spatial and temporal resolution along with its polarimetric capability provide us an opportunity for the comprehensive exploration of the snow cover properties. Complete utilization of full polarimetric information from SAR data with extensive field campaigns could help to better understanding of the snow cover properties and development of new methods for the estimation of snow parameters.
In this study, multi-temporal RADARSAT-2 Polarimetric Synthetic Aperture Radar (PolSAR) data is used to investigate the snow properties over the Alps. Spitzingsee, Bayern, Germany and Davos, Switzerland are the two study sites selected for the field campaign. Totally 11 Radarsat-2 scenes were acquired in the same orbit with the same incidence angle during Dec 2017 - Mar 2018. The Automatic Weather Station (AWS) data and near real-time in-situ snow measurements with the SAR data are collected. Snow probe measured snow depth and the limited snow stratigraphy profile measurements are used for the investigation.
Various PolSAR decomposition parameters, backscattering coefficients, coherency matrix elements, generalized snow surface and volume parameters [1] are derived from Radarsat-2 data. These parameters are exhaustively investigated with the AWS and in-situ snow depth measurements along with the temperature and precipitation information. It is interesting to see that, the depth of the snowpack above the first thick ice crust layer is having a high correlation with the certain parameters. Particularly, the dominant scattering type phase (Φs1) from Touzi incoherent polarimetric decomposition method [2] and (HH+VV) (HH-VV) Phase from coherency matrix is having a high correlation with the snow depth above the ice crust layer. It seems even though the snow surface is dry, the presence of a single ice layer in the snowpack acts like an underlying ground surface for C-band microwave frequency. These parameters are further investigated with different snow conditions for the estimation of snow water equivalence.



[1]. M. Surendar, A. Bhattacharya, G. Singh, Y. Yamaguchi, and G. Venkataraman, “Development of a snow wetness inversion algorithm using polarimetric scattering power decomposition model,” Int. J. Appl. Earth Observ. Geoinf., vol. 42, pp. 65–75, 2015.

[2]. R. Touzi, "Target Scattering Decomposition in Terms of Roll-Invariant Target Parameters," in IEEE Transactions on Geoscience and Remote Sensing, vol. 45, no. 1, pp. 73-84, Jan. 2007

Area D - Board 490