Validation datasets for Dynamical segmentation (DyNA) and RGB-HSV based approaches for avalanches using Sentinel-1 SAR images

The dataset contains multiple sets of validation files for the extraction of avalanches using the dynamical segmentation and RGB-based colour detection approach on Sentinel-1 images. The study is performed over a 180 sq km area around Davos, Switzerland, for two years: 2018 and 2019. This is the updated version based on new results generated after better optimisation of the DyNA parameters. The dataset can be described as: SPOT_2018/ 2019 : Avalanche shapefile polygons originally coming from Hafner et al. 2020 and downloaded from: www.doi.org/10.16904/envidat.202. The original dataset was edited to remove some avalanche polygons and used to compare the results of the Sentinel-1 SAR dynamic segmentation mapping. S1_2018/ 2019_ref : Avalanche polygons were also downloaded from www.doi.org/10.16904/envidat.202. These polygons serve as a reference for comparing the results of dynamical segmentation with previous Sentinel-1 SAR mapping results. DyNA_avalanche_mask_ASC/DES/ASC_DES_2018/2019 : These are .tif files containing binary masks for avalanches generated by the dynamical segmentation approach. Rasters with the prefix ASC stand for avalanche mask generated solely by using the ascending orbit data, DES represents results from only descending orbit, and ASC_DES stands for a union mask combining results from both ascending and descending orbits. For comparison of our results, the ASC_DES avalanche mask was considered. RGB_avalanche_mask ASC/DES/ASC_DES_2018/2019 : These are .tif files containing binary masks for avalanches generated by the RGB-HSV-based approach. Rasters with the prefix ASC stand for avalanche mask generated solely by using the ascending orbit data, DES represents results from only descending orbit, and ASC_DES stands for a union mask combining results from both ascending and descending orbits. For comparison of our results, the ASC_DES avalanche mask was considered.