Thresholds and distances to better detect wet snow over mountains with Sentinel-1 image time series

Fatima Karbou (1) , Guillaume James (2) , Philippe Durand (3) , Abdourrahmane Atto (4)

1 CEN - Centre d'Etudes de la Neige
2 TRIPOP - Modélisation, simulation et commande des systèmes dynamiques non lisses
3 CNES - Centre National d'Études Spatiales [Toulouse]
4 LISTIC - Laboratoire d'Informatique, Systèmes, Traitement de l'Information et de la Connaissance

In mountain regions, seasonal snow monitoring is very important for many applications such as hydrology, mountain ecosystems, meteorology and avalanche forecasting. This chapter discusses wet snow extent estimation using T. Nagler's backscatter thresholding method. It examines some other interesting options to improve wet snow detection by exploring the use of similarity metrics. Several distances are calculated and tested on a database of Sentinel-1 Synthetic Aperture Radar (SAR) image time series of different sizes over the French Alps. The resulting metrics are discussed and compared with the reference method, as well as with independent data. Change detection is widely used for wet snow detection from SAR images. The different metrics studied show variations that can be linked to significant changes in the SAR signal which is promising for implementing inversion methods integrating the time series of images as a whole.

Domaines: Interfaces continentales, environnement; Mathématiques générales [math.GM]

Fatima Karbou, Guillaume James, Philippe Durand, Abdourrahmane Atto. Thresholds and distances to better detect wet snow over mountains with Sentinel-1 image time series. Change Detection and Image Time Series Analysis, 1: Unsupervised Methods, Wiley, pp.1-18, 2021, ⟨10.1002/9781119882268.ch5⟩. ⟨hal-03483209⟩