Unsupervised Snow Cover Classification Using Dual-Polarized SCATSAT-1 Satellite Data

The Ku-band (13.5 GHz)-based Scatterometer Satellite (SCATSAT-1) launched on September 26, 2016, by the Indian Space Research Organization (ISRO) offers the enhanced spatial resolution (~2.2 km) data products in form of Sigma-naught (σ0) and Gamma-naught (γ0). Both forms are available in two polarizations modes: Horizontal transmission and horizontal reception (HH) and vertical transmission and vertical reception (VV) that impact the strength of backscatter. In the present work, we have analyzed the performance of the dual-polarized (HH and VV)-backscattered coefficient over the snow cover region. To test the efficacy, we have implemented the unsupervised classification on both backscattered coefficients (σ0 and γ0). The k-mean as an unsupervised classifier takes advantage of the independence of user input. The study area is a part of Western Himalayas and acquired using the SCATSAT-1 level-4 (L4) product. Results have been validated using Moderate Resolution Imaging Spectroradiometer (MODIS) MOD10 product and shown the effectiveness of k-means clustering on SCATSAT-1 L4 both backscattered coefficients (σ0 and γ0).