Abstract

A strong windthrow affected many forests in Slovenia and northern Italy on the night of the 29th and the 30th of October, 2018. In this paper, we propose a new methodology for windthrow identification, comprising two new indexes, PCA-CI for evaluating change detection and PCA-WI for the identification of windthrows by analysing the time series of C-SAR Sentinel-1 data. The proposed methodology, unlike the previous studies on the identification of windthrows, does not subtract the values of dual and single polarised VH-VV data combinations, but proposes a new approach with a multivariate transformation of dimensionality reduction of the time series of the dual and single polarised data in two different time frames.

Key words: windthrow, time series, SAR, C-SAR, radar, Sentinel-1, remote sensing