Bi-temporal change analysis of satellite imagery to detect landslides triggered by intense rainfall events

Leonardo Disperati (a,b), Filippo Gregori (a), Massimo Perna (c), Francesco Manetti (d), Guido Lavorini (e) & Carlo Villoresi (e)
(a) Università degli Studi di Siena, Dipartimento di Scienze Fisiche, della Terra e dell'Ambiente, Strada Laterina, 8, 53100, Siena, Italy. E-mail: disperati@unisi.it (b) CNR, Istituto di Geoscienze e Georisorse, Via G. Moruzzi, 1, 56124, Pisa, Italy. (c) CNR Ibimet/Consorzio LaMMA, Via Madonna del Piano, 10, 50019, Sesto Fiorentino (Fi), Italy. (d) Consorzio LaMMA, Via Madonna del Piano, 10, 50019, Sesto Fiorentino (Fi), Italy. (e) Regione Toscana – SITA - P.O. Geologia, Via di Novoli, 26, 50127 Firenze, Italy.

This paper presents the results of implementation of bi-temporal change analysis methods to RapidEye satellite imagery to support the detection, at regional scale, of landslides caused by two intense rainfall events of 2009 and 2011, in Northern Tuscany. Image geometric and radiometric pre-processing were applied. Then, bi-temporal image ratioing of Difference Vegetation Index (DVI) and bi-temporal spectral transformations (Principal Component Analysis - PCA; Independent Component Analysis - ICA) were implemented. Finally, unsupervised and supervised classification allowed us to obtain thematic representation of areas of changes which supported the identification of almost hundred landslides in the study area.

intense rainfall event, shallow landslides detection, satellite images, bi-temporal change analysis, classification.