Seasonal landslide mapping and estimation of landslide mobilization rates using aerial and satellite images

We tested the possibility of using digital, color aerial ortho-photographs and monoscopic, panchromatic satellite images of comparable spatial and radiometric resolution, to map recent landslides in Italy and to update existing measures of landslide mobilization. In a 90-km(2) area in Umbria, central Apennines, rainfall resulted in abundant landslides in the period from September 2004 to June 2005. Analysis of the rainfall record determined the approximate dates of landslide occurrence and revealed that the slope failures occurred in response to moderately wet rainfall periods. The slope failures occurred primarily in cultivated terrain and left subtle morphological and land cover signatures, making the recognition and mapping of the individual landslides problematic. Despite the difficulty with the identification of the landslides without the use of stereoscopic visualization, visual analysis of the aerial and satellite images allowed mapping 457 new landslides, ranging in area 3.0 x 10(1) < A(L) < 2.5 x 10(4) m(2), for a total landslide area A(LT) = 6.92 x 10(5) m(2). To identify the landslides, the investigators adopted the interpretation criteria commonly used to identify and map landslides on aerial photography. The result confirms that monoscopic, very high resolution images taken by airborne and satellite sensors can be used to prepare landslide maps even where slope failures are difficult to detect, provided the imagery has sufficient geometric and radiometric resolutions. The different dates of the aerial (March 2005) and the satellite (June-July 2005) images allowed the temporal segmentation of the landslide information, and studying the statistics of landslide area and volume for different periods. Compared to pre-existing information on the abundance and size of the landslides in the area, the inventory obtained by studying the aerial and satellite images proved more complete. The new mapping showed 145% more landslides and 85% more landslide area than a pre-existing reconnaissance inventory. As a result of the improved mapping, the rate of landslide mobilization for the 2004-2005 landslide season was determined to be phi(L) = 27.1 mm year(-1), 30% higher than a previous estimate for the same period. This seasonal rate of landslide mobilization is significantly larger than the long-term regional erosion rate in the central Apennines. The accelerated rate is attributed to agricultural practices that favor slope instability. (C) 2011 Elsevier B.V. All rights reserved.