干旱区生态环境知识资源中心

The aim of this study is to assess the sensitivity to buried archaeological structures of C- and L-band Synthetic Aperture Radar (SAR) in various polarisations. In particular, single and dual polarised data from the Phased Array type L-band SAR (PALSAR) sensor on-board the Advanced Land Observing Satellite (ALOS) is used, together with quadruple polarised (quad pal) data from the SAR sensor on Radarsat-2. The study region includes an isolated area of open fields in the eastern outskirts of Rome where buried structures are documented to exist. Processing of the SAR data involved multitemporal averaging, analysis of target decompositions, study of the polarimetric signatures over areas of suspected buried structures and changes of the polarimetric bases in an attempt to enhance their visibility. Various ancillary datasets were obtained for the analysis, including geological and lithological charts, meteorological data, Digital Elevation Models (DEMs), optical imagery and an archaeological chart.

For the Radarsat-2 data analysis, results show that the technique of identifying the polarimetric bases that yield greatest backscatter over anomaly features, and subsequently changing the polarimetric bases of the time series, succeeded in highlighting features of interest in the study area. It appeared possible that some of the features could correspond with structures documented on the reference archaeological chart, but there was not a clear match between the chart and the results of the Radarsat-2 analysis. A similar conclusion was reached for the PALSAR data analysis. For the PALSAR data, the volcanic nature of the soil may have hindered the visibility of traces of buried features. Given the limitations of the accuracy of the archaeological chart and the spatial resolution of both the SAR datasets, further validation would be required to draw any precise conclusions on the sensitivity of the SAR data to buried structures. Such a validation could include geophysical prospection or excavation. (C) 2013 Elsevier B.V. All rights reserved.