Despite the recognized effectiveness of LiDAR in penetrating forest canopies, its capability for archaeological prospection can be strongly limited in areas covered by dense vegetation for the detection of subtle remains scattered over morphologically complex areas. In these cases, an important contribution to improve the identification of topographic variations of archaeological interest is provided by LiDAR-derived models (LDMs) based on relief visualization techniques. In this paper, diverse LDMs were applied to the medieval site of Torre Cisterna to the north of Melfi (Southern Italy), selected for this study because it is located on a hilly area with complex topography and thick vegetation cover. These conditions are common in several places of the Apennines in Southern Italy and prevented investigations during the 20th century. Diverse LDMs were used to obtain maximum information and to compare the performance of both subjective (through visual inspections) and objective (through their automatic classification) methods. To improve the discrimination/extraction capability of archaeological micro-relief, noise filtering was applied to Digital Terrain Model (DTM) before obtaining the LDMs. The automatic procedure allowed us to extract the most significant and typical features of a fortified settlement, such as the city walls and a tower castle. Other small, subtle features attributable to possible buried buildings of a habitation area have been identified by visual inspection of LDMs. Field surveys and in-situ inspections were carried out to verify the archaeological points of interest, microtopographical features, and landforms observed from the DTM-derived models, most of them automatically extracted. As a whole, the investigations allowed (i) the rediscovery of a fortified settlement from the 11th century and (ii) the detection of an unknown urban area abandoned in the Middle Ages.

Abstract:The use of GIS and Spatial Analysis for predictive models is an important topic in preventive archaeology. Both of these tools play an important role in the Support Decision System (SDS) for archaeological research and for providing information useful to reduce archaeological risk. Over the years, a number of predictive models in the GIS environment have been developed and proposed. The existing models substantially differ from each other in methodological approaches and parameters used for performing the analysis. Until now, only few works consider spatial autocorrelation, which can provide more effective results. This paper provides a brief review of the existing predictive models, and then proposes a new methodological approach, applied to the neolithic sites in the Apulian Tavoliere (Southern Italy), that combines traditional techniques with methods that allow us to include spatial autocorrelation analysis to take into account the spatial relationships among the diverse sites.

No abstract

This paper deals with a UAV LiDAR methodological approach for the identification and extraction of archaeological features under canopy in hilly Mediterranean environments, characterized by complex topography and strong erosion. The presence of trees and undergrowth makes the reconnaissance of archaeological features and remains very difficult, while the erosion, increased by slope, tends to adversely affect the microtopographical features of potential archaeological interest, thus making them hardly identifiable. For the purpose of our investigations, a UAV LiDAR survey has been carried out at Perticara (located in Basilicata southern Italy), an abandoned medieval village located in a geologically fragile area, characterized by complex topography, strong erosion, and a dense forest cover. All of these characteristics pose serious challenge issues and make this site particularly significant and attractive for the setting and testing of an optimal LiDAR-based approach to analyze hilly forested regions searching for subtle archaeological features. The LiDAR based investigations were based on three steps: (i) field data acquisition and data pre-processing, (ii) data post-processing, and (iii) semi-automatic feature extraction method based on machine learning and local statistics. The results obtained from the LiDAR based analyses (successfully confirmed by the field survey) made it possible to identify the lost medieval village that represents an emblematic case of settlement abandoned during the crisis of the late Middle Ages that affected most regions in southern Italy.