Conference Agenda

This work outlines recent results from various activities and projects aimed at monitoring, preserving, and proactively safeguarding cultural heritage (CH) sites against both natural and man-made threats, including extreme events driven by climate change. The primary goal was to develop a multi-risk assessment approach for these areas, starting with a review of existing data. This phase considers the vulnerabilities, exposure levels of the elements at risk, and conservation states of the sites, as well as the interactions between different geo-hazards and cultural heritage.

The following phase involved the use of Earth Observation (EO) data, specifically through multi-temporal InSAR processing, analysis and interpretation. Additionally, field surveys are integrated with other EO and in situ monitoring systems to produce an effective mitigation and conservation plan.

Several case studies are discussed to demonstrate the feasibility of this site-specific approach, which takes into account the unique characteristics of each site, such as its natural and man-made context, geological setting, and the different geohazards it faces. These case studies include the ancient Port of Classe (Emilia Romagna), the Archaeological area of the Phlegrean Fields, the Archaeological Park of Paestum and Velia (Campania), the historical towns of Volterra and Pienza (Tuscany), Orvieto and Civita di Bagnoregio (Lazio), the Villa Romana del Casale in Piazza Armerina (Sicily), and the Historic Quarter of the Seaport City of Valparaíso in Chile.

Depending on the sensor characteristics (e.g. spatial resolution, microwave band, revisit time) SAR images acquired from different constellations such as Cosmo-SkyMed (first and Second Generation) and Sentinel-1 were processed using D-InSAR approaches to assess the spatial and temporal evolution of deformations induced mainly by subsidence and slope instability. Based on ground and site conditions the Permanent Scatterer (PS) and Small Baseline (SB) algorithms were employed to produce maps of measurement points, average displacement rates, and time series for the analysed period. Where available, local GNSS data were used to calibrate EO data with ground truth. The results emphasize the value of a well-tested approach offering to site managers a powerful tool for protecting cultural heritage sites from geo-hazards related damage.

The Fucino area (Abruzzo region), once the third largest lake in Italy, is a territory that has undergone drastic changes and witnessed numerous natural disasters, such as floods and earthquakes, which have shaped the relationship between humans and the natural environment.

The presence of a lake with an unstable regime has led to the emergence of a culture of water management, developed since prehistoric times. This culture entails a strong adaptation to the lacustrine environment, but also a constant aspiration to reverse the natural order by transforming water into land.

While ancient texts are an exceptional source for understanding the history of Fucino, the interpretation of satellite images provided by commercial platforms (e.g. Google Earth, Mapbox, Esri) represents the most effective tools for 'visualizing' the technological endeavors of the Roman era aimed at draining the lake and obtaining cultivable land.

Starting from the 2nd century AD, a complex system of channels has been set up and enhanced over the centuries, leading to a regular pattern in the land division (i.e. cadaster). The (still) humid bottom of the lake has fossilized part of these monumental anthropic operations.

Our research begins with the acquisition of these fossil traces revealed by satellite images. In a further step, all the elements are automatically classified by their orientation in Morphal environment, to identify coherent morphological grids, such as Roman land division models (centuriations) applied to the drained lands.

This work aims at narrating a chapter of the story of this “intangible” cultural heritage whose traces are detectable only through remote sensing techniques. The Fucino area represents a site worth to be better investigated (e.g. through aerial campaign), documented and preserved.

Space-borne remote sensing has increasingly supported archaeological and cultural heritage applications over the past century, and Synthetic Aperture Radar (SAR) imaging has played a key role in advancing this application field. In this work, SAR data capabilities for archaeological prospection and cultural heritage site monitoring are investigated and demonstrated through two case studies from the wider Province of Rome (Italy). Medium to very high resolution SAR scenes acquired by the RADARSAT-2 and Copernicus Sentinel-1 missions using the C-band, ALOS-1 using the L-band and COSMO-SkyMed using the X-band are exploited to trial the detection of crop marks at (semi-)buried and sub-surface archaeological features in the archaeological landscape of Ostia-Portus. The analysis highlights a progressive increase in the detectability of smaller archaeological features when moving from L to X band, from co- to cross-polarizations, and toward finer spatial resolutions. Big data stacks of Sentinel-1 imagery are also processed with the parallelized Small BAseline Subset (SBAS) multi-temporal Interferometric SAR (InSAR) method to extract ground displacement time series and monitor the stability of cultural heritage assets within the UNESCO World Heritage Site (WHS) of Rome. Several hotspots showing significant land deformation (mainly indicating the occurrence of subsidence) can be identified, such as in the area of Fiumicino International Airport and along the Tiber River alluvium, involving monuments and heritage assets. Land subsidence patterns distributed along the Tiber River course are spatially confined due to subsurface geology and interactions with urban development. In addition to the well-known subsidence patterns within the UNESCO WHS, at Basilica Saint Paul Outside the Walls and Grotta Perfetta valley, Sentinel-1 InSAR results highlight a clear subsidence pattern with rates up to -0.9 cm/year affecting the residential quarter of Valco San Paolo. This work was carried out in the framework of ESA and Chinese MOST Dragon-5 SARchaeology project.

Heritage, both cultural and natural, is a complex ecosystem made up of heterogeneous and interrelated components. Thus, its preservation requires a multidisciplinary approach, providing a foundation of knowledge for the development of safeguarding strategies and policies.

In this sense, the growing availability, accuracy, and temporal continuity of remote sensing and Earth observation data represent essential tools for heritage monitoring and risk assessment.

Starting from two pilot cases, the Island of Mozia (Sicily) and the Valleys of Tortona (Piedmont), methodologies based on the integration of multi-sensor satellite imagery from the Copernicus program, Sentinel-1, Sentinel-2 and Sentinel-3, with aerial imagery at different spatial and temporal resolutions have been developed. Data processing based on advanced geospatial analytics, including change detection and time-series analysis, enables to monitor and quantify environmental and anthropogenic risks for the areas and to derive technical requirements on the scalability and replicability of the methodology.

Both pilot sites present significant heritage elements: on the Island of Mozia, located in a protected marine area, an archaeological site coexists with the wine cultivation, while Tortona is a 25,000 inhabitants Roman age settlement at the outskirts of the UNESCO Wine Cultura Landscape of site Langhe, Roero and Monferrato.

By applying big data analytics, we implement a scalable and automated workflow capable of processing large volumes of satellite data in near real-time. A detailed assessment of land-use changes, vegetation stress, and surface deformation integrating Copernicus data and products with high-resolution aerial datasets. The risk assessment, through a multi-criteria analysis framework, incorporates machine learning techniques to simulate “what-if” scenarios using Copernicus climate datasets.

The modularity and scalability of the approach, implementing automated data processing pipelines in a dedicated SW platform, enables its adaptation to various cultural and natural heritage contexts, offering a replicable, cost-effective solution for near real-time monitoring and early warning systems, accessible by the user, supporting also preservation efforts.

The PERSEO project (Prisma hyperspectral image Enhancement for Revealing cultural heritage Sites from Earth Observation) aims to ascertain the suitability of PRISMA hyperspectral data for applications in the Cultural Heritage domain. In collaboration with the Italian Space Agency, the project leverages the PRISMA satellite and advanced pansharpening techniques to address the limitations of lower-resolution satellite imagery, allowing for the detection of small-scale archaeological features.

Key case studies include Aquileia (Italy), known for its archaeological and geomorphological richness. In this area high-resolution LiDAR Digital Terrain Models, high-resolution RGB data and multitemporal satellite images were interpreted to create a georeferenced dataset of over 400 geoarchaeological features. This dataset is used to compare the performance of machine learning models in automatically detecting sub-soil traces using Sentinel-2 data against the usage of PRISMA data. The project has evaluated quantitatively and qualitatively some state-of-the-art pansharpening techniques (GSA, MTF-GLP and HySure) in enhancing PRISMA’s 30m hyperspectral data’s spatial resolution to a 5m resolution. A PyTorch reimplementation of HySure has accelerated processing times by 6x, significantly improving the detection of subsoil features while preserving the spatial and spectral quality of the original implementation.

The development of multitemporal analysis techniques is another aspect of the project deemed critical by the nature of the crop and soil marks interpreted when evaluating the presence of a sub-soil archaeological object. By analysing time series of images taken across different seasons, the project addresses the challenges of seasonal vegetation and soil variations, improving the visibility of archaeological traces throughout the years.

As PERSEO progresses, these enhanced detection methods and automated workflows will streamline the use of hyperspectral satellite imagery in archaeological prospection, contributing significantly to cultural heritage preservation. Future work includes extending these methods to additional cases, improving the efficiency of detection with PRISMA imagery, and further refining machine learning applications.

question time