Fusion of Satellite Multispectral Images Based on Ground-Penetrating Radar (GPR) Data for the Investigation of Buried Concealed Archaeological Remains

Abstract

Abstract:The paper investigates the superficial layers of an archaeological landscape based on the integration of various remote sensing techniques. It is well known in the literature that shallow depths may be rich in archeological remains, which generate different signal responses depending on the applied technique. In this study three main technologies are examined, namely ground-penetrating radar (GPR), ground spectroscopy, and multispectral satellite imagery. The study aims to propose a methodology to enhance optical remote sensing satellite images, intended for archaeological research, based on the integration of ground based and satellite datasets. For this task, a regression model between the ground spectroradiometer and GPR is established which is then projected to a high resolution sub-meter optical image. The overall methodology consists of nine steps. Beyond the acquirement of the in-situ measurements and their calibration (Steps 1-3), various regression models are examined for more than 70 different vegetation indices (Steps 4-5). The specific data analysis indicated that the red-edge position (REP) hyperspectral index was the most appropriate for developing a local fusion model between ground spectroscopy data and GPR datasets (Step 6), providing comparable results with the in situ GPR measurements (Step 7). Other vegetation indices, such as the normalized difference vegetation index (NDVI), have also been examined, providing significant correlation between the two datasets (R = 0.50). The model is then projected to a high-resolution image over the area of interest (Step 8). The proposed methodology was evaluated with a series of field data collected from the Vésztő-Mágor Tell in the eastern part of Hungary. The results were compared with in situ magnetic gradiometry measurements, indicating common interpretation results. The results were also compatible with the preliminary archaeological investigations of the area (Step 9). The overall outcomes document that fusion models between various types of remote sensing datasets frequently used to support archaeological research can further expand the current capabilities and applications for the detection of buried archaeological remains.

Recent Trends and Long-standing Problems in Archaeological Remote Sensing

“…While we share this ambition, we can cite only limited examples of attempts to undertake multi-sensor fusion that go beyond layering in GIS, and note that the authors of these studies report limited success to date e.g. Agapiou et al (2017), where the authors state that the "proposed regression approach provided acceptable but not completely satisfactory results" (p. 16). Multi-instrument integration where the data types are quite different remains a challenge, and should be an active area for future research.…”

A European-Scale Investigation of Soil Erosion Threat to Subsurface Archaeological Remains

“…Over the last years, a plethora of scientific efforts attempted a better detection and characterisation of subsurface archaeological remains through a variety of remote-sensing techniques. The latest include satellite imageries [23,24], aerial observations [25][26][27], unmanned aerial vehicles (UAVs) [28], and ground-based geophysical prospections [29,30]. In addition, predictive modelling was used through Geographical Information System (GIS) to identify areas with potential archaeological evidence [31].…”

Remote Sensing and Geosciences for Archaeology

“…Some of the papers also explore the benefits of combining different approaches, e.g., optical remote sensing, aerial imagery and geophysical prospection [27,28], 3D surveys, and reconstruction in GIS environment [29]. In other cases, the focus is the demonstration of automatic detection tools and workflows [14,15].…”

“…In addition, Agapiou et al [27] focus on integration. In particular, in the demonstration site of Vésztő-Mágor Tell in the eastern part of Hungary, the authors run a workflow of data integration and fusion consisting of nine steps, including data capture, regression models to examine more than 70 different vegetation indices, cross-check and validation with GPR, and in situ magnetic gradiometry measurements.…”