Ground penetrating radar survey of the Coqueiro underwater archaeological stilt village, eastern Amazon, Brazil

This paper presents the results of a Ground Penetrating Radar (GPR) survey conducted at the Coqueiro underwater archaeological stilt-house settlement in the Baixada Maranhense, eastern Amazon, Brazil. The region contains pre-colonial settlements built on wooden pillars between approximately AD 100 and 1000. The main objective of this study was to identify and map GPR anomalies compatible with wooden pillars preserved within the water column, at the lakebed, and buried by sediments below the lakebed, thereby defining priority areas for future archaeological investigations. A total of 58 GPR profiles were acquired using a 270 MHz shielded antenna installed in a wooden canoe, with survey trajectories recorded by GNSS. The complete GPR profiles were displayed in two-way travel time, whereas selected intervals containing the most prominent anomalies were converted to depth using a fixed two-layer velocity model, with the interface positioned at 0.8 m depth, a velocity of 0.033 m/ns assigned to the freshwater column, and 0.050 m/ns assigned to the underlying water-saturated sediments beneath the lakebed. The radargrams revealed a strong and relatively continuous reflector associated with the lakebed, together with localized diffraction hyperbolas concentrated in specific sectors of the surveyed area. Profiles T24 and T50 contained the clearest responses, including anomalies located within the water column and below the lakebed reflector. Two-dimensional GPR numerical modeling using the Finite-Difference Time-Domain method reproduced the main geometrical characteristics and relative positions of the observed diffraction patterns, supporting their physical compatibility with submerged wooden pillars. A bathymetric map generated from the GPR-derived lakebed reflector indicated water depths ranging from approximately 0.2 to 1.2 m. The selected diffraction hyperbolas occurred in sectors with water depths near 0.8 m, although their spatial distribution may reflect differences in preservation, sediment cover, GPR detectability, and survey coverage. Ceramic fragments collected near profile T50 further reinforce the archaeological potential of this sector. Overall, the results demonstrate the usefulness of GPR for the initial geophysical characterization of extensive and poorly documented archaeological sites in shallow Amazonian freshwater environments.