Hyperbolic diffractions in Ground Penetrating Radar (GPR) data are caused by a variety of subsurface objects such as pipes, stones, or archaeological artifacts. Supplementary to their location, the propagation velocity of electromagnetic waves in the subsurface can be derived. In recent years, it was shown that deep learning tools can automatically detect hyperbola in radargrams using data measured over urban infrastructure, which are relatively clear. In contrast, in this study, we used an archaeological dataset with diverse underground structures. In the first step we used the deep learning network RetinaNet to detect hyperbola automatically and achieved an average precision of 0.58. In the next step, 10 different approaches for hyperbola fitting and thus velocity determination were applied. The derived information was validated with manually determined velocities and apex points. It was shown that hyperbola extraction by using a threshold and a column connection clustering (C3) algorithm followed by simple hyperbola fitting is the best method, which had a mean velocity error of 0.021 m/ns compared to manual determination. The average 1D velocity-depth distribution derived in 10 ns intervals was in shape comparable to the manually determined one, but had a systematic shift of about 0.01 m/ns towards higher velocities.

We performed geophysical and geoarchaeological investigations in the Wadden Sea off North Frisia (Schleswig-Holstein, Germany) to map the remains and to determine the state of preservation of the medieval settlement of Rungholt, especially its southern dyke segment, called theNiedamdyke. Based on archaeological finds and historical maps, Rungholt is assumed to be located in the wadden sea area around the island Hallig Südfall. During medieval and early modern times, extreme storm events caused major land losses, turning cultivated marshland into tidal flats. Especially the 1^stGrote Mandrenke (or St. Marcellus’ flood), an extreme storm surge event in 1362 AD, is addressed as the major event that flooded and destroyed most of the Rungholt cultural landscape. Cultural traces like remains of dykes, drainage ditches, tidal gates, dwelling mounds or even plough marks were randomly surveyed and mapped in the tidal flats by several authors at the beginning of the 20^thcentury. Due to the tidal flat dynamics with frequently shifting tidal creeks and sand bars, the distribution of cultural remains visible at the surface is rapidly changing, making it hard to create a comprehensive map of the cultural landscape by surveying. Today, the Niedam dyke area is fully covered by tidal flat sediments, depriving any remains from further archaeological investigation. Since little is known about the precise location or state of preservation of these remains, our investigation aimed at the rediscovery of the medieval dyke system and associated structure with modern and accurate geophysical, geodetical and geoarchaeological methods. Magnetic gradiometry revealed a large part of the medieval dyke, confirming two tidal gates and several terps connected inland with the dyke, providing a detailed example of a Frisian medieval dyke system. Based on our results, the so far inaccurate and incomplete maps of this part of Rungholt can now be specified and completed. Beyond that, seismic reflection profiles give a first depth resolving insight in the remains of the dyke system, revealing a severe threat to the medieval remains by erosion. The site is exemplary for the entire North Frisian coast, that was influenced by multiple flood events in the middle ages to modern times.

Inad1362, a major storm surge drowned wide areas of cultivated medieval marshland along the north‐western coast of Germany and turned them into tidal flats. This study presents a new methodological approach for the reconstruction of changing coastal landscapes developed from a study site in the Wadden Sea of North Frisia. Initially, we deciphered long‐term as well as event‐related short‐term geomorphological changes, using a geoscientific standard approach of vibracoring, analyses of sedimentary, geochemical and microfaunal palaeoenvironmental parameters and radiocarbon dating. In a next step, Direct Push (DP)‐based Cone Penetration Testing (CPT) and the Hydraulic Profiling Tool (HPT) were applied at vibracore locations to obtainin situhigh‐resolution stratigraphic data. In a last step, multivariate linear discriminant analysis (LDA) was successfully applied to efficiently identify different sedimentary facies (e.g., fossil marsh or tidal flat deposits) from the CPT and HPT test dataset, to map the facies' lateral distribution, also in comparison to reflection seismic measurements and test their potential to interpolate the borehole and CPT/HPT data. The training dataset acquired for the key site from coring and DP sensing finally allows an automated facies classification of CPT/HPT data obtained elsewhere within the study area. The new methodological approach allowed a detailed reconstruction of the local coastal landscape development in the interplay of natural marsh formation, medieval land reclamation and storm surge‐related land losses.

We present a case study of multi-coil frequency-domain electromagnetic (FD-EMI) prospection of a wooden ship wreckage from the 17th century. The wreckage is buried in a sandbar in the German part of the tidal flat area of the North Sea. Furthermore, the wreckage was excavated in advance and covered again after investigation. This ground truthing background and the position of the wreckage makes it a unique investigation object to test the feasibility of FD-EMI for prospecting wooden archaeological objects in the high conductive sediments of tidal flat areas. Our results reveal the shape and position of the wreckage in terms of conductivity maps. The resulting signal change caused by the wreckage in conductivity is only 10% of the value of the water-saturated sandy background, respectively, making a cautious process necessary, including a precise height correction. The data, furthermore, reveals a sensitivity to the vertical shape of the wreckage and thus sufficient depth sensitivity, but with reduced sensing depth. The study highlights the great potential of EMI for both in situ heritage management and archaeological research in the Wadden Sea.

Along the southern North Sea coast from the Netherlands to Denmark, human cultivation efforts have created a unique cultural landscape. Since the Middle Ages, these interactions between humans and natural forces have induced major coastal changes. In North Frisia (Germany), storm floods in 1362 AD and 1634 AD turned wide areas of embanked cultural land into tidal flats. Systematic geoarchaeological investigations between Nordstrand and Hallig Südfall comprise coring, trenching, sedimentary, geochemical and microfaunal palaeoenvironmental parameter analyses and radiocarbon dating. Together with geophysical prospection results and archaeological surveys, they give insights into the landscape’s development and causes for land losses. Results reveal that fens and bogs dominated from c. 800 BC to 1000 AD but are mostly missing in the stratigraphy. Instead, we found 12th to 14th cent. AD settlement remains directly on top of a pre-800 BC fossil marsh. This hiatus of c. 2000 years combined with local ‘Hufen’ settlements implies an extensive removal of peat during cultivation eventually resulting in the use of underlying marshland for agricultural purposes. Fifteenth cent. AD tidal flat deposits on top of the cultivated marsh prove that human impact lowered the ground surface below the mean high water of that time, clearly increasing the coastal vulnerability. We consider these intensive human–environment interactions as a decisive trigger for the massive loss of land and establishment of the tidal flats in North Frisia that are currently part of the UNESCO World Heritage “Wadden Sea”.

Hyperbolae in radargrams are caused by a variety of small subsurface objects. The analysis of their curvature enables the determination of propagation velocity in the subsurface, which is important for exact time-to-depth conversion and migration and also yields information on the water content of the soil. Using deep learning methods and fitting (DLF) algorithms, it is possible to automatically detect and analyze large numbers of hyperbola in 3D Ground-Penetrating Radar (GPR) datasets. As a result, a 3D velocity model can be established. Combining the hyperbola locations and the 3D velocity model with reflection depth sections and timeslices leads to improved archaeological interpretation due to (1) correct time-to-depth conversion through migration with the 3D velocity model, (2) creation of depthslices following the topography, (3) evaluation of the spatial distribution of hyperbolae, and (4) derivation of a 3D water content model of the site. In an exemplary study, we applied DLF to a 3D GPR dataset from the multi-phased (2nd to 12th century CE) archaeological site of Goting on the island of Föhr, Northern Germany. Using RetinaNet, we detected 38,490 hyperbolae in an area of 1.76 ha and created a 3D velocity model. The velocities ranged from approximately 0.12 m/ns at the surface to 0.07 m/ns at approx. 3 m depth in the vertical direction; in the lateral direction, the maximum velocity variation was ±0.048 m/ns. The 2D-migrated radargrams and subsequently created depthslices revealed the remains of a longhouse, which was not known beforehand and had not been visible in the unmigrated timeslices. We found hyperbola apex points aligned along linear strong reflections. They can be interpreted as stones contained in ditch fills. The hyperbola points help to differentiate between ditches and processing artifacts that have a similar appearance as the ditches in time-/depthslices. From the derived 3D water content model, we could identify the thickness of the archaeologically relevant layer across the whole site. The layer contains a lot of humus and has a high water retention capability, leading to a higher water content compared to the underlying glacial moraine sand, which is well-drained.