1A well-defined area that was once part of the territory of the Roman colony of Narona, bordered on the north-east and south-west by karst highlands and hills, was surveyed with the goal of evaluating the state of preservation of the ancient settlement landscape. The area lies in the Teskera and Hardomilje microregion within the Trebižat river valley (Ljubuški municipality, Bosnia and Herzegovina). A regular survey of the Trebižat river valley had been made in the late 19th/early 20th century (Patsch, 1907). Some sites were excavated during the Yugoslav period, but the breakup of Yugoslavia stopped virtually all archaeological research in the region. Roman settlement in this territory has been studied since 2016 by Tomasz Dziurdzik, Principal Investigator in a Preludium 10 grant (2015/19/N/HS3/00886) of the National Science Centre of the Republic of Poland.
2Geophysical surveys were conducted in four microregions (Fig. 1A) where Roman sites have been reported and survey conditions were satisfactory. The results concerning the fort at Gračine have already been published (Pisz & Dziurdzik, 2019), while publications of discoveries from Studenci and Donji Radišići are in preparation.
3This paper presents the survey in the alluvial plains of Teskera and Hardomilje on the southern bank of the Trebižat river (Fig. 1B). Earlier research had revealed numerous Roman remains (with no exact spatial relations), including a major road (serving also as a levee) as well as several gravestones and miscellaneous surface finds. The data needed to be verified and expanded in an effort to assess the state of preservation of archaeological remains and the threats posed by agricultural activity.
Figure 1. Top (A): location of the four surveyed microregions; bottom (B): Teskera-Hardomilje area. UAV orthophotomap overlaid on satellite imagery and extent of the geophysical survey.
4The research approach and methodology were determined by the state of research (Schmidt et al., 2015). The survey started in 2016 with extensive fieldwalking documented with a handheld GPS. This revealed several concentrations of finds, mostly fragmented pieces of ceramics. An orthophotomap and DEM, covering roughly 60 ha, were created with DJI Phantom 4 UAV and Agisoft Metashape software (Fig. 1B). No visible crop marks, soil marks or landforms were captured.
5Extensive magnetic susceptibility measurements were made with a Bartington MS3 and MS2D field loop. The measured values (~20-400 SI) were recorded together with their location in a RTK GNSS controller. No particular concentrations of higher magnetic susceptibility were noted; local changes relate to fields with soil reportedly brought from elsewhere.
6The next step, carried out in 2018, was a magnetometer survey using a Multichannel Sensys MXV3 gradiometer in a pushed-cart configuration with six sensors in an array with a 0.5 m separation distance and RTK GNSS. An area of approximately 9 hectares was measured.
7The magnetic survey data was preprocessed with Sensys MAGNETO® software (sensor offset removal) and processed further in Geoplot 4 software (despiking, GPS Gap Fill, interpolation).
8The final stage was an earth resistance survey (ER), carried out with a Geoscan Research RM85 resistance meter, in a Wenner-α probe array with a 0.5 m probe separation distance. Approximately 2.5 ha were measured in total. The sample and traverse interval was 1 m with the exception of the Teskera–Nuga area where an urgent out-of-sequence ER survey had to be carried out in 2016-2017 to help develop protection plans of potential remains (Fig. 2C). Small-scale excavation by local archaeologists in 2017 verified the identified structure, putting a stop to the landowner’s plans. The structure was then measured again with 0.5 m sample and traverse intervals.
Figure 2. Top (A): magnetometer survey results from Teskera–Nuga and Hardomilje (West) areas; left bottom (B): orthophotomap of the area in Teskera–Nuga where a Roman structure was found; centre bottom (C): high-resolution ER survey (2017) overlaid on ER survey from 2016; right bottom (D): magnetometer survey results overlaid on an orthophotomap of the excavated Roman structure.
9The ER data was processed in Geoplot 4 software. Despike, Low Pass, High Pass and interpolation functions were applied subsequently.
10The geophysical data from all stages of the survey were visualised, georeferenced and uploaded to a QGIS spatial database.
11The geophysical methods applied proved effective, capturing even weak anomalies from barely preserved structures that were later verified in salvage excavations conducted by local archaeologists.
12The magnetic map is dominated by dipolar anomalies from ferrous objects and by strong positive zones (Fig. 2A). The former are caused by small ferrous debris dumped in the fields by farmers (with distinctive differences between particular plots), and field infrastructure like fences and reinforced concrete posts. The latter are most probably related to karst geology and might be produced by the magnetic fill of sinkholes etc. They occur near the southern edge of the surveyed area, where limestone bedrock outcrops are present. In the whole area we identified only two Roman structures.
13The ER survey delivered similar and complementary data. Apparent resistivity values were roughly 30-150 Ωm. Numerous high-resistance linear anomalies crossing the area correspond probably to alluvial gravel deposits. The two Roman structures identified with the magnetometer survey were also captured with this method.
14The magnetic anomalies on the Teskera–Nuga map were clear and strong (Fig. 2D). The limestone building wall produced a negative linear anomaly. The highly dynamic positive anomaly observed inside the outline of the building corresponds to layers of burning and is present despite their thinness.
15At Hardomilje–Kratina, a new part of the already known Roman grave monument (AE 2000, 1175) was found reused as a field boundary marker (Fig. 3A). A series of linear negative magnetic anomalies was captured in the immediate vicinity; they could be caused by the limestone remains of a road and building (Fig. 3B). ER produced less clear, but still complementary data (Fig. 3C). Overall, the multi-method geophysical data proved the presence of Roman remains (Fig. 3D).
Figure 3. Hardomilje-Kratina area. Top Left (A): extent of the geophysical survey; bottom left (B): magnetometer survey results, top right (C): ER survey results, bottom right (D): archaeological interpretation of geophysical data.
16During fieldwalking in Hardomilje (West), a long patch of small limestone fragments was found along the course of a well-preserved Roman road reported in 19th-century research. However, the area lacked any geophysical anomalies that could be interpreted as archaeological features.
17Sufficient soil magnetic susceptibility in the valley provides favourable conditions for magnetometry. Non-magnetic limestone features are detectable as negative linear anomalies, layers of burning produce strong thermo-remanent magnetic anomalies, but they should not be confused with common anomalies produced by karst-related features.
18The low resistivity of the soil creates proper conditions for detecting architectural remains with ER. However, anomalies of natural origin, probably alluvial gravel dyke deposits, predominate.
19Both methods effectively detected even poorly preserved remains, with ER data complementing magnetometer data.
20While only two buildings and a stretch of an ancient road can be identified with certainty, the survey provided valuable verification and spatial referencing of data from earlier research. Elements of Roman settlement landscape preserved in the late 19th century, including a long segment of a road through Hardomilje (West) were found to be destroyed by intensive farming in the area. Remains of architecture and archaeological layers are in extremely bad condition and located very close to the surface.
21The available data is by no means comprehensive or dated for that matter. It suggests a scattered settlement pattern with small villas, far apart, but centred on the road, probably lined by a cemetery.
22This landscape is vanishing. Intensive farming in the fertile Trebižat river valley presents a constant threat to the preservation of ephemeral archaeological remains. A large scale survey, continuing the present research, should be considered as top priority.