Finding the Early Christian Basilica of Ancient Messene — Before the First Trench

The Challenge

Ancient Messene was founded in 369 BCE at the pre-existing site of Ithome and became the civic heart of the liberated Messenian nation — an economic and cultural hub that adapted to Roman rule, survived into late antiquity, and was likely struck hard by the famous earthquake of 365 CE.

Between 2020 and 2023 the Open University of Cyprus team excavated east of the Asklepieion complex, where a large semicircular building stood visible at the surface with tall standing walls. What emerged was a spacious early Byzantine ecclesiastical complex: a basilica with a 6.35 m semicircular chord, side annexes incorporating earlier Roman structures, and — layered above the ruins — a medieval cemetery whose last burials date to the 12th–14th century CE.

Four seasons in, the constraint was no longer archaeological knowledge. It was time and money. Every trench opened in the wrong place costs a season. The team needed to know where the buried architecture continued, how deep it lay, and — equally valuable — where there was nothing at all, so that no effort would be wasted on empty ground.

This is precisely the question facing ministries of antiquities, heritage authorities, port operators and infrastructure developers from Athens to Nicosia, from Jerusalem to Abu Dhabi: what lies beneath the surface, exactly where, and at what depth — before anyone breaks ground?

Why Ground Penetrating Radar

GPR emits an electromagnetic pulse into the ground. The pulse travels at a speed governed by the host material’s dielectric constant until it meets a contrast — a wall, a void, a grave fill, a geological interface — and part of its energy reflects back. By measuring two-way traveltime, the instrument converts the echo into depth; by reading amplitude and phase, it distinguishes soil layers from buried objects from voids.

The dielectric constant is the governing variable, ranging from 1 to 81. The higher it climbs, the harder it is for the signal to reach depth. Values between 1 and 20 — air, dry soils, low-salinity ground — are ideal. At Messene the prominent dielectric was 6–8 and conditions were very dry, which is why the survey performed as well as it did.

Our Method

A single radar profile shows a shape. Only a grid shows a building. We treated the standing church as the origin of a local x–y coordinate system and collected dense parallel transects in both directions — horizontal profiles along x, vertical profiles along y — so that every target could be resolved as a plan-view form with orientation and extent, not merely as an isolated hyperbola.

A 450 MHz antenna was selected as the standard compromise for archaeological prospection: enough resolution for masonry and grave-scale features, enough penetration to reach 3–6 m in favourable ground. The dry soils allowed a maximum depth of about 4 m. Knowing the targets lay between 0 and 3 m, gain and time window were tuned accordingly (0–60 ns) to maximise signal-to-noise ratio.

Four grids were laid out — west, east, north and the church floor. Profile spacing was 50 cm along both axes, dense relative to target size. In total 268 GPR profiles were acquired, including test lines and repeated passes with varied settings. Every trace was synchronised with GPS, giving each target both local and geographic coordinates so results could be exported directly onto maps.

Processing: where the value is created

Raw radar is not evidence. Every dataset passed through a standardised pipeline:

• Time-zero correction to remove the direct-wave effect — the energy travelling straight from transmitter to receiver, which creates a false reflection.
• Automatic gain curve to recover amplitude with depth.
• Bandpass filtering and background removal to suppress environmental noise and horizontal banding.
• 2D migration to collapse diffraction hyperbolae back to their true source positions.
• Topographic correction — critical on a sloping archaeological site, and routinely skipped by less careful practitioners.
• Finsler imaging techniques, a modified-Finslerian ray-tracing approach developed in-house, for optimal handling and visualisation of the wavefield.

Finally, the excavated architectural remains were 3D scanned, so that the radar findings could be digitally correlated — geometry against geometry — with what the archaeologists had already exposed. This step turned interpretation from an argument into a measurement.

Results

Of the four grids, the west grid and the church floor produced significant findings. The south side could not be surveyed — there was no access. Potential artefacts were identified at depths from 0 to 1.5 m on the west side of the excavated walls, extending roughly 30 m beyond the last excavation point.

The church floor: a probable hidden grave

Inside a very restricted space, we collected dense lines at 20 cm inline and crossline spacing with a 50 ns time window — a shallower window justified because the floor had already been excavated down below the level of the surrounding ground.

The data revealed a high-amplitude target appearing consistently at 60 cm depth. On the basis of its amplitude response, its lateral continuity across adjacent lines, and the well-documented burial practice at this church — where a medieval cemetery evolved around and over the earlier complex — the target is interpreted as a probable hidden grave or series of graves. It is now a defined excavation objective rather than a lucky discovery waiting to happen.

The west grid: the basilica’s long walls, projected

West of the excavated church we applied a 2.5D survey — 2D profiles interpolated into a 3D data volume, from which horizontal amplitude slices (time slices or depth slices) can be extracted. Thirty-two vertical profiles were collected on this grid, supported by six horizontal profiles that constrained the interpretation.

Two unambiguous targets emerged. We interpret them as the buried extension of the church’s east–west long walls, on four independent grounds:

Four independent arguments — geometric, geophysical, stratigraphic and architectural — converging on a single interpretation. That is the standard we hold radar interpretation to, and it is what separates a defensible target from a coloured blob on a screen.

The Negative Result Is Also a Deliverable

The most under-valued output of a well-designed GPR survey is the confident identification of empty ground. At Messene the survey did not only say where to dig; it said where not to dig. For a project budget measured in excavation seasons, or a construction programme measured in delay days, negative evidence is worth as much as positive evidence — and only a complete, dense grid can produce it credibly.

Why This Matters Beyond Greece

Ancient Messene is a dry, well-drained inland site with masonry targets in the first three metres of the subsurface. That description fits an enormous proportion of the archaeological and heritage estate across Cyprus, Israel, Jordan, Lebanon, Egypt, the United Arab Emirates, Saudi Arabia, Qatar and Bahrain — regions where rapid development and dense, deep archaeology occupy the same ground.

The methodology transfers directly to:

• Pre-excavation targeting for research and rescue archaeology — spend the trench budget where it will pay.
• Pre-construction archaeological screening ahead of infrastructure, pipelines, roads, metro alignments and coastal development.
• Cemetery and grave mapping, including unmarked burials, in heritage and forensic contexts.
• Heritage-site management plans that require documented subsurface baselines before intervention.
• Church, mosque, monastery and museum floor surveys, where excavation is effectively impossible and radar is the only option.
• Integration with UXO and utility clearance on sites where archaeology and modern hazards overlap.

Geomorph Imaging Solutions is hardware-agnostic. We select the antenna frequencies, positioning systems and processing chain that the geology, the target depth and the project budget actually demand. The value we deliver sits in survey design, processing and interpretation — not in a logo on an antenna. Our results are peer-reviewed, GIS-ready and built to support real decisions by real project managers.

Frequently Asked Questions

Can ground penetrating radar find buried walls before excavation?

Yes. At Ancient Messene, GPR imaged two strong linear reflectors roughly 1 m below the surface, separated by about 4 m — matching the 3.9 m separation of the already-excavated church walls. The radar targets were interpreted as the buried continuation of the basilica’s east–west long walls, and this was established before any new trench was opened.

How deep can GPR see on an archaeological site?

Depth of investigation depends on antenna frequency and soil electrical properties. At Messene a 450 MHz antenna in very dry soils with a dielectric constant of 6–8 reached a maximum depth of about 4 m. In wetter or clay-rich ground, penetration falls sharply. Most archaeological targets in the Eastern Mediterranean lie in the top 0–3 m, which is exactly the range these systems image best.

Can GPR detect graves and burials?

GPR can detect the disturbance, voids and material contrasts associated with graves. On the floor of the Messene church, dense 20 cm grid lines revealed a consistent high-amplitude anomaly at 60 cm depth, interpreted on the basis of its amplitude response and archaeological context as a probable hidden grave or series of graves.

Is GPR survey suitable for protected heritage sites in Cyprus, Israel and the Gulf?

Yes. GPR is entirely non-invasive: nothing is dug, drilled or removed, which makes it acceptable to antiquities authorities and heritage regulators. The dry, low-conductivity soils typical of Cyprus, Israel, Jordan, the UAE and Saudi Arabia are among the most favourable conditions for radar penetration.

What deliverables come from an archaeological GPR survey?

Typical deliverables are processed 2D radargrams, interpolated 2.5D and 3D data volumes, horizontal depth slices, GNSS-referenced target coordinates in both local and geographic systems, GIS-ready overlays on site plans and aerial imagery, and a written interpretation ranking each anomaly with recommended trench locations.

How long does a survey like this take?

Acquisition of a dense, bi-directional grid of the scale described here is typically a matter of days rather than weeks; processing, correlation and reporting follow. Mobilisation across Greece, Cyprus, Israel and the Gulf is routine for our teams.

Planning a project over sensitive or unknown ground?

Geomorph Imaging Solutions delivers non-invasive subsurface mapping for archaeology, infrastructure and heritage projects across Greece, Cyprus, Israel, the Eastern Mediterranean and the Gulf. Tell us about your site — we will tell you honestly whether radar can answer your question.

Discuss your site with us

Full publication: Arvanitis, M., Deligiannakis, G., and Tsivikis, N. (2024). “GPR survey for the precise location of early Christian ecclesiastical components in ancient Messene, Greece.” The Leading Edge, 43(10), 665–669. Society of Exploration Geophysicists. https://doi.org/10.1190/tle43100665.1
Figures reproduced from the above, © Society of Exploration Geophysicists.

Related case study: Mapping Two Ancient Greek Cities Without Digging: GPR at Lechaion and Sikyon · Our geophysical survey services