Revealing the Hidden Archaeological Landscape of Ancient Colchis Using Multispectral and Thermal UAV Remote Sensing

Abstract

This study investigates the potential of integrated multispectral and thermal UAV remote sensing for detecting and mapping hidden archaeological structures of Proto-Colchian and Colchian settlement systems of the 3rd–1st millennium BCE, within the Rioni River basin in western Georgia. Characteristic settlement mounds (Dikhagudzuba), moats/ditches, and possible paleochannel networks represent a unique example of large-scale adaptation to waterlogged environments, and limited archaeological excavations have significantly hindered the identification and interpretation of these buried landscape features. Pilot UAV surveys were conducted at four archaeological sites differing in location, geomorphological setting, land use, and degree of preservation using DJI Mavic 3M multispectral and DJI Matrice 4T thermal UAV platforms. Multispectral imagery enabled the detection of vegetation and soil-moisture anomalies associated with buried ditches, embankments, canals, and settlement structures, while thermal data revealed differences in thermal inertia related to subsurface archaeological remains and palaeohydrological features. The study covered approximately 15.4 ha and produced 5836 RGB, 2219 multispectral, and 3617 thermal images with spatial resolutions ranging from 1.87 to 4.61 cm. The integration of high-resolution elevation models, RGB imagery, vegetation indices (ExG and normalized EVI2), and thermal data enabled the identification and characterization of microtopographic, vegetation, moisture, and thermal anomalies associated with archaeological and palaeohydrological features. The investigated mounds cover 2754–4685 m2, reach approximately 9–10 m in height, and are surrounded by moats/ditches covering 4265–6520 m2 and ramparts reaching approximately 2–3.5 m in height. The results demonstrate that combined UAV-based multispectral and thermal remote sensing provides a highly effective non-invasive tool for archaeological prospection in dynamic wetland landscapes. The proposed approach significantly improves the detection of buried archaeological features and paleochannels, while also supporting landscape-scale analyses, monitoring of archaeological heritage, and the development of predictive archaeological models for western Georgia.

The authors' abstract, as published at the source. Remote Sensing, 2026 · DOI ↗