The Byzantine settlements in the Negev Desert, particularly those surrounding ancient Shivta, have been the subject of various archaeological, historical, and geographical studies. Despite these efforts, the question of why Shivta was abandoned by its inhabitants remains under discussion. The reasons for abandonment are typically divided into two main categories: those related to violent conquest, epidemics, and man-made disasters, and those caused by natural events such as climate changes and earthquakes.
The proximity to the Dead Sea Transform – an area frequently affected by earthquakes – has led to the hypothesis that one or more of these seismic events may have destroyed the village. However, the literature surrounding this hypothesis remains ambiguous, with evidence both supporting and refuting it.
This study aims to address this gap by investigating whether potential earthquakes in late antiquity hit the village. To answer this, various damage patterns observed on the walls of the remaining structures were mapped and analyzed. Additionally, a novel methodology was developed that tailors Geo-Informatics tools and statistical tests to identify spatial and repetitive damage patterns. Specifically, V-shaped damage patterns were analyzed to assess whether (i) they resulted from a shared phenomenon and (ii) whether there is evidence linking this phenomenon to an earthquake.
The results from both statistical tests and machine learning clustering techniques, such as K-Means, reveal a repetitive damage pattern across the site’s walls, suggesting that these damages are likely correlated with one or more earthquake occurrences. Although it is not possible to precisely date the earthquake or link it to a historical event, the analysis offers support for the hypothesis that an earthquake contributed to the site destruction.
This preliminary research enhances our understanding of potential seismic events at the site of Shivta and provides a valuable methodology that will need to be examined at other archaeological sites.