3rd Croatian Conference on Earthquake Engineering 3CroCEE

The world has witnessed widespread damages and losses suffered by precious cultural heritage structures during recent earthquakes e,g. Emilia-Romagna, Italy 2012; Central Italy 2016; Turkey /Syria 2023; Morocco 2023; Tohoku, Japan 2011; Puebla, Mexico 2017; Nepal 2015; Christchurch, New Zealand 2011; Iran 2017, among others. Based on traditional surveys and reconnaissance using 3D documentation technologies e.g photogrammetry, TLS etc. following the destructive earthquakes, extensive amounts of observed damage data about the damages of historical churches and mosques etc. has been accumulated over the years.
This paper presents the results of attempts to systematically analyze the damage data for historical churches and mosques to identify the lessons we can learn therefrom, determine their patterns of damage, and evaluate the possible causes of damage. The paper first documents the deep experience and knowledge base of performance of Churches in Italy during the past major earthquakes. The results of an analysis of the observed damage data from the historical colonial churches that suffered extensive damages and losses during the 2017 Puebla, Mexico earthquake, are presented next. The paper then analyzes the available data on the performance and damages suffered by historical mosques during the catastrophic Turkey earthquake of 2023; and the Al Houz, Morocco earthquake of 2023, to classify the types of damage, identify the patterns of damage, and possible causes of damage. The results of this study will help in the structural assessment of damaged historical churches and mosques and the development of the plans for their repair and restoration.
This paper further plans to explore the potential application of AI/ML systems for the systematic analysis of observed damage image data for historical Churches and Mosques, including classification of features, types of damage, and determining the patterns of damage of historical churches and mosques during catastrophic earthquakes.
Many data types and sources describe historical buildings, including 3D scans, photogrammetry, survey images and videos. This wealth of data requires the support of deep learning systems to identify types of damage due to natural disasters. The analysis of this large volume of data also offers the opportunity to compare before-and-after images to determine requirements of restoration projects.