Predicting ancient city sites using GEE coupled with geographic element features and temporal spectral features: a case study of the Neolithic and Bronze Age of Jianghan region

Abstract

As an important birthplace of civilization in China, the Yangtze River Basin has the necessary to discover further and investigate the ancient remains, and the archaeological site prediction model is significant for discovering and investigating archaeological remains. In this paper, we focused on the ancient city sites of the Neolithic and Bronze Age in Jianghan region in the middle reaches of the Yangtze River, annotated the specific locations and ranges of 33 ancient city sites using the Google Earth Engine (GEE) cloud platform, and proposed a machine learning ancient city site prediction model by coupling geographic element features and temporal spectral features. Results indicated that the ancient city sites were recognizable in different geographic elements and separable in Sentinel-2 multispectral bands and spectral indices; the coupled time series spectral features could improve the ability of the model to recognize the regions of the ancient city sites, the percentage of pixels with a high probability of prediction (greater than 0.57) within the range of the ancient city sites was 80.0%, and the distribution of the ancient city sites could be obtained from the precise high probability regions. The model proposed can be used to predict the potential geographic locations of ancient city sites and indicate the key areas for future field archaeological survey work.