Scholarlink

<p>Managing water resources in arid and semi-arid regions has been a fundamental human endeavor, essential for the establishment and sustainability of both urban and rural communities. Qanats—ancient underground water systems—have long been a cornerstone of water supply and distribution in the arid Middle East. These subterranean systems are typically identified from the surface by vertical shafts that mark the path of the underground water channel. In this paper, we apply a machine learning approach to identify qanat shafts and their underground channels from historical aerial imagery. Our primary objective is to automatically detect qanats within the historical landscape of the city of Zuzan, Iran, using convolutional neural networks. Specifically, we trained the You Only Look Once (YOLO) object detection algorithm, a type of convolutional neural network, with aerial images from the 1960s covering various areas of the Iranian Plateau where qanat remains have been verified by direct inspection. The trained model enabled the detection of several qanat channels leading to the city of Zuzan, a region where no intensive archaeological surveys had previously been conducted. Historical analysis suggests that in this region, these qanats were the primary factor in the formation and sustenance of Zuzan during both historical and Islamic periods, and they continue to function today.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Using convolutional neural networks to identify qanats and analysis of their role in the formation of Zuzan City (Iran, Islamic period)

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Fereshte Azarkhordad,
  • Juan A. Barceló,
  • Hasan Hashemi Zarajabad,
  • Abed Taghavi

摘要

Managing water resources in arid and semi-arid regions has been a fundamental human endeavor, essential for the establishment and sustainability of both urban and rural communities. Qanats—ancient underground water systems—have long been a cornerstone of water supply and distribution in the arid Middle East. These subterranean systems are typically identified from the surface by vertical shafts that mark the path of the underground water channel. In this paper, we apply a machine learning approach to identify qanat shafts and their underground channels from historical aerial imagery. Our primary objective is to automatically detect qanats within the historical landscape of the city of Zuzan, Iran, using convolutional neural networks. Specifically, we trained the You Only Look Once (YOLO) object detection algorithm, a type of convolutional neural network, with aerial images from the 1960s covering various areas of the Iranian Plateau where qanat remains have been verified by direct inspection. The trained model enabled the detection of several qanat channels leading to the city of Zuzan, a region where no intensive archaeological surveys had previously been conducted. Historical analysis suggests that in this region, these qanats were the primary factor in the formation and sustenance of Zuzan during both historical and Islamic periods, and they continue to function today.