Abstract Torrential rainfall can generate landslides, flash floods, and debris flows which might become disasters, causing loss of life and damage to property and infrastructure. To respond opportunely to hydrometeorological hazards, it is necessary to assess, rapidly and accurately, damage to the affected area. This is commonly done through time-consuming reconnaissance visits to obtain detailed field information. This paper proposes a methodology which uses: i) high resolution satellite and RGB images from unmanned aerial vehicles (UAV), ii) digital elevation models (DEM), and iii) object-based image analysis (OBIA) for rapid urban flood damage assessment and estimation of the number of houses washed away, or with a total or partial roof collapse, by comparing pre- and post-event data. The case study was Tropical Storm Earl in 2016 that affected the town of Chicahuaxtla, Puebla, Mexico, due to the overflow of the Zempoloantongo River that cuts through the town causing several loss of life and severe property damage. The results indicate that the three-pronged approach proposed herein is able to discriminate changes before and after the event and improve image classification of washed-away or destroyed houses. The overall accuracy of the proposed automatic classification obtained with UAV data had a value of 97.4%. Structural damage was not assessed in this study.

中文翻译：

---

使用高分辨率遥感数据和基于对象的方法快速城市洪水灾害评估

摘要 暴雨会引发山体滑坡、山洪暴发和泥石流等灾害，造成人员伤亡，财产和基础设施受损。为了及时应对水文气象灾害，有必要快速准确地评估受影响地区的损失。这通常是通过耗时的勘察访问来获得详细的现场信息来完成的。本文提出了一种方法，该方法使用：i) 来自无人机 (UAV) 的高分辨率卫星和 RGB 图像，ii) 数字高程模型 (DEM)，以及 iii) 基于对象的图像分析 (OBIA)，用于快速城市洪水灾害评估通过比较事前和事后的数据，估计被冲毁或完全或部分屋顶倒塌的房屋数量。案例研究是 2016 年影响墨西哥普埃布拉州奇卡瓦斯特拉镇的热带风暴厄尔，因为横穿该镇的 Zempoloantongo 河泛滥，造成数人死亡和严重财产损失。结果表明，本文提出的三管齐下的方法能够区分事件前后的变化，并改善被冲毁或毁坏房屋的图像分类。使用无人机数据获得的拟议自动分类的总体准确率为 97.4%。本研究未评估结构损坏。结果表明，本文提出的三管齐下的方法能够区分事件前后的变化，并改善被冲毁或毁坏房屋的图像分类。使用无人机数据获得的拟议自动分类的总体准确率为 97.4%。本研究未评估结构损坏。结果表明，本文提出的三管齐下的方法能够区分事件前后的变化，并改善被冲毁或毁坏房屋的图像分类。使用无人机数据获得的拟议自动分类的总体准确率为 97.4%。本研究未评估结构损坏。