Scholarlink

<p>Flooding has long-lasting impacts on rural communities, where recovery tends to be slow, costly, and difficult. This study introduces a hybrid methodology to track long-term recovery from the 2019 Midwest floods across 59 affected communities in eastern Nebraska. Using a pre-trained Siamese U-Net deep learning model with manual digitization of high-resolution satellite imagery (2020–2024). Findings reveal stark disparities in recovery progress across infrastructure categories and geographic regions. While transportation infrastructure demonstrates substantial recovery, housing recovery lags significantly reaching only 28% in the most flood-affected rural communities even six years after the 2019 flood disaster underscoring the disproportionate challenges these areas continue to face. Spatial differences are evident with Cherry and Pierce counties nearly recovered, while Sarpy and Holt continue to lag. Community case studies highlight varied approaches, including property demolition and managed retreat, yet comprehensive recovery remains limited. Although the model accurately detects demolition, it struggles with new construction, reinforcing the need for manual validation. The findings underscore uneven recovery trajectories and the importance of place-based, resilience-focused strategies.</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

Uneven recoveries: a deep learning assessment of the 2019 Midwest floods and their impact on rural communities

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月31日

查看全文
  • Rao Nargis Jahan,
  • Jenny B. Mason,
  • Jahangeer Jahangeer,
  • Jiyoung Lee,
  • Pranjay Joshi,
  • Zhenghong Tang

摘要

Flooding has long-lasting impacts on rural communities, where recovery tends to be slow, costly, and difficult. This study introduces a hybrid methodology to track long-term recovery from the 2019 Midwest floods across 59 affected communities in eastern Nebraska. Using a pre-trained Siamese U-Net deep learning model with manual digitization of high-resolution satellite imagery (2020–2024). Findings reveal stark disparities in recovery progress across infrastructure categories and geographic regions. While transportation infrastructure demonstrates substantial recovery, housing recovery lags significantly reaching only 28% in the most flood-affected rural communities even six years after the 2019 flood disaster underscoring the disproportionate challenges these areas continue to face. Spatial differences are evident with Cherry and Pierce counties nearly recovered, while Sarpy and Holt continue to lag. Community case studies highlight varied approaches, including property demolition and managed retreat, yet comprehensive recovery remains limited. Although the model accurately detects demolition, it struggles with new construction, reinforcing the need for manual validation. The findings underscore uneven recovery trajectories and the importance of place-based, resilience-focused strategies.