Division of Forestry and Natural Resources, Davis College of Agriculture, Natural Resources and Design, West Virginia University, Morgantown, WV, 26506, USA.
Environ Manage. 2023 Sep;72(3):671-681. doi: 10.1007/s00267-023-01842-9. Epub 2023 Jun 21.
The Appalachian region of the United States has experienced significant growth in the production of natural gas. Developing the infrastructure required to transport this resource to market creates significant disturbances across the landscape, as both well pads and transportation pipelines must be created in this mountainous terrain. Midstream infrastructure, which includes pipeline rights-of-way and associated infrastructure, can cause significant environmental degradation, especially in the form of sedimentation. The introduction of this non-point source pollutant can be detrimental to freshwater ecosystems found throughout this region. This ecological risk has necessitated the enactment of regulations related to midstream infrastructure development. Weekly, inspectors travel afoot along new pipeline rights-of-way, monitoring the re-establishment of surface vegetation and identifying failing areas for future management. The topographically challenging terrain of West Virginia makes these inspections difficult and dangerous to the hiking inspectors. We evaluated the accuracy at which unmanned aerial vehicles replicated inspector classifications to evaluate their use as a complementary tool in the pipeline inspection process. Both RGB and multispectral sensor collections were performed, and a support vector machine classification model predicting vegetation cover were made for each dataset. Using inspector defined validation plots, our research found comparable high accuracy between the two collection sensors. This technique displays the capability of augmenting the current inspection process, though it is likely that the model can be improved further. The high accuracy thus obtained suggests valuable implementation of this widely available technology in aiding these challenging inspections.
美国阿巴拉契亚地区的天然气产量增长显著。为了将这种资源输送到市场，必须开发基础设施，这在多山的地形中既需要建立井场，也需要建立运输管道。中游基础设施包括管道的通行权和相关基础设施，可能会导致严重的环境退化，特别是以泥沙淤积的形式。这种非点源污染物的引入可能会对整个地区的淡水生态系统造成危害。这种生态风险要求制定与中游基础设施发展相关的法规。每周，检查员徒步沿着新的管道通行权进行检查，监测地表植被的重新建立，并确定未来管理中出现问题的区域。西弗吉尼亚州地形复杂，给徒步检查员的检查工作带来了困难和危险。我们评估了无人机复制检查员分类的准确性，以评估它们在管道检查过程中作为补充工具的使用。我们对 RGB 和多光谱传感器进行了采集，并为每个数据集制作了预测植被覆盖的支持向量机分类模型。使用检查员定义的验证图，我们的研究发现两种传感器采集的结果具有可比性，都达到了很高的精度。这项技术展示了增强当前检查过程的能力，尽管模型可能还可以进一步改进。如此高的准确性表明，广泛应用这种技术可以帮助进行具有挑战性的检查。
