Vegetation management along powerlines with remote sensing

Ching Man Yung

Research & Development Group
Hitachi, Ltd.

Introduction

Vegetation management (VM) plays a crucial role in minimizing forest fire risks and preventing power outages caused by tree encroachment along power lines. With the increasing frequency of severe weather events, transmission line operators who have traditionally conducted labor-intensive, cyclical maintenance such as on-site tree height measurement and pruning during outages, are now urgently looking at remote sensing technologies to facilitate predictive and preventative VM.

How can remote sensing be used in vegetation management?

Remote sensing involves collecting data on the Earth’s surface from a distance using satellites or unmanned aerial vehicles (UAVs) such as drones. This technology is particularly beneficial for VM, as it enables large-scale monitoring of plant health, disease detection, invasive species management, and vegetation mapping. However, while UAVs and LiDAR provide accurate imagery for monitoring tree encroachment, their high costs limit their commercial viability for frequent updates.

At Hitachi, my colleagues and I are investigating a cost-effective and scalable solution by using high-resolution satellite imagery for powerline risk assessment which we reported on at 2024 IEEE International Geoscience and Remote Sensing Symposium.[1]

Individual tree-based risk assessment approach using satellite imagery

The framework that we are looking at, individual tree-based risk assessment consists of four components: pixel-based tree height estimation, treetop detection, crown delineation, and risk evaluation. In this approach, we generated a tree height map using a Random Forest model trained on LiDAR ground truths collected via drone flights. The map facilitates treetop detection through local maximum filtering, and our adaptive window size proposal further allows for accurate crown delineation.

This novel risk analysis method includes neighborhood searching, potential failure zone identification, and risk calculation. For each powerline segment, a buffer is drawn to capture neighboring trees and calculate the potential failure zone using USDA's Hazard Tree Guideline[2]. This step is crucial for calculating the collision probability between trees and powerlines. Risk is classified into four levels—Negligible, Minor, Moderate, and High—and a color-coded risk map is generated to highlight critical areas. Such visualizations will empower powerline operators to quickly identify and address high-risk sections, and thereby enhance their maintenance strategies.

Through evaluation using LiDAR inputs as ground truth, we demonstrated that our framework achieves performance comparable to high-resolution LiDAR inputs. Specifically, our framework obtained a precision of 0.922 and a recall of 0.819, resulting in a high F1 score of 0868. Additionally, the Macro Averaged Mean Absolute (MAMAE) is 0.669, indicating an average classification order difference of approximately 0.7.

Conclusion

This innovative approach to vegetation management underscores Hitachi’s commitment to deliver transformative digital solutions that address our customers’ challenges with Lumada and cutting-edge technology. By leveraging remote sensing and high-resolution satellite imagery for tree-based risk assessment, we can provide a scalable, cost-effective solution that enhances the reliability and safety of powerline infrastructure. This development will empower energy industries, particularly transmission line operators, to embrace digital transformation, creating smarter and more resilient systems that effectively respond to the challenges of climate change and evolving environmental conditions. This work enhances Hitachi Energy’s Vegetation Manager [3], enabling continuous monitoring and proactive prevention of vegetation-related risks for utility management.

Acknowledgements

I would like to thank the members of my team and colleagues in Hitachi, Yu Zhao, Tomonori Yamamoto, Koichiro Yawata, Shinji Matsuda, and Norihiko Moriwaki, who contributed their expertise and effort in this project. I would also like to express my gratitude for the constructive feedback from Hitachi Energy and their invaluable support and insights, which has significantly enhanced the quality of this work.