Overstory
Use satellite-imagery AI to help power companies manage the trees beside power lines
During a typhoon, a power outage is often not because a pylon fell but because a tree fell on the power line. Overstory uses satellite and aerial imagery plus AI to tell power companies which line segment's vegetation is most dangerous and which to fix first, replacing "cycle-based full-line trimming" with "risk-based prioritization."
Features and application scenarios
The system analyzes satellite and aerial imagery, identifying vegetation height, growth rate, and clearance from the line, then combines asset locations, wildfire-risk maps, and terrain data to produce a risk priority ranking. The interface has both desktop and mobile dashboards, letting office planners and field crews see the same data, and it provides SAIDI (System Average Interruption Duration Index) impact analysis, directly linking trimming decisions to outage hours. Customers include six of the top ten North American power companies, plus many operators in the Americas and Europe.
Suited to power companies' asset-maintenance, line-operations, and wildfire-risk-management departments. Taiwan's situation is a bit different — we don't have California-scale wildfires, but the tree-fall outages caused by typhoons and heavy rain each year are a real pain. If Taipower and local line maintenance could prioritize with similar logic, resources would be used more precisely. The barrier for this kind of tool isn't the technology but whether the unit treats trees as an asset that needs managing.
Main features
- AI vegetation-risk analysis of satellite and aerial imagery
- Risk-ranked trimming work-order priority
- Combines asset location, wildfire maps, and terrain data
- Desktop and mobile dashboards
- SAIDI outage-metric impact analysis
Common uses
- Distribution-line vegetation-trimming scheduling
- Risk-hotspot inventory before wildfires and typhoons
- Maintenance-budget allocation and outcome tracking
- Power-supply reliability-metric improvement
Key Features
- AI vegetation-risk analysis of satellite and aerial imagery
- Risk-ranked trimming work-order priority
- Combines asset location, wildfire maps, and terrain data
- Desktop and mobile dashboards
- SAIDI outage-metric impact analysis
Pros
- Replaces cyclical trimming with risk-oriented, cutting more outages on the same budget
- Actually adopted by large power companies, not a proof of concept
- Data links directly to the reliability metrics regulators care about
Cons
- Serves power utilities; general enterprises can't use it
- No public adoption case in Taiwan yet; local map data and tree species need validation
- Pricing requires inquiry; it's an enterprise-grade project
Use Cases
- Distribution-line vegetation-trimming scheduling
- Risk-hotspot inventory before wildfires and typhoons
- Maintenance-budget allocation and outcome tracking
- Power-supply reliability-metric improvement
Editor's Note
Full-line trimming is the fairest and also the most wasteful. Sending limited crews to the few truly dangerous kilometers — that's what data should do.
FAQ
Taiwan's line density differs a lot from the US — is it still applicable?
The algorithm's core is the spatial relationship between vegetation and lines, and the principle is universal, but Taiwan's tree species, terrain, and line density all differ, so the model needs local-data calibration. The reasonable approach is a small-scale pilot in a single area first to validate accuracy.
Is satellite imagery's resolution enough to see a single tree?
Overstory mixes multi-source imagery like satellite and aerial, and the resolution obtainable varies by region. When evaluating, ask clearly what image grade your service area can actually get and how fast the update frequency is — frequency is often more critical than resolution.