WISPR: Modeling Tree-Induced Power Outages

This is a test article! I am still working on the longer, more serious version that walks through everything from processing the LiDAR, tree growth model, all the way to risk reduction optimization strategy.

This project builds a physics-informed and data-driven model to understand how trees interact with power infrastructure under high-wind conditions.
Using LiDAR-derived tree geometry, species attributes, and historical outage records, we estimate per-tree failure probabilities and aggregate them into grid-level risk maps.

The work is part of the WISPR (Wind & Infrastructure Spatial Risk) initiative, which aims to enhance climate resilience by identifying vulnerable areas across regional power networks.

Method Snapshot

Tree-level failure modeling
We estimate tree failure probabilities using logistic regression trained on tree geometry, species, and local wind exposure.
Spatial aggregation
Tree-level probabilities are aggregated onto power-grid segments to map the likelihood of storm-induced outages.
Wind super-resolution
A diffusion model generates minute-level wind profiles from hourly observations to drive high-resolution simulations.

This system helps utilities and policymakers anticipate climate-induced outage risks and plan targeted vegetation management, balancing safety, cost, and environmental goals.