PG&E is extracting data from high resolution satellite imagery to quantify wildfire risk and its drivers along their 100,000+ electric distribution line miles to identify the best risk mitigation measures available within given cost and other restraints.

Vegetation failures are a top contributor to ignitions on the distribution system presenting significant wildfire risk. Several types of vegetation-related data identified through satellite imagery have proven advantageous as electric utility companies search for solutions to minimize electric grid failures that can lead to catastrophic wildfires including identification of trees within striking distance of the line, tree and canopy height, and indicators of tree health.

PG&E will reveal their machine learning predictive model using vegetation-related data extracted from high resolution satellite imagery to help inform prioritization of costly vegetation management work.

PG&E will teach participants about:

• their framework for quantifying wildfire risk.
• tips for utilizing high resolution satellite imagery as an ML model input.
• useful data to extract from high resolution satellite imagery.