Improving Winter Power Outage Forecasts with a Snow Index

Winter storms produce a wide range of forecast challenges when it comes to predicting power outages. Electric grids are susceptible to high winds, ice accretion, and hefty snow loads. Due to the complex nature of how it occurs, wet snow, snow that has a high density, is an extremely difficult phenomena to forecast. Wet snow typically has low snow-to-liquid ratios and occurs when falling snow particles encounter a very shallow warm surface layer. Even though this problem is well documented, there still remains a large gap in the ability to accurately identify when, where, and how much wet snow will occur. In this presentation, we will display the connection between the density of snow and power outages across the Northeast and Mid-Atlantic United States. This snow index looks to isolate the true impact of snow density by removing external factors like strong wind gusts and significant levels of ice accumulation that can occur during the same winter storms. By applying a probability density function to mirror the impact of snow density and total outages, snow density can be graded on an impact-based scale from 0 to 1 rather than just a numeric value or as a qualitative condition (dry or wet). Improvements in power outage forecasts will be shown using the University of Connecticut’s Outage Prediction Model. Future research objectives for the snow index, including its connection to numerical modeling, will be highlighted.