Analyzing Spatiotemporal Trends and Anomalies in Power Outage Data Across the United States

Power outages have far-reaching and often disproportionate impacts on communities and the economy of the United States. Weather-related events, which account for approximately 80% of major power outages, have doubled in frequency over the past decade compared to the early 2000s. In this study, power outage data from 2015-2022 at 15 minute interval is analyzed to understand the spatio-temporal patterns of power outages across the US. In addition a Bayesian Gaussian Mixture Model was used to detect anomalous power events in the country. Outage events were analyzed in 3 categories of less than 2 hours, 2-8 hours and greater than 8 hours. Results of analysis show that all outage events show an increasing trend with the large outage events (8+ hours) showing the highest increasing trend amongst the 3 categories. All the states in the US shows an increase in the longer duration power outage (8+ hours) with the states of Texas and California showing the highest trend. Similarly anomalous power events also show an increasing trend with most anomalous values occurring in the month of June, July, and August and concentrated in the south-eastern and north-eastern part of the US. Results of this study underscore that as climate change intensifies extreme weather events, the aging electrical grid faces increasing pressure, highlighting the urgent need for infrastructure upgrades and improved resilience strategies to mitigate the growing challenges posed by power outages across the country.