Statistical and Spatial Analysis of Hurricane-induced Power Outage and Restoration

Tropical cyclones (TC) cause billions of dollars in damage to coastal areas in the United States annually. Global climate change is increasing favorable environmental conditions for TCs which produce heavy flooding and precipitation. It is important to understand the communities that will be most affected, and will likely suffer the longest power outages. County-level power outage data from the Department of Energy’s Environment for Analysis of Geo-Located Energy Information (EAGLE-I) were used to analyze the relationships of environmental and socioeconomic variables on power outage trends, response, and recovery for power outages caused by two North Atlantic Basin hurricanes: Hurricane Florence (September 2018) and Hurricane Sally (September 2020). Hurricane Florence reached Category 4 wind speeds, but ultimately made landfall as a Category 1 along Wrightsville Beach, North Carolina. Hurricane Sally was a category 2 system that made landfall along Gulf Shores, Alabama. A daily average power outage baseline was created from four years of power outage records. From this baseline, power outages were identified that could be attributed to either hurricane. Quasi-Poisson and linear regression models were used to assess relationships between power outage trends, hurricane characteristics, and socioeconomic variables. Kruskal-Wallis tests were used to assess land classification and helped assess whether methods used were suitable for studying hurricane-related power outage trends. Relationships identified within data were found to be inconsistent across both study areas. Analyses of Florence’s study area indicated a statistically significant relationship between the number of consecutive days a county experiences power outages and daily average power outage with TC strength and percent of population below poverty. However, in Sally’s analyses, the affected county’s distance from the TC track was the only statistically significant variable found across the analyses. This paper analyzes the complex interactions between county level power outages caused by TCs and the influence of socioeconomic variables on them using EAGLE-I data.