Analysis of model-based electrical transmission disaster risk and extreme weather preparedness for the Southeastern Electric Reliability Council.

The transmission network of the North American power grid is susceptible to extreme weatherevents. The coastal states of the South Eastern Reliability Corporation (SERC) region:Louisiana, Mississippi, Alabama, Georgia, South Carolina, and Florida, experience a highfrequency of major electrical disturbances during between June and August. Understandingand forecasting the number of outages in this region has important benefits for planning andmanagement of the transmission network. To this end, we develop a regression model to predictthe number of power outages to expect during this season of interest based on anomalous seasurface temperatures in the Atlantic and Pacific Oceans, as these oscillations are known to havea direct impact on regional weather. Assumptions of constant variance and normality ofresiduals were satisfied. To address multi-collinearity, a principal component regression of theindependent variables was conducted as well as leave-one-out cross validation to ensurefairness and integrity of the model while reducing any potential sources of bias. Forecastensemble techniques enabled uncertainty analysis and a corresponding ranked probability skillscore (RPSS) of the model writ large. The generated hindcast from this work displayed strongpredictive capability of electrical outages in a novel fashion. The results of this work, therefore,demonstrate the ability to forecast seasonal power outages and transmission failures resultingfrom extreme weather, allowing for regional planners to allocate resources and personnelaccordingly.