Weather Regime Drivers of Power Outages over the Continental U.S.

Abstract Power outages are a common, costly occurrence in the United States and pose risks to human health and safety. Transmission and distribution equipment is exposed to many weather hazards, placing additional stress on aging grid components. Utility-specific outage datasets cover single regions, and national-scale datasets such as OE-417 Emergency reports only capture large events exceeding reporting thresholds, restricting the ability to construct large-scale outage patterns. Here, we present a novel approach for extracting outage days from the NOAA Storm Events Database to characterise outage patterns over 1996–2024. We compare these days to a benchmark large-outage dataset derived from OE-417 reporting (Purdue dataset). We find NOAA-derived outage days exhibit similar temporal and spatial patterns to the Purdue dataset. Both NOAA and Purdue outage days are thunderstorm driven, although NOAA outage days account for a larger share of thunderstorm-related events (73%) than Purdue (48%), highlighting differences between large-scale and small-scale outage events. Grouping outage anomalies by North American weather regime, we observe a significant 3% increase in outage probability relative to climatology over MISO-N states during Pacific Ridge days and significant increases over Southeastern TSOs during Greenland High days. These increases align with the leading edge of upper-level troughs in each regime. Further, we find strong seasonal variability in regime-level outage probabilities, with Western and Northeastern TSOs peaking in winter, Midwestern TSOs in spring, and Southern TSOs in summer and fall. These results establish a climatological baseline of regime- and season-specific outage risk, improving understanding of atmospheric conditions associated with power outages.