Loading data sources and coverage…
Every figure on this site traces to a named source and a stated method. When a utility does not report, we show the gap rather than estimate around it, and live and historical figures are never added together.
The national base is ODIN(DOE/Oak Ridge National Laboratory), which collects near-real-time reports from participating utilities. On top of it we add statewide emergency-management aggregators (e.g. MEMA, Cal OES) and major utilities’ public outage maps. Where one source authoritatively aggregates a whole state, other sources’ rows for that state are dropped so a customer is never double-counted. Where a feed reports only a utility- or state-wide total, that total counts toward the state figure but is never assigned to a specific county; it is shown as unmapped rather than guessed.
County alert thresholds compare customers out (and % out, where a denominator exists) against the values you set. For New Jersey utilities we also flag NJ BPU’s major-event line, at least 10% of a utility’s customers in an operating area out (N.J.A.C. 14:5-8.9), applied per utility per county. This is an indicator computed from public map data, not the utility’s or the BPU’s official determination.
History comes from EAGLE-I (DOE/ORNL), which recorded county-level outages at 15-minute resolution from 2014-11 onward. We fold those readings into outage eventsand sum customer-minutes without power over a year, then divide by the customers tracked in that county, giving an annual “average hours without power per customer” figure analogous to the utility reliability metric SAIDI. Because EAGLE-I coverage varies by year, each county page shows the share of customers tracked so the figure can be read in context.
Shown only during a genuinely meaningful active outage: at least 3% of the county out, or 1,200 customers. It is the median duration of past outages of a similar size in that same county, drawn from the EAGLE-I event history, with the sample size stated on the page. “Taking longer than usual” is judged by how long the current outage has actually been running versus that historical median, not a static size comparison. This is a historical estimate, not a utility’s official ETA, and it is labeled as such wherever it appears.
Reliability metrics and service territory come from the US EIA Form 861. SAIDI is the average duration of interruptions per customer per year; SAIFI is the average number of interruptions; CAIDI (SAIDI ÷ SAIFI) is the average length of a single interruption. We report the IEEE-standard figures, which include major event days. Provider rankings order utilities by SAIDI (lower is more reliable) among those of comparable territory size.
County pages can overlay HHS emPOWER public, de-identified counts of Medicare beneficiaries who rely on electricity-dependent medical equipment. These are aggregate counts only; identifiable data is available to public health authorities through state and local health departments.
Retail electricity prices are not published at the county level. We derive a county estimate by taking each serving utility’s residential price from EIA-861and combining them, weighted by each utility’s residential customer count where that count is available (a simple average otherwise). Because utilities report customers at the state level, a true per-county customer share is not available; this is an estimate, labeled as one on every page it appears.
Per-hazard ratings (hurricane, tornado, ice storm, wildfire, flooding, and others) and the social-vulnerability rating come from FEMA’s National Risk Index. The composite outage-risk tier shown on each county page combines FEMA’s weather-hazard exposure with that county’s own decade of observed outage hours, so it reflects both what the county is exposed to and what it has actually experienced.
Found a number that looks wrong? Email [email protected] with the page and what you saw; we fix errors promptly.