READINESS IS A PRACTICE. START HERE.INDEPENDENT PREPAREDNESS // EST. 2026
Build my kit ↗

LOCAL OUTLOOK / METHOD VERSION 1

Evidence before urgency.

What our weather-concern levels mean, where the data comes from, and what we cannot infer.

A weather indicator for preparation

We use National Weather Service forecasts and official point alerts to highlight conditions that can disrupt electricity delivery. Levels are our editorial interpretation of weather evidence. They are not NWS outage forecasts, probabilities, a utility status report, or a prediction of how long an outage will last.

A location represents a forecast grid cell. ZIP lookup uses an approximate representative area, not a street address. Preset places use public city-center coordinates. Both are rounded to two decimal places. Conditions and utility circuits can vary within a city. An alert applying to that point may not apply to every address.

How levels are assigned

We examine the next 24 or 72 hours from the current time. The strongest available indicator sets the level. We do not add points for product clicks, shopping choices or commission rates.

  • No elevated concern detected: no rule below is triggered, with fresh continuous wind-gust, sustained-wind and temperature coverage and a successful official alert check. This is not zero outage risk.
  • Keep an eye on conditions: gusts at least 35 mph, sustained winds at least 25 mph, forecast interval ice at least 0.01 inches, or temperature at least 100°F. Relevant official watches and advisories also qualify.
  • Prepare for possible disruption: gusts at least 45 mph, sustained winds at least 35 mph, forecast interval ice at least 0.10 inches, or temperature at least 110°F. Relevant high-wind, severe-thunderstorm, tropical-storm, winter-storm, blizzard, flood, storm-surge, extreme-heat and Red Flag warnings also qualify.
  • Serious weather concern: gusts at least 58 mph, sustained winds at least 40 mph, forecast interval ice at least 0.25 inches, or official hurricane, extreme-wind, tornado or ice-storm warnings.

These thresholds are transparent screening choices, not a calibrated outage model or universal utility operating limits. The 58 mph gust threshold aligns with the NWS severe-thunderstorm wind criterion. Local warning criteria vary. Heat can increase demand but does not establish a supply shortage. Forecast ice is the largest overlapping reported interval amount, not a sum for the whole window.

Overlapping gusts of at least 30 mph and relative humidity at or below 20% add a watch-level fire-weather indicator. Dry fuels, terrain and utility decisions are not modeled. A Red Flag Warning is not a power-shutoff notice. Check the utility's official PSPS information in California.

Only active, actual alerts overlapping the selected window are considered. Marine, air-quality, heat advisories and other official bulletins are displayed as applicable; only the documented power-related event rules affect the concern level. The time window spans qualifying indicators and does not imply continuous hazardous conditions.

Unknown is different from quiet

Forecasts older than six hours and assessment responses older than 20 minutes are excluded from concern levels. Separately, dated NWS weather values may be displayed for up to 24 hours after issue when they cover upcoming periods; these are labeled as an earlier forecast and never supply a quiet concern assessment. Older or missing values remain unavailable. Failed alert checks and incomplete essential forecast coverage cannot produce a quiet assessment. Fresh evidence may still show an elevated concern with an explicit incomplete-data label. Ice and humidity availability are disclosed separately. Short, localized thunderstorms and microclimates can be missed by the forecast grid.

California county counts come from the public-domain California OES Power Outages by County layer, which draws from PG&E, SCE, SDG&E and SMUD maps. We show counts only when the state layer reports an update within 45 minutes; missing or stale data never means zero. The display separates planned and unplanned affected-customer counts. The source groups customers by the incident's location; affected customers can cross county lines, so these are not counts of county residents without power. Unplanned does not mean weather-caused, and a county report cannot establish that an address is affected. Selected active incident reports may show a utility-reported start time and estimated restoration time. The start field means when an outage was reported, not necessarily when power failed at an address; estimates can change. Incident detail failures do not erase fresh county totals. These reports do not confirm power status at a home, identify the customer's utility, or establish an address-level outage start time. Maryland's public-domain county dataset reports customers without power, customers tracked by participating utilities, and their percentage. It uses a floating local timestamp, interpreted as Eastern time; the site withholds incomplete or stale editions. The state application says data arrives every 15–30 minutes and does not include customers in Hagerstown. The percentage describes tracked customers in a county at a reported time, not the future chance of an outage or status at an address. Across the 50 states and DC, a separate 2025 EIA-861 county service-territory list offers possible delivery utilities where the EIA name matches Census county geography; counties can contain several utilities and the list cannot identify a provider for a household. Connecticut uses the Census 2020 county geography because EIA still reports the former eight counties; some Alaska legacy names and one dissolved Virginia city do not map to 2025 counties. The server uses the Census coordinate geocoder to identify the county from rounded coordinates and caches successful results for a day. This list is annual inventory, not live outage status. For all areas, use the utility's official service for address-level status and restoration estimates. Equipment failures, maintenance, accidents and other non-weather causes are not predicted.

Updates with restraint

Automatic checks run every 10 minutes while the page is visible unless paused. The selected area remains only in this open page. The site has no browser notifications, push alerts, SMS, email subscription or saved visitor profile. Use official local warning channels.

What historical validation would add

ORNL's historical EAGLE-I county outage records and NOAA storm records can support future research. Version 1 is not trained on them and has no measured outage-prediction accuracy. Before publishing probabilities, we would define the affected-area threshold and forecast horizon, test on held-out storms using information available before each event, assess false alarms and missed events, and check calibration separately by region and weather type. County outage records cannot establish individual household outcomes.

Free sources, traceable use

No paid API or proprietary modeling dependency is used by this feature. Free source data does not guarantee unlimited hosting capacity or continuous upstream availability. Existing global tools list their own sources and licences.

Method reviewed October 12, 2026. Independent interpretation; no government or utility endorsement.