The forecast combines fire and weather inputs
HRRR-Smoke starts with estimates of fire emissions, then uses a high-resolution weather model to predict transport, plume rise, mixing, deposition, and near-surface concentration. A forecast can change when a new satellite fire detection appears, an emission estimate is revised, or the weather analysis changes. The result is a physically based model field, not a simple extension of the current map.
Near-surface smoke in Smoke Window is primary fire-emitted fine-particle mass near the lowest model level. It does not include every source of ambient PM2.5. Dust, urban pollution, secondary particle formation, and other pollutants can affect observed air quality without appearing in the smoke field.
Read a run as a sequence
Smoke Window waits for a complete extended HRRR run and reads hours 0 through 48 at three-hour intervals. The opening answer is withheld until all 17 frames are verified, because an incomplete sequence could miss a later smoky period or falsely shorten a lower-smoke window. Once complete, the bar timeline and map use the same selected forecast hour.
Compare adjacent frames rather than anchoring on a single timestamp. If several samples show the same broad pattern, confidence in the trend is more useful than confidence in one exact value. Near plume edges or complex terrain, small changes in model wind can move a band across the selected cell.
Turn guidance into a practical choice
For flexible plans, look for a block of lower modeled-smoke samples and then check the nearby AirNow observation shortly before leaving. The model can help compare a morning walk with an afternoon hike, but it cannot certify that either time is healthy for a particular person.
When conditions are rapidly changing, reloading the page checks NOAA for the newest complete model cycle and refreshes AirNow and fire reports. A newer run may revise the answer. For health-sensitive decisions, official advisories and local guidance take precedence over a model-planning tool.
Why Smoke Window waits for a complete run
A forecast that has downloaded only the first few frames can describe the immediate map but cannot support a 48-hour conclusion. A later smoky frame might shorten an apparent window, and a missing lower frame might hide an improvement. Smoke Window therefore marks partial bars as provisional and withholds its headline until hours 0 through 48 are all verified. The selected current frame downloads first so the map becomes useful while the chronological sequence continues in the background.
The role of forecast lead time
Hour 3 begins from a much newer atmospheric state than hour 48. Farther-ahead values depend on a longer chain of predicted wind, mixing, plume rise, precipitation, and fire behavior. That does not make the second day useless, but it changes how it should be read. Near-term frames can guide a specific outing; later frames are better for identifying a broad tendency and deciding when to check again. Exact transition times deserve less confidence as lead time increases.
What the forecast leaves out
The near-surface MASSDEN record used here represents primary fine particles emitted by fires. It is not a forecast of every component contributing to ambient PM2.5 and says nothing directly about ozone or coarse particles. Smoke above the lowest model layer may also create visible haze without a matching surface band. The app avoids filling those gaps with an AQI conversion, because doing so would combine unlike quantities and imply health meaning the model does not provide.
Checking a changing fire day
A rapidly growing incident can change emissions between satellite looks, and clouds or dense smoke may obscure fire detections. A new HRRR cycle can therefore revise both where smoke begins and how the atmosphere moves it. Refreshing the page checks for the newest fully completed extended cycle rather than adopting a partial run. If the answer changes, compare run times and patterns; the revision is new guidance, not proof that an earlier model was deceptive.
A forecast checklist for outdoor plans
Choose the exact activity location, inspect several adjacent frames, and note whether the point lies near a sharp plume edge. Confirm that the 17-frame sequence is complete. Then read the nearby preliminary AirNow NowCast AQI, including reporting-area distance and primary pollutant. Recheck shortly before leaving and follow official advisories. This workflow uses model timing without pretending that a lower smoke-specific value guarantees healthy air for a particular person or activity.
Bottom line: wildfire smoke forecast
The best use of a smoke forecast is comparison: one time against another, one new run against its predecessor, and the model against current observed context. It is not a certificate for an activity. Keep the forecast location specific, favor stable multi-frame trends, and refresh as the plan approaches. That discipline turns uncertain guidance into a practical planning aid without overstating what the atmospheric model knows.