Research forecast · 12 km WRF-CMAQ

California air quality, next 72 hours

Hourly ozone, PM2.5 and the full six-pollutant EPA Air Quality Index for every incorporated city in California, from my own forecast pipeline on Northeastern's Explorer cluster.

Loading the current forecast…

Loading the forecast…

Spatial view

Where the pollution is, hour by hour

Click the map to jump to the nearest city.

How this is made

What you are looking at, and what it cannot tell you

The model chain

GFS forecast → WRF → MCIP → SMOKE with EPA's 2022v1 emissions platform → CMAQ, on a 12 km California grid, run daily on Northeastern's Explorer cluster.

Every value here is a volume average over a 144 km² grid cell, layer 1 of the model. Cities that share a cell share a forecast; street-scale gradients, near-road exposure and local sources smaller than the cell are not resolved.

The Air Quality Index

Computed on EPA's breakpoints (40 CFR 58 Appendix G, Table 2, 2024 revision) from all six criteria pollutants. The overall AQI is the maximum sub-index — that is the definition, not a shortcut — and the "driver" names whichever pollutant attains it. Hourly values use trailing windows: 8-hour for ozone and CO, 24-hour for PM2.5 and PM10, 1-hour for SO2 and NO2.

Weather fields, and why there is no UV

Temperature is the model's 2-m air temperature. Relative humidity is derived from that temperature together with the 2-m mixing ratio and surface pressure. Solar radiation is broadband downward shortwave at the ground.

“Rain nearby” is area coverage, not a chance of rain. It is the share of the area within 48 km of the city where the model has precipitation falling at that hour. A conventional probability of precipitation expresses forecast uncertainty and needs an ensemble of runs; this is a single deterministic forecast, so no honest probability can be derived from it. 100% here means “it is raining across the whole neighbourhood in this one forecast”, not “certain rain”.

There is no UV forecast here, deliberately. Neither WRF nor CMAQ outputs an ultraviolet field in this configuration, and photolysis diagnostics are switched off, so there is no actinic flux to weight either. UV-A+B is only about 3–5% of broadband shortwave and that fraction shifts with sun angle, ozone column and cloud, so scaling the radiation shown here would produce a number that looks like a UV Index without being one. For UV, use the EPA UV Index.

Limitations you should know about

    For health decisions, official advisories and current measurements, use AirNow and your local air district. This is a research forecast published to show what the model produces, not a public health service.