CALPUFF

Quintero’s forecast is produced by CALPUFF, a Lagrangian puff dispersion model: each source emits a succession of puffs that the wind carries, turbulence widens, and chemistry and deposition consume, over real terrain and with weather that changes hour to hour and place to place. It is the model environmental authorities accept for complex domains — coasts, basins, industrial belts — where a straight-plume model will not do.

The chain

1. WRF — the weather model produces the forecast of wind, temperature and stability over the domain, from a global forecast. It is the expensive part: hours of compute per demo.
2. MMIF and CALMET — WRF’s weather is translated to the fine grid CALPUFF uses, with the domain’s terrain and land use.
3. CALPUFF — runs once per source and per scenario, and the results are summed at publication. Modelling per source is what makes the source group possible as a dimension: each one’s contribution exists separately before being summed.
4. Publication — concentrations every 15 minutes on the grid and at every receptor, the averaged series, the statistics and the alerts.

The configuration, in the open

CALPUFF is governed by dozens of switches — dispersion, chemistry, deposition, terrain treatment. The ones this demo uses are read on the dashboard, under Domain → CALPUFF: every switch with its effective value and where it came from — the platform’s default, the demo’s configuration, or a decision of this run — so that what is shown is exactly what ran.

The manual

In preparation

The CALPUFF user’s manual will be published in this section of the guide, to be read beside each demo’s configuration.

What the demo does not do

The demo is static: the 4.5 modelled days are presented as a running forecast, with a now that advances and repeats — see the forecast cycle. It does not re-run the model every day; that is what the operational AirDesk Forecast adds.

→ The forecast cycle · This demo