An ensemble smoke dispersion forecast system for management of agricultural field burning

ClearSky, a decision support tool for management of smoke from agricultural burning, is used in both eastern Washington and northern Idaho. ClearSky predicts surface level PM_2.5 concentrations for MM5-forecast meteorology and user-proposed agricultural field burns. MM5 meteorological forecasts are reformatted using CALMET and CALPUFF is used to simulate the dispersion of PM_2.5 emissions calculated based on expected (or possible, or actual) field burning. CALPUFF, a Lagrangian puff model, generates hourly surface layer PM_2.5 concentration predictions throughout a receptor grid.

Previous ClearSky evaluations have shown that CALPUFF is extremely dependent upon accurate forecast meteorology to skillfully predict the transport and dispersion of plumes produced by field burns. Small differences in predicted versus actual winds (speed and direction) determine whether a smoke plume impacts a receptor as predicted, or not.

To address this source of error, the Ensemble ClearSky System (ECS) uses multiple (≤) different MM5 forecasts produced by the University of Washington Mesoscale Ensemble System. In the ECS, each field-burning scenario of PM_2.5 emissions is simulated in parallel through the set of available meteorological forecasts, to translate the uncertainty represented in the spread of forecast winds, into multiple PM_2.5 forecasts. The resulting forecasts of hourly PM_2.5 concentrations can be reviewed individually to observe the details of the different forecasts, or the results can be reviewed collectively to characterize the degree of spread, hence uncertainty, of the forecast winds. Also, the PM_2.5 results can be combined into an ensemble average or other probabilistic product. Select days during the 2004 burn season were simulated using the ensemble ClearSky system, and ensemble results were compared to single meteorological forecast driven ClearSky results to evaluate the value of the Ensemble ClearSky System.