Forecasting Dust Storms using HYSPLIT
Roland R. Draxler
Prepared for the Sino-US Workshop on Dust Storms and Their Effects on Human Health, November 25-26, 2002, Raleigh, North Carolina
The model was initially tested over Kuwait, Iraq and Saudi Arabia, where it predicted about the right number of dust events (18%). The model results also agreed quantitatively with measurements at four locations in Saudi Arabia and one in Kuwait for one major dust event (>1000 µg/m3). However, for several smaller scale dust events (200-1000 µg/m3) the model substantially over-predicted the air concentrations. Part of the over-prediction was attributed to the model's sensitivity to the threshold friction velocity and the surface soil texture coefficient (the soil emission factor), and the difficulty in accurately representing these parameters in the model.
To apply the model over other domains where detailed digital soil characteristics are not available, the emission module was modified to use HYSPLIT's one-degree land-use file by assuming that a "desert" land-use grid cell corresponds to the Kuwait "active sand sheet" soil type category. To compensate for the greater number of potential dust emission cells, the original PM10 flux equation was replaced by a relationship not dependent upon soil type and which had a substantially lower emission flux.
The modified model was tested for two recent Chinese dust storms and one Australian event, the results of which are discussed in more detail in this presentation. The meteorological data for the examples and model executables are provided on a limited number of CD-ROMs for distribution to workshop participants.
Email me the full report /div>