Dust Storm Simulations

Previous

HOME
Next

HYSPLIT contains two different wind-blown dust emission algorithms. The original procedure is available through the GUI, while the newest procedure must be configured manually. For both versions, the dust emission algorithm is invoked by setting the namelist parameter ICHEM=3 in the file SETUP.CFG. In both procedures, potential dust emission points are defined in the input CONTROL file by the two additional fields of threshold friction velocity and area associated with each dust emission location.

  1. The original dust emission algorithm (an option not currently not available) was based upon describing the vertical dust flux as a fraction of the horizontal transport:
    • Q = K ρ g-1 U* (U*2-U*t2)

    where the soil emission factor K and threshold friction velocity U*t varied by soil type and land use. This detailed approach proved to be impractical in locations where these details are unknown, hence a simplified method was applied:
    • Q = 0.01 U*4

    when the threshold friction velocity field is set to -1.0 and the area field represents the dust emission area (m2) centered about that point. Dust emissions would occur at that location when the "hardwired" threshold friction velocity (28 cm/s) is exceeded. The use of the GUI option Special Runs / Dust-storm to set three data points (bottom left, upper right, resolution) for dust emissions, just invokes a pre-processor to create a CONTROL file for all the desert area land-use locations within that domain. The CONTROL file is configured for this fixed threshold dust emission algorithm. The GUI procedure cannot be used in the continental U.S. because there are no desert locations covering a full one-degree grid point.

  2. Before proceeding, if this tutorial is being run through the web rather from a CD, it will be necessary to download the meteorological data file NAM20100330.bin for this example. Any pre-configured CONTROL files used in this example will need to be edited to point to the location of the downloaded data file.

  3. To demonstrate the dust-storm computation, we will focus on an event that occurred in Utah from 30-31 March 2010. From a satellite derived data base of dust sources, we have already identified 21 locations (0.25 degree grid) in the Sevier Lake region of Utah that are frequent dust sources to the Salt Lake City region. Open the Setup Run and retrieve the pre-configured control file CONTROL_dust1.txt from the \Tutorial\dust directory. Note that the 21 potential dust locations contain two additional fields, the -1 indicating the use of the first algorithm and area value representing a 0.25 degree square emission area. All the other input values should already be set. Note that in the Concentration Grid menu the sampling start time is set for 00 00 00 06 00, which means that sampling will start 6 hours after the start of the simulation, for a 24-hour duration sample. The model results wil be compared with the daily averaged (local midnight to midnight) AirNow measurements. Then save to exit.

  4. Next open the Advanced / Concentration / menu #10 and check the Enable PM10 dust storm emission algorithm radio-button. Save to exit and then open menu #4 to increase the particle emission rate to 100000 with a maximum of 50000. These changes can also be invoked by retrieving the SETUP_dust.txt file. Save the changes to exit.

  5. Now go ahead and press the Run Model menu tab. The standard output message will indicate that the PM10 dust emission algorithm has been enabled. Note we are just doing a standard run and not through the Special Runs / Dust-storm menu which configures the control file. On your own try this option and you will see that it doesn't work because there are no desert-only land-use cells in the HYSPLIT data base. When the run completes, open the concentration display menu and set the units conversion to 1.0E+06 to convert grams/m3 to ug/m3. Also set the contour interval to fixed values of 200 100 50 20 10, then execute the display to show the dust plume.

  6. The calculation can be compared to the AirNow measurements in the Salt Lake City area for that day. Open the Utilities / Convert to DATEM menu and set the measured data input file to AirNow_dust.txt and with the conversion factor of 1.0E+06. The statistical results show under-prediction by about a factor of 10 and a correlation coefficient of 0.30. A performance level confirmed by the scatter diagram.

  7. In the revised dust emission algorithm, the threshold friction velocity and dust emission area vary for each location and dust locations are predefined by month based upon a climatology of dust sources developed from satellite data. In this approach, the dust emissions are defined by:
    • Q = K A (U*-U*t)

    where the positive value in the threshold friction velocity field (cm/s) triggers the use of the new algorithm. The second field is the product of the soil-dust density (K) and emission area (A).

  8. To use the alternate emission algorithm, open the Setup Run and retrieve the pre-configured control file CONTROL_dust2.txt from the \Tutorial\dust directory. Note that the 21 potential dust locations are the same but the two additional fields now show a varying threshold friction velocity and the product of the dust density and emission area (KA). This approach provides more user control of the emission configuration. Save the changes and run the model. Once it completes go right to the Convert to DATEM step. The model results show a higher correlation (0.51), but the model under-prediction has increased to about a factor of 400!

  9. There are a lot of uncertainties in these calculations, especially when more detailed soil and land-use information is required. In this case, the climatological values for U* and KA may not be representative for 2010, perhaps because of flooding in previous years built up erodible sediment that was not present during the period of the climatological analysis of the dust emission sources. A simple test of the sensitivity would be to increase the emission factor. We don't need to rerun the model, but can simply change the conversion factor in the Convert to DATEM menu to 4.0E+08, 400 times larger. The statistical results show a comparable correlation (0.45). The corresponding scatter diagram shows values for all the measurement stations in the Salt Lake City area over a wide range of air concentrations, suggesting that varying the emission factors by location can provide a method to improve model performance.

The dust storm emission calculations are very sensitive to the emission parameters and the module should not be treated as a black box. However, using the revised module with spatially varying characteristics, it may be possible to "calibrate" the model, refining the emission factors using measured data, and then using the calibrated model for predictions for a specific location. To configure the model for other locations or times, sample monthly control files, with only the source locations, are provided for the continental U.S. or all global points. The threshold velocities tabulated in the continental and global files were derived from different meteorological data sources.