17.3 Short-Range Dispersion Simulations

Previous

HOME


Next

The first phase of Project Sagebrush took place at ARL's Field Research Division's tracer release facility, located on the Department of Energy's Idaho National Laboratory (INL) during September and October of 2013. We will configure HYSPLIT to reproduce the results of the last release on the 18th of October. Sampling was conducted over 10 min intervals at multiple locations along arcs, the most distant of which was 1600 m downwind of the source location.

  1. After opening the GUI press, Reset and then open the Setup Run Menu. Enter the simulation start time of 13 10 18 19 with a run duration of 3 hours. Select the 1-km 5 min resolution WRF meteorological data file sage5_wrf01.bin in the sagebrush directory. Open the Starting Locations sub-menu and enter the location of the tracer release point 43.59066 -112.938 10.0. Then save to exit.

  2. Open the Concentration Grid menu and enter the center of the grid 43.59 -112.94, a grid resolution of about 100 m - 0.001 0.001 and span of about 20 km - 0.2 0.2. For many of the previous longer-range simulations, we used a vertical grid size of about 100 m. In this situation, we are so near the source that the plume is not expected to be well mixed, a smaller vertical dimension is required. Set the height of the level to 25 meters. Examine the sampling data for release #5 to determine the required start 13 10 18 20 00 and stop 13 10 18 22 00 times as well as the 10 min averaging time so that we can match the model calculations to the measurements. Save to exit the menu.

  3. Next open the Pollutant release menu to enter the text description of the sulfur hexafluoride tracer SF6E and the release rate of 3708 grams over 2.5 hours which started at 13 10 18 19 30. The model calculation starts at 1900 but no tracer particles are released until 1930 and the calculation continues for 3 hours until the end of the last measurement time. Save to exit the menu and then save to exit the Setup Run menu.

  4. In the Advanced / Configuration Setup / Concentration tab open Menu #4 for the particle release number and set the value to 50000 and the maximum particle number to something slightly larger such as 100000. The large release rate is required to insure a sufficient number of particles are sampled in the 25 x 100 meter grid cell over each 10 min period. Save to exit the menu.

  5. Previously when examining the measured data you should have noticed that the measurement units were ppt. The release rate is defined in grams so the default model output is g/m3 and therefore a units conversion is required. In the previous section on units conversion two approaches were discussed. A fixed conversion factor based upon a standard atmosphere or a dynamic approach where the conversion occurs during the calculation based upon ambient conditions. We will configure the model for the latter method by opening the In-Line Conversion Menu #10 and selecting the Divide output mass by air density radio-button. Save to close all menus and press the Run Model tab. The run takes a few minutes and the standard output should show that mass conversion was enabled.

  6. Once the run completes open the Utilities / Convert to / DATEM menu and select the measured data file sage5_meas.txt and provide a unique output file name such as sage5.txt for the model results at the measurement locations. Knowing the molecular weight of SF6 (32 + 6 x 19) we can compute the conversion factor (29/146 109) and enter the value of 1.986E+08 in the field to convert from g/kg to ppt. Note that if we were doing a standard atmosphere conversion of the default output g/m3 to ppt, the required entry would be 1.536E+08.

  7. Press the Create DATEM file button and then the Compute Statistics button to see that over all the 1363 samples the correlation of model and measurements was 0.80 which can also be illustrated by pressing the Scatter Plot button.

  8. Although the standard HYSPLIT plotting routines are not optimized for very short distances we can still get an approximate visualization of the plume by opening the Display / Concentration menu and selecting the rings check-button and setting the display for 5 1, that is five rings at 1 km intervals. This and 100% zoom will help optimize the display for short range applications. Set user contours at about factor of two intervals from 10000 to 10 and enter the measured data file sage5_meas.txt for display. Prior to creating the display, open the Map Borders Label menu and the mass units ppt, convert the volume field to blanks, and set the layer field to between. Now save and execute the display.

  9. The results for the first sampling time period clearly show the plume direction and the inadequacies of the display program at this scale. Right clicking on the map zooms the display further and although the measured concentrations are not readable, there are sections of overlap explaining the good statistical performance. However, there are areas of disagreement suggesting the underlying meteorological data wind directions could be further improved. Note that a higher resolution display can be achieved by using Google Earth. Before exiting the GUI save the control file to sage5_control.txt and the namelist file to sage5_setup.txt.

This example demonstrated how a standard HYSPLIT simulation could be applied to very short range dispersion simulations. Although the internal model integration time step can not be less than one minute, this internal time step only applies to the transport and dispersion integration with respect to the meteorological data. When the time step is at one minute, a special code section is invoked to linearly interpolate the particle positions to the concentration grid at less than one minute time steps depending upon the size of the concentration grid. Another option is to use puff rather than particle dispersion. Puffs are forced to intersect at least one concentration grid point and the concentrations are only sensitive to puff size rather than grid cell size.

External Meteorology Links:    sage5_wrf01.bin