The STILT inverse modeling approach by Lin et al. (2003) introduced several significant changes that can also be applied to most standard forward dispersion calculations. These included several items noted previously, such as the Hanna turbulence parameterization (KBLT=5), the modified Richardson number for the mixed layer depth (KMIXD=3), and the variable Lagrangian time scale (VSCALES=1.0). Two additional options will be reviewed in this section, one enhances the dispersion computation and the other modifies how the input meteorological data are interpolated to the HYSPLIT internal coordinate system.
 To insure all options are at their default values, first Reset, then retrieve conc_case_control.txt and conc_case_setup.txt. We will continue using the aircraft sampling case for these calculations.
 One of the core components of STILT is insure massconsistency and that wellmixed layers stay well mixed. The STILT mixing scheme can be selected by opening Configuration Setup / Concentration / ParticlePuff Release Menu #3. This STILT mixing scheme sets up several additional computations:
 an internal forwardbackward transport scheme to correct for violations of mass consistency in the meteorological fields
 the definition of additional layers near top of PBL to reduce particle trapping in that stable environment
 a probability scheme for particle reflection/transmission across interfaces with step changes in turbulence
 a finer internal time step, 10% of T_{L} to reduce the errors introduced by operator splitting
Although the peak concentration is slightly closer to the measured peak value, overall the STILT mixing result is comparable to the original calculation using HYSPLIT mixing.
 The second option is the interpolation of the meteorological data to the internal HYSPLIT coordinate system. However, this alternative interpolation option only works with WRF meteorological data fields that contain various mass flux variables. Continue using the STILT mixing scheme. Therefore, the first step is to rerun the base calculation using wrf27mc rather than the default wrf27uw. Go to the Setup Run menu and select the masscoupled WRF data file. In this context, masscoupled means that the velocity fields are products of air density and velocity averaged over the output period. Therefore, the computational results reflect the differences between timeaveraged winds and instantaneous winds.
 The WRF interpolation scheme can be selected by opening Configuration Setup / Concentration / WRF Vertical Interpolation Menu #13. The WRF vertical interpolation scheme follows the coordinate transformations within the WRF model: HYSPLIT calculates the change in height using the hypsometric equation while the WRF the data on half and full ETA coordinates are interpolated to the HYSPLIT terrain following sigma levels using dry mass and density. Although it shows a slightly lower peak concentration, the final computation result is not much different than the result using the HYSPLIT interpolation.
Although for this particular example the more complex STILT mixing and interpolation procedures did not show dramatic differences from the standard base calculation, other meteorological situations may be more sensitive to these computational options. However, more important than determining which is the best turbulence or computational method to use, multiple options permit the development of a robust ensemble. Ensembles will be addressed in a later section.
