HYSPLIT Boundary Layer Issue

[sgonzalezh]

Hello!!

I'm launching HYSPLIT trajectories from polar regions and I chose to launch the trajectories from 25%, 50% and 75% of the boundary layer height using the option KMSL=2. As it is well known, boundary layer in polar regions is very thin, especially during the winter.

My first question is how HYSPLIT calculates the boundary layer height? There is no much information about this topic nowhere, even in the user's guide.

Related with this question I have another issue. The user's guide states:

KMIX0 is a related parameter that sets the minimum mixing depth. The default value is 250 meters and is related to the
typical vertical resolution of the meteorological data. A resolution near the surface of 25 hPa is typical of pressure-level
data files. This suggests that it is impossible to infer a mixed layer depth of less than 250 m (10 m per hPa) for most
meteorological input data

I have the KMIX0 set as default 250 and I'm using the GDAS dataset, but I've found boundary layer depths less than 250 m, sometimes even only few meters. Qualitatively HYSPLIT performes well: it is winter... we have a strong inversion in the first few meters... But, why does it happens if in theory the mixing layer should be above 250 m? And again, how HYSPLIT calculates the boundary layer height? And what extent can we be confident in the boundary layer height calculated by HySPLIT?

Thank you so much!!!

[ariel.stein]

The simulated PBL affects the dispersion and the resulting air concentrations by changing the rate of dispersion as well as providing a cap to vertical mixing of pollutants. Currently the modeling system includes the following options for estimating the PBL height(zi): a) use the meteorological model (uses the zi value from the meteorological model), b) from the temperature profile (computes zi the height at which the potential temperature is at least two degrees greater than the minimum potential temperature). The computation is made from the top down , c) from the TKE profile (uses the meteorological model's TKE profile to estimate the zi at the height at which the TKE either decreases by a factor of two or falls to a value of less than 0.21), and d) set to a constant value (to match external observations the mixing depth can also be set to a constant value).

Here you have further information from the manual:
Mixed Layer Depth Computation
KMIXD is used to control how the boundary layer depth is computed. In addition as acting as a vertical lid to particle
dispersion (advection is not affected), the mixed layer depth is also used to scale the boundary layer mixing coefficients
and computing turbulent fluxes from wind and temperature profiles. The default is to use the value provided by the
meteorological model through the input data set. The profile can also be used if available. The computation defaults to
use temperature profiles if the mixed layer field is not available. Another option to use for testing is to replace the index
value (0,1,2) with a value greater than 10. In this situation, that value will be used as the mixed layer depth and will be
constant for the duration of the simulation.
0 = Use meteorological model MIXD if available (DEFAULT)
1 = Compute from the temperature profile
2 = Compute from the TKE profile
> = 10 use this value as a constant
KMIX0 is a related parameter that sets the minimum mixing depth. The default value is 250 meters and is related to the
typical vertical resolution of the meteorological data. A resolution near the surface of 25 hPa is typical of pressure-level
data files. This suggests that it is impossible to infer a mixed layer depth of less than 250 m (10 m per hPa) for most
meteorological input data.
250 = The minimum mixing depth (DEFAULT)

[ariel.stein]

If KMIX0 is set to the default your mixing layer will be at least 250 m, even though your meteorological dataset might indicate that its PBL height is lower than that.

[sgonzalezh]

Thank you Ariel.

This information will be very useful.

Sergi