We want to use HYSPLIT to model both a Chlorine and an Ammonia spill. We
have looked all over the internet for the Henry's Actual and Effective constants,
as well as the surface reactivity and diffusivity ratios for both Chlorine
and Ammonia. Do you know where we can find this information?
Using a model -- e.g., HYSPLIT -- to model a new compound is generally a complicated
task. One must understand that physics and chemistry of the compound well enough
to determine: (a) if the model in question can be made to work for the compound;
(b) if any modifications to algorithms must be made; and (c) values for key model
parameters.
I have adapted the HYSPLIT model for dioxin, atrazine and mercury and in each
case, substantial model development work was required. It was not simply a case
of picking a few parameters.
Nevertheless, it can be instructive to start with a 0th order approach and begin
getting a feel for the results.
The parameters you are asking about all refer to the dry (HL, diff, surf react)
and wet (HL) deposition algorithms.
First, for dry deposition, you could simply specify an average dry deposition
velocity. If you want to use the resistance-based formulation, you need HL, diffusivity,
and surface reactivity.
The first piece of advice I can give you is to conduct a series of bounding or
sensitivity analyses. For each parameter you are unsure of, conduct simulations
with a variety of values and see what differences there are in the results. In
some cases, there will be little or no difference and so you won't have to worry
so much about the value.
Effective Henry's Law coefficients are not the pure HL coeff., but reflect the
fact that the compound can dissociate in solution and appear to have a higher
HL coeff than pure equilibrium alone. Therefore, the effective HL coeff. will
be pH dependent, for example, for compounds that dissociate in water. This factor
will be important for both NH3 and Cl2 as they are very "active" in solution,
i.e., they participate in a number of association and dissociation processes.
I once found a value of the pure HL coefficient for Cl2 to be 0.076 molar/atm,
but this should be checked. You will have to do a literature search and should
be able to find this parameter for Cl2 and for NH3. However, as mentioned above,
the pure HL coeff. will generally be lower than the effective HL coefficient
-- sometimes orders of magnitude lower.
One reference I have gives the following value for the pure HL coeff of NH3 (Ammonia,
National Academy of Sciences, University Park Press, Baltimore, 1979):
log(10) HL = 1477.8/T - 1.6937 (where T is in degree's Kelvin)
The units are not given... but I think they may be molar/atm (not sure). That
reference also gives info regarding dissociation, etc, and you could estimate
the effective HL for this data.
For diffusivity in air, you can just use a guess-timate of the diffusivity of
some other small molecule, or, you can try to find the value in the literature.
Alternatively, there is a book giving methods to estimate this -- "Handbook of
Chemical Property Estimation Methods", W. Lyman, W. Reehl, and D. Rosenblatt,
Amerian Chemical Society, Washington DC 1990. I don't think the simulation will
be overly sensitive to this parameter as long as you get in the right ballpark.
For surface reactivity, this is always a very uncertain parameter. I'd try sensitivity
analyses with values of 0, 0.1, and 1 and see if there is a big difference.
Dr. Mark Cohen
http://www.arl.noaa.gov/ss/transport/cohen.html