Hello,
I'm trying to compare HYSPLIT simulation concentration results (Horizontal Gaussian with vertical particle motion) to satellite-inferred concentration data. As the satellite data is integrated over the total column (according to documentation approx. 35-40km), the dispersion model is set at the same top of model.
Here is where my question comes in:
As I'm using the GDAS meteorological dataset, the lowest pressure level is 20 hPa, which corresponds to about 24km. How is HYSPLIT calculating the (average) column concentration above 25km? On a particle faction basis (pollutant to air, molar) , the results of maximum concentration difference is ~40% when comparing dispersion simulations conducted at 25 vs 35km.
From my understanding, the amount (moles) of air should not be contributing that much to the average concentration.
Thank you in advance.
Top of Model Setting
Re: Top of Model Setting
In the CONTROL file (or through GUI) you set the concentration grid.
https://www.ready.noaa.gov/hysplitusersguide/S313.htm
I assume then that you are setting the vertical levels in the concentration grid up to 40 km?
Since the GDAS does not give model winds above 20 hPa, then HYSPLIT will extrapolate the value of the winds above that level.
If you have computational particles in this region then you will see a message "Start height in extrapolation region..."
or in the MESSAGE file "NOTICE prfecm: extrapolation from level"
The way HYSPLIT calculates concentrations above the 20hPa level is the same way it calculates concentrations everywhere else,
https://www.ready.noaa.gov/documents/Tu ... _eqns.html
Air concentrations are computed by summing the contribution of each 3D particle's mass (m) each time step that the particle resides in the grid cell and dividing the sum by the cell's volume, where x, y, and z define the grid cell's dimensions.
I'm not sure I am following your process.
when you say " the results of maximum concentration difference is ~40% when comparing dispersion simulations conducted at 25 vs 35km. From my understanding, the amount (moles) of air should not be contributing that much to the average concentration."
Are you saying that in one simulation you use a concentration grid up to 25 km and in one you use concentration grid up to 35 km?
Then you compare the maximum concentration from the two simulations? Are you keeping vertical resolution of the grid the same -
e.g. one grid has levels 15,25,35 and the other has 15, 25? Or are you changing it e.g. one grid has levels 0,25 and one has levels 0,35?
Maximum concentrations could be changing for a variety of reasons here which depend on how you have changed your concentration grid from one simulation to the other as well as how the mass is initialized and then transported in the simulation.
For instance if your maximum concentrations are occurring above 25 km then you won't see them if you only have concentration grid defined up to 25 km. If you make your concentration grid coarser then you would be calculating average concentration over a larger volume which could cause average concentration to be smaller.
https://www.ready.noaa.gov/hysplitusersguide/S313.htm
I assume then that you are setting the vertical levels in the concentration grid up to 40 km?
Since the GDAS does not give model winds above 20 hPa, then HYSPLIT will extrapolate the value of the winds above that level.
If you have computational particles in this region then you will see a message "Start height in extrapolation region..."
or in the MESSAGE file "NOTICE prfecm: extrapolation from level"
The way HYSPLIT calculates concentrations above the 20hPa level is the same way it calculates concentrations everywhere else,
https://www.ready.noaa.gov/documents/Tu ... _eqns.html
Air concentrations are computed by summing the contribution of each 3D particle's mass (m) each time step that the particle resides in the grid cell and dividing the sum by the cell's volume, where x, y, and z define the grid cell's dimensions.
I'm not sure I am following your process.
when you say " the results of maximum concentration difference is ~40% when comparing dispersion simulations conducted at 25 vs 35km. From my understanding, the amount (moles) of air should not be contributing that much to the average concentration."
Are you saying that in one simulation you use a concentration grid up to 25 km and in one you use concentration grid up to 35 km?
Then you compare the maximum concentration from the two simulations? Are you keeping vertical resolution of the grid the same -
e.g. one grid has levels 15,25,35 and the other has 15, 25? Or are you changing it e.g. one grid has levels 0,25 and one has levels 0,35?
Maximum concentrations could be changing for a variety of reasons here which depend on how you have changed your concentration grid from one simulation to the other as well as how the mass is initialized and then transported in the simulation.
For instance if your maximum concentrations are occurring above 25 km then you won't see them if you only have concentration grid defined up to 25 km. If you make your concentration grid coarser then you would be calculating average concentration over a larger volume which could cause average concentration to be smaller.