[RegCNET] Two questions about modelling dust in RegCM4

[Solmon Fabien]

Hi Carolina

Carolina D C Cavazos Guerra wrote:
> Hi,
>
> I have used RegCM3 for dust modelling for the last couple of years so I
> might say I have certain experience. This time I tried to do some
> experiments using the newest RegCM4. I used the regcm.in.DUST included in
> the examples folder as template to configure my regcm.in in the run folder
> and used the same parameterization as in RegCM3. My first question comes
> from the fact that the suggested values for the chem param in the regcm.in
> for regcm4 is configured as follows:
>
>  &chemparam
>  idirect   =    2,
>  inpchtrname  =   'DUST','DUST','DUST','DUST',
>  inpchtrsol   =  0.10, 0.10, 0.10, 0.10
>  inpchtrdpv  
> =0.00000,0.00000,0.00000,0.00000,0.00000,0.00000,0.00000,0.00000,
>  inpdustbsiz(:,1)  = 0.01, 1.00, 2.50, 5.00
>  inpdustbsiz(:,2)= 1.00, 2.50, 5.00,20.00
>  /
>
> The parametres I have been using in regcm3 are:
> chtrsol   =  0.50, 0.60, 0.70, 0.80,
>  chtrdpv   = 0.00025,0.00050,0.00075,0.00100,0.00025,0.00050,0.00075,0.00100,
>  dustbsiz  =  0.01, 1.00, 2.50, 5.00, 1.00, 2.50, 5.00,20.00,
>
> As far as I remember the chtrsol and chtrdpv (inpchtrsol and inpchtrdpv in
> regcm4) correspond to wet and dry deposition, so my question is: which
> ones are the correct values?
>
>   

Dry deposition : it does not matter , since anyway there is a specific
sedimentation/deposition routine for dust.
Wet deposition : I don't think the values of RegCM3 example were
representative. Dust are not very hydrophilic.
The 0.1 parameter makes dust more hydrophobic so less sensitive to wet
deposition  ( I think I use 0.2 for the first bin, but maybe you can do
some sensitivity tests ) ... the only way to correctly assess wet
deposition would be to really validate deposition flux against data ..
We are working on scheme that takes also into account a better
representation of the rainout for different class of dust diameters (
and assuming raindrop distributions) , would you be interested to be
involved in these development since you are an experienced user ? .
Anyway , you can make some sensitivity test on your domain

> My second query regards about outputs for a dust experiment. I did run an
> experiment using the default values and everything appeared to be fine
> until I started to analyse the resulting products. I noticed that the
> parametres have still the same units as in regcm3 so I would expect if not
> equal outputs values with similar magnitudes. What I notice is that all
> values in certain parametres are unrealistically big (e.g. AOD eq 900)
> whereas in other parametres such as emiss or colb, values have an order of
> magnitude of around 10E2 or 10E3 bigger than in regcm3. So my question is,
> am I doing something wrong with the configuration or is this related to a
> bug in the dust emission scheme?
>
>   
That is more surprising .! . The official version of RegCM3 had many
issues I think ( except for some modified version that were circulating
around) so it is good that you upgrade to RegCM4 which has anyway a much
better management in term of maintenance of the code and bug
backtracking . 
RegCM4 has some changes in the dust emissions concerning the subgrid
treatment of emissions by different texture, the soil humidity
corerctions ( that never worked properly in RegCM3 official version, and
also some modification of soil aggregate distribution for texture class
2 and 3.. Now this last modif can be auite sensitive depending on your
domain. I did some test over the Sahara and I no problem of
overestimation of AOD , I had even slightly lower dust burdens etc.

So my questions are :
Which version of RegCM4 are you working with ( 4.0 I presume)
What is your domain and resolution ?
Do you use BATS or CLM ?
Did you compare instantaneous or averaged fields ?
What kind of convection scheme are you using ( it can be very sensitive
on emsission iof you use emmanuel over west africa for example)  ?
 
For info , very soon will be released a verion RegCM4.1 where dust work
also with CLM.

Cheers

Fabien





For information

> I would really appreciate any reply about these queries.Fo
>
> Regads
>
> Carolina
>
>   
>> <((((>`..`..`...><((((>`..`..`...><((((>
>>     
> Carolina Cavazos Guerra
> Department of Geography
> University College London
> Pearsons Building, Gower Street
> London. WC1E 6bt, UK
>
>
> _______________________________________________
> RegCNET mailing list
> RegCNET at lists.ictp.it
> https://lists.ictp.it/mailman/listinfo/regcnet
>