doubts regarding fine tunning, bulk forcings ....

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
mashinde
Posts: 135
Joined: Mon Jun 22, 2009 3:46 pm
Location: Indian Institute of Tropical Meteorology, Pune, INDIA

doubts regarding fine tunning, bulk forcings ....

#1 Unread post by mashinde »

Dear Friends,


(1) I tried ROMS with climatological forcing in the Mediterranean sea. And it works well after initial settings. The large scale surface circulation features seems to be comparable when validated with the MEDATLAS data with some biases. It further needs some fine tunning as the current model configuration use 1/16 deg resolution and 40 vertical levels. The fine tunning is required in TNU2,TNU4,VISC2,VISC4 for eddy resolved simulations. I tried to put small number of TNU2 and VISC2, but I still confused with this numbers. Is there any thumb rule to determine the values of above matrices so that the flow should be turbulent and to get a flavour of the Mediterranean eddies ?

(2) In second set of experiments, before starting, I want to smooth the bathymetry using "LP Bathymetry Method" (http://www.liga.ens.fr/~dutour/Bathymetry/index.html), as there are some problems in current simulations regarding the smoothing of the bathymetry. I am working on this. Is there any reliable smoothing technique for complex basin such as the Mediterranean sea? Is shapiro/laplacian filters are good one ?

(3) Which vertical mixing parametrization scheme is good? In current simulations, I applied KPP, but in some paper I read that the KPP is good for global ocean modelling. Any recommendations ? What about MY2.5 ?

(4) In addition to this I want to use bulk forcing from ERA-40. I also have doubts regarding this. They are as follow ....
(4.1) While using bulk forcing, should I need to change clim, ini and boundary files. This files are created from Levitus and COADS data sets. I just want to use bulk forcing to study the air-sea interactions and also for other process studies.

(4.2) In such case what will be the time stamps ? For example, I want to use daily bulk forcings from 01-01-1958 and to run the model for next five years. So, in this case what will be the value of NTIMES and also the time stamps ? The DT I am using is 200s.

(4.3) How can I force river runoff ?

(5) After this simulations, I want to couple ROMS with AGCM to make regional coupled system for further study. Is there any documentation on how to couple the ROMS with AGCM ?


Any suggestions/recommendations ???



Thanking you

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: doubts regarding fine tunning, bulk forcings ....

#2 Unread post by kate »

Good questions all, let's see what I can answer:
mashinde wrote:(1) I tried ROMS with climatological forcing in the Mediterranean sea. And it works well after initial settings. The large scale surface circulation features seems to be comparable when validated with the MEDATLAS data with some biases. It further needs some fine tunning as the current model configuration use 1/16 deg resolution and 40 vertical levels. The fine tunning is required in TNU2,TNU4,VISC2,VISC4 for eddy resolved simulations. I tried to put small number of TNU2 and VISC2, but I still confused with this numbers. Is there any thumb rule to determine the values of above matrices so that the flow should be turbulent and to get a flavour of the Mediterranean eddies ?
Our rule of thumb was always that TNU2 can be a factor of ten less than VISC2. When picking VISC2, think about the timescales of the eddies, the timescales of damping on the 2 dx spatial scale and the eddy spatial scale. You can probably damp the 2 dx on a one day timescale.
(2) In second set of experiments, before starting, I want to smooth the bathymetry using "LP Bathymetry Method" (http://www.liga.ens.fr/~dutour/Bathymetry/index.html), as there are some problems in current simulations regarding the smoothing of the bathymetry. I am working on this. Is there any reliable smoothing technique for complex basin such as the Mediterranean sea? Is shapiro/laplacian filters are good one ?
Bathy smoothing is a fine art. I have the most experience with an ancient Fortran code that smooths more where the r-value is large. Then I do a plain shapiro filter everywhere. Try it and see if it runs...
(3) Which vertical mixing parametrization scheme is good? In current simulations, I applied KPP, but in some paper I read that the KPP is good for global ocean modelling. Any recommendations ? What about MY2.5 ?
Again, try things. We use KPP in the Bering Sea because the GLS schemes don't get the thermocline right. We use GLS (k-kl) in another application because KPP was blowing up, as was k-omega.
(4) In addition to this I want to use bulk forcing from ERA-40. I also have doubts regarding this. They are as follow ....
(4.1) While using bulk forcing, should I need to change clim, ini and boundary files. This files are created from Levitus and COADS data sets. I just want to use bulk forcing to study the air-sea interactions and also for other process studies.
No need to change any of those files to go to bulk fluxes.
(4.2) In such case what will be the time stamps ? For example, I want to use daily bulk forcings from 01-01-1958 and to run the model for next five years. So, in this case what will be the value of NTIMES and also the time stamps ? The DT I am using is 200s.
You can do the math - steps per day at dt=200s times 5 years.
(4.3) How can I force river runoff ?
You need to create a point sources forcing file or else define ANA_PSOURCE and specify it analytically. Do you know what the river fluxes should be? Where they should be in your grid units?
(5) After this simulations, I want to couple ROMS with AGCM to make regional coupled system for further study. Is there any documentation on how to couple the ROMS with AGCM ?
I don't know AGCM, but coupling to NCAR's CESM has been quite a job. We run a global POP ocean and extract side boundary conditions from it for ROMS. Both oceans pass their SST to the atmosphere as a merged product. Get familiar with a mapping program called scrip. How familiar are you with AGCM? Are you starting with one-way off-line coupling?

mashinde
Posts: 135
Joined: Mon Jun 22, 2009 3:46 pm
Location: Indian Institute of Tropical Meteorology, Pune, INDIA

Re: doubts regarding fine tunning, bulk forcings ....

#3 Unread post by mashinde »

kate wrote:Good questions all, let's see what I can answer:

Thanks Kate for the valuable suggestions.
mashinde wrote:(1) I tried ROMS with climatological forcing in the Mediterranean sea. And it works well after initial settings. The large scale surface circulation features seems to be comparable when validated with the MEDATLAS data with some biases. It further needs some fine tunning as the current model configuration use 1/16 deg resolution and 40 vertical levels. The fine tunning is required in TNU2,TNU4,VISC2,VISC4 for eddy resolved simulations. I tried to put small number of TNU2 and VISC2, but I still confused with this numbers. Is there any thumb rule to determine the values of above matrices so that the flow should be turbulent and to get a flavour of the Mediterranean eddies ?
Our rule of thumb was always that TNU2 can be a factor of ten less than VISC2. When picking VISC2, think about the timescales of the eddies, the timescales of damping on the 2 dx spatial scale and the eddy spatial scale. You can probably damp the 2 dx on a one day timescale.

I used small number of TNU2 and VISC2 in current simulation and stored daily HIS and monthly AVG files. In some daily HIS file, model simulation shows some eddies in the basin. Unfortunately, no seasonal variation. Since my application is eddy resolving, there should be presence of eddies in the Mediterranean basin. That is why, I should choose the most appropriate values of these parameters. I am finding more reference and will try to use in my next simulations. I am still doing analysis with this data to encounter the problems in the model simulations.
(2) In second set of experiments, before starting, I want to smooth the bathymetry using "LP Bathymetry Method" (http://www.liga.ens.fr/~dutour/Bathymetry/index.html), as there are some problems in current simulations regarding the smoothing of the bathymetry. I am working on this. Is there any reliable smoothing technique for complex basin such as the Mediterranean sea? Is shapiro/laplacian filters are good one ?
Bathy smoothing is a fine art. I have the most experience with an ancient Fortran code that smooths more where the r-value is large. Then I do a plain shapiro filter everywhere. Try it and see if it runs...

Can you please tell me where I can get this ancient Fortran code ?
(3) Which vertical mixing parametrization scheme is good? In current simulations, I applied KPP, but in some paper I read that the KPP is good for global ocean modelling. Any recommendations ? What about MY2.5 ?
Again, try things. We use KPP in the Bering Sea because the GLS schemes don't get the thermocline right. We use GLS (k-kl) in another application because KPP was blowing up, as was k-omega.

I found one problem with this mixing scheme after computing MKE and TKE. I use KPP as a vertical mixing parametrization scheme. In case of MKE plots there always persists a strange patch of circulation in the Ionian Sea. I don't understand the occurrence of this patch. I speculate that this may happen due to KPP scheme. I have to try other schemes and look again.

(4) In addition to this I want to use bulk forcing from ERA-40. I also have doubts regarding this. They are as follow ....
(4.1) While using bulk forcing, should I need to change clim, ini and boundary files. This files are created from Levitus and COADS data sets. I just want to use bulk forcing to study the air-sea interactions and also for other process studies.
No need to change any of those files to go to bulk fluxes.
(4.2) In such case what will be the time stamps ? For example, I want to use daily bulk forcings from 01-01-1958 and to run the model for next five years. So, in this case what will be the value of NTIMES and also the time stamps ? The DT I am using is 200s.
You can do the math - steps per day at dt=200s times 5 years.

I understand the time stepping and date stamps now.
(4.3) How can I force river runoff ?
You need to create a point sources forcing file or else define ANA_PSOURCE and specify it analytically. Do you know what the river fluxes should be? Where they should be in your grid units?

I don't know about river fluxes. And also to define the location of points in grid units.
(5) After this simulations, I want to couple ROMS with AGCM to make regional coupled system for further study. Is there any documentation on how to couple the ROMS with AGCM ?
I don't know AGCM, but coupling to NCAR's CESM has been quite a job. We run a global POP ocean and extract side boundary conditions from it for ROMS. Both oceans pass their SST to the atmosphere as a merged product. Get familiar with a mapping program called scrip. How familiar are you with AGCM? Are you starting with one-way off-line coupling?
Yes, I am familiar with the AGCM. Where do I get mapping program scrip, I searched on the internet but didn't get it. I think it will be easy to use Oasis coupler.

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: doubts regarding fine tunning, bulk forcings ....

#4 Unread post by kate »

I used small number of TNU2 and VISC2 in current simulation and stored daily HIS and monthly AVG files. In some daily HIS file, model simulation shows some eddies in the basin. Unfortunately, no seasonal variation. Since my application is eddy resolving, there should be presence of eddies in the Mediterranean basin. That is why, I should choose the most appropriate values of these parameters. I am finding more reference and will try to use in my next simulations. I am still doing analysis with this data to encounter the problems in the model simulations.
Seasonal variability should be coming from your forcing.
Can you please tell me where I can get this ancient Fortran code ?
http://www.arsc.edu/~kate/grids/gridpak6.tar.gz
Look at the algorithm in bathsuds (user-defined smoothing).
I found one problem with this mixing scheme after computing MKE and TKE. I use KPP as a vertical mixing parametrization scheme. In case of MKE plots there always persists a strange patch of circulation in the Ionian Sea. I don't understand the occurrence of this patch. I speculate that this may happen due to KPP scheme. I have to try other schemes and look again.
I am no expert on the mixing stuff.
I don't know about river fluxes. And also to define the location of points in grid units.
You need a tool which tells you the i,j points on the grid, referenced to a map or the land mask or something. The matlab editmask is nice, or perhaps you can get it from the land mask in ncview. Then you need to create the sources netcdf file with the locations of the sources, the vertical structure, the flow. Getting flow data can be a challenge. Making netcdf files can be fun - I don't remember if I used NCL or python for the last one.
Yes, I am familiar with the AGCM. Where do I get mapping program scrip, I searched on the internet but didn't get it. I think it will be easy to use Oasis coupler.
http://climate.lanl.gov/Software/SCRIP/

Post Reply