Data Assimilation while running jobs in a chain – in #12: CCLM Starter Package Support

in #12: CCLM Starter Package Support

Dear colleagues,

I am trying to perform an assimilation of GPS data with the CCL m starter package for a 6 month simulation.
Since the size of the assimilated data is too big to be allocated in a normal, single simulation of a 6 month period I am going to split the simulation into several one-week runs.
To this end I will use the subchain scripts provided within the starter package. Since I am new to these scripts I would like to have your advice on the following points.

- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.

- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?

Thank you for your help and answers

Regards

Alberto

  @albertocaldas-alvarez in #02df59a

Dear colleagues,

I am trying to perform an assimilation of GPS data with the CCL m starter package for a 6 month simulation.
Since the size of the assimilated data is too big to be allocated in a normal, single simulation of a 6 month period I am going to split the simulation into several one-week runs.
To this end I will use the subchain scripts provided within the starter package. Since I am new to these scripts I would like to have your advice on the following points.

- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.

- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?

Thank you for your help and answers

Regards

Alberto

Data Assimilation while running jobs in a chain

Dear colleagues,

I am trying to perform an assimilation of GPS data with the CCL m starter package for a 6 month simulation.
Since the size of the assimilated data is too big to be allocated in a normal, single simulation of a 6 month period I am going to split the simulation into several one-week runs.
To this end I will use the subchain scripts provided within the starter package. Since I am new to these scripts I would like to have your advice on the following points.

- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.

- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?

Thank you for your help and answers

Regards

Alberto

View in channel
”- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.”

It does not matter in which folders the source codes of cclm, int2lm and cfu are installed. Just make sure the paths to the executables are set in the run scripts correctly.

”- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?”

You need to put these lines in the beginning of the cclm.job.tmpl script file, since this is the only script that can be called in a higher frequency than monthly (i.e. the INC_DATE setting does only effect this script). All other scripts are called monthly.

  @burkhardtrockel in #ddf13f5

”- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.”

It does not matter in which folders the source codes of cclm, int2lm and cfu are installed. Just make sure the paths to the executables are set in the run scripts correctly.

”- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?”

You need to put these lines in the beginning of the cclm.job.tmpl script file, since this is the only script that can be called in a higher frequency than monthly (i.e. the INC_DATE setting does only effect this script). All other scripts are called monthly.

”- When jobs are run in a chain which executable and runscripts of the int2lm and cclm are being called at each time? In my computer I have installed in three separate folders the int2lm, the cclm and the chain routines (cfu). My question is whether the subchain scripts access the gcm2clm or the cclm_run scripts from the previously installed int2lm and cclm folders. I ask it since, given the fact that I am using the nudging scheme, nudging settings have to be defined in the original cclm scripts, and provided that the subchain routines do not interfere with the nudging settings I need to know if the right runscripts and executables are being called.”

It does not matter in which folders the source codes of cclm, int2lm and cfu are installed. Just make sure the paths to the executables are set in the run scripts correctly.

”- For my assimilation an ASCII file named ‘gps’ is to be placed in the same group of folders that contains the cclm runscript and its executable (cclm.exe). The assimilation code is written in such a way that only files named ‘gps’ are read for this type of nudging. Since my simulation will be split now into weekly jobs, I need to exchange the previous ‘gps’ file for that of the next simulated week. Could you suggest me where in the subchain scripts I could add extra code lines to delete the older ‘gps’ ASCII file and replace it for the new observation input file? For example, once the restart files of the former week have been written?”

You need to put these lines in the beginning of the cclm.job.tmpl script file, since this is the only script that can be called in a higher frequency than monthly (i.e. the INC_DATE setting does only effect this script). All other scripts are called monthly.