Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

CLM4.5 - Own surface dataset creation

M

matthias_demuzere@ees_kuleuven_be

New Member
Dear all,I am currently tryingo to create high-resolution inputdata for CLM4.5, on approximately 1 km resolution for a regional domain centered above Melbourne (Australia), which includes ocean. This is done on our local Australian HPC system and thus not on supported machines.For this I am following the workflow here. Going through these steps I end up having some errors and general questions, which I will try to clarify below (indicated with bullet points). I hope all of this is clear and that someone will be able to point me in the good direction on how to create all the necessary input files for CLM4.5.Please note that only CLM4.5 input data is necessary since this model is coupled to another regional climate model. 1) Create SCRIP grid datasets:This I do in the following folder: cesm1_2_2/models/lnd/clm/tools/shared/mkmapdataWith the following statement: mknoocnmap.pl -p -37.85,145 -n 0.01x0.01_Melbourne -dx 0.01 -dy 0.01 -nx 300 -ny 300This results in three files: cesm1_2_2/models/lnd/clm/tools/shared/mkmapdata/map_0.01x0.01_Melbourne_noocean_to_0.01x0.01_Melbourne_nomask_aave_da_150428.nccesm1_2_2/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_0.01x0.01_Melbourne_nomask_c150428.nccesm1_2_2/models/lnd/clm/tools/shared/mkmapgrids/SCRIPgrid_0.01x0.01_Melbourne_noocean_c150428.nc
  • I assume that I should use SCRIPgrid_0.01x0.01_Melbourne_nomask_c150428.nc in the following step?
  • Since I have ocean in my domain, it seems I should do something else? Although the phrase "you could modify the mask in the SCRIP grid file for ocean, and then use ESMF_RegridWeightGen to create the mapping file, and gen_domain to create the domain file" is not clear to me.
 2) Domain datasetUnfortunately, this step is unclear to me:
  • In the folder cesm1_2_2/tools/mapping/gen_mapping_files/ I do not see a gen_domain script. Only gen_cesm_maps.sh and gen_ESMF_mapping_file/create_ESMF_map.sh.
  • If I have to use those, would it be possible to explain me what exactly they do, what input they need and what kind of output they produce?
 3) mksurfdata_mapHere, I altered models/lnd/clm/tools/shared/mkmapdata/mkmapdata.sh to make sure the libraries and folders are properly set for our system. Also all input data is available in a local inputdata/lnd/clm2/mappingdata/grids folder.As target grid, I provide the one created in step 1. The command to interactively use the script is as follows:mkmapdata.sh -f /PATHTO/SCRIPgrid_0.01x0.01_Melbourne_nomask_c150428.nc --res 0.01x0.01 -t regional -p clm4_5 -v This goes well for a few files:map_0.5x0.5_AVHRR_to_0.01x0.01_nomask_aave_da_c150428.ncmap_0.5x0.5_MODIS_to_0.01x0.01_nomask_aave_da_c150428.nc , and then I get an error (for full log, see attachment):./mkmapdata.sh: line 107: 11208 Segmentation fault      ${cmd}Error status returned from mkmapdata script
  • Here, I am not sure why this fails? Any ideas?
 4) mksurfdataSince I failed in creating the maps above, I did not yet had a chance to execute this script.Nevertheless, I see that mksurfdata_map.namelist holds a long list of raw input data files.
  • I was wondering what the most easy way is to get these data onto my local machine? For Step 3, I just used svn export from this link: https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/lnd/clm2/mappingdata/grids/.
  • But I really wonder whether this is the most efficient method, especially since the input data list is a lot longer for this step?
  • Also I was wondering what the all_urban      = .false. and no_inlandwet   = .true. mean? Would it be possible to provide some more information on this?
 5) More stepsAfter the above steps, should all input data be available for CLM4.5 (or 4.0). Or is there another step missing? 
  • Finally: Since all of these scripts (+ available inputdata) seem to be tailored for a number of specific machines (yellowstone, jaguar, ...), I was wondering whether an outside researcher can also get access to one of these machines to create the necessary user-defined input files for a model simulation (the latter being performed on a different platform). If not, then I hope someone could help me with the issues above.
 Thanks & kind regards.      
 
erik

erik

Erik Kluzek
CSEG and Liaisons
Staff member
Hi MathiasI've snipped some of your text, but have answers to your questions below...1) Create SCRIP grid datasets:[/i][/b]This I do in the following folder: cesm1_2_2/models/lnd/clm/tools/shared/mkmapdata[/i][/b]With the following statement: mknoocnmap.pl -p -37.85,145 -n 0.01x0.01_Melbourne -dx 0.01 -dy 0.01 -nx 300 -ny 300[/i][/b]…[/i][/b]·       I assume that I should use SCRIPgrid_0.01x0.01_Melbourne_nomask_c150428.nc in the following step?[/i][/b]Yes.Since I have ocean in my domain, it seems I should do something else? Although the phrase "you could modify the mask in the SCRIP grid file for ocean, and then use ESMF_RegridWeightGen to create the mapping file, and gen_domain to create the domain file" is not clear to me.[/i][/b][/list][/list] Really there are only two issues for which the ocean in your domain matters. 1.)   If the ocean is extensive enough, that you don’t want CLM to spend time running over ocean2.)   If you need to generate files to run an ocean model (even a data ocean model)The area over ocean will just be set as “wetland” for CLM. If you are only running CLM, this just means CLM is doing extra work. If the area is small, this probably doesn’t matter. And since using “mknoocnmap.pl” is so much easier, even if the ocean area is large, we recommend that you start this way, and maybe do something different if you get tired of wasting resources later.The other reason is if you are going to run an ocean model (even a data ocean model) over the ocean area, In this case you’ll need to define ocean areas, and fractions, as well as define mappings between ocean and atmospheric grids. This is of course mainly important if you are running an active atmosphere model, and need to have reasonable results over ocean as well as land-areas.2) Domain dataset[/i][/b]Unfortunately, this step is unclear to me:[/i][/b]·       In the folder cesm1_2_2/tools/mapping/gen_mapping_files/ I do not see a gen_domain script. Only gen_cesm_maps.sh and gen_ESMF_mapping_file/create_ESMF_map.sh.[/i][/b] [/i][/b]“gen_domain” is a FORTRAN program that you have to build in order for it to be created. The CLM UG page on it is at…http://www.cesm.ucar.edu/models/cesm1.2/clm/models/lnd/clm/doc/UsersGuide/x11812.htmlAlso read the README file in the tools/mapping/gen_domain directory for more information on how to build it…http://www.cesm.ucar.edu/models/cesm1.2/clm/tools/mapping/gen_domain_files/README·       If I have to use those, would it be possible to explain me what exactly they do, what input they need and what kind of output they produce?The page above explains that “gen_mapping_files/gen_cesm_maps.sh to create a mapping file between the atmosphere and ocean”. So this is important if you have significant ocean in your grid, and you are going to use some sort of ocean model for it. The “create_ESMF_map.sh” is a script to run the ESMF mapping program from ocean and land SCRIP grid files. 3) mksurfdata_map[/i][/b]mkmapdata.sh -f /PATHTO/SCRIPgrid_0.01x0.01_Melbourne_nomask_c150428.nc --res 0.01x0.01 -t regional -p clm4_5 -v [/b]This goes well for a few files:[/i][/b]map_0.5x0.5_AVHRR_to_0.01x0.01_nomask_aave_da_c150428.nc[/i][/b]map_0.5x0.5_MODIS_to_0.01x0.01_nomask_aave_da_c150428.nc[/i][/b] [/i][/b], and then I get an error (for full log, see attachment):[/i][/b]./mkmapdata.sh: line 107: 11208 Segmentation fault      ${cmd}[/b]Error status returned from mkmapdata script[/b]·           Here, I am not sure why this fails? Any ideas?[/i][/b] It fails when working on the 3x3 minute LandScan grid. I can’t tell anything beyond that.4) mksurfdata[/i][/b]·       I was wondering what the most easy way is to get these data onto my local machine? For Step 3, I just used svn export from this link: https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/lnd/clm2/mapping....[/i][/b]But I really wonder whether this is the most efficient method, especially since the input data list is a lot longer for this step?[/i][/b][/list]
See this example, on how to transfer to your local machine... http://www.cesm.ucar.edu/models/cesm1.2/clm/models/lnd/clm/doc/UsersGuide/x11868.html#getmksurfdata_map_datasets [/i][/b]·       Also I was wondering what the all_urban      = .false. and no_inlandwet   = .true. mean? Would it be possible to provide some more information on this?[/i][/b]The CLM UG talks about this on http://www.cesm.ucar.edu/models/cesm1.2/clm/models/lnd/clm/doc/UsersGuide/x11868.html Look just under Figure 2.6.5) More steps[/i][/b]After the above steps, should all input data be available for CLM4.5 (or 4.0). Or is there another step missing?[/i][/b] Yes. The complete process is on this page in the CLM UG.. http://www.cesm.ucar.edu/models/cesm1.2/clm/models/lnd/clm/doc/UsersGuide/x11573.html ·       Finally: Since all of these scripts (+ available inputdata) seem to be tailored for a number of specific machines (yellowstone, jaguar, ...), I was wondering whether an outside researcher can also get access to one of these machines to create the necessary user-defined input files for a model simulation (the latter being performed on a different platform). If not, then I hope someone could help me with the issues above.[/i][/b]  If you could get access to yellowstone, it would likely be easier for you. I know you can apply for access. But, probably you'd want to contact the Land Model Working Group LMWG Co-chairs and see if they'd be willing to work with you.
 
You must log in or register to reply here.
Top