Couldn't use ESMF regridding tool due to out of memory

[tianyaobu@gmail_com]

To whom it may concern,I tried to run a single point simulation by using PTCLM in CLM 4.5 (CESM 1.2.0). There came a memory problem when I tried to use ESMF regridding tool in mkmapdata.sh. The program failed due to out of memory, when it started to work on the file "SCRIPgrid_3minx3min_LandScan2004_c120517.nc".I looked through ESMF memory related information in this forum. It seems 200 GB memory per node is required for running ESMF regridding tool, but the cluster I am working at only provides 64 GB memory per node. I understand due to memory limitation, ESMF regridding tool is only supported in selected machines with enough memory per node.My question is if I am not albe to use ESMF regridding tool, is there another way to produce the mapping data? Is there another tool I can use?Thank you very much in advance. Your input will be greatly appreciated.Tian

 

[yfan1@uni-goettingen_de]

Hi Tian,I had similar memory issue with mkmapdata.sh when it processes the HYDRO-1K topostats (10GB) and GLOBE-Gardner 3x3min glacier (3GB) dataset. They are the only data source for extracting glacies and topographic information. I tried to subset them to a smaller region to lower down the data size but didn't succeed because of the unique data structure that is different from normal netcdf files. I hope the NCAR scientists can give a solution to the memory problem or provide smaller subsets for different regions or other alternative sources. At last, I modified the codes in mkmapdata.sh a little bit so that I could put a constant value for slope for my case and let the script omit the glacier dataset, and finally created the mapping files.Yuanchao 

 

[tianyaobu@gmail_com]

Hi Yuanchao,
Thank you very much for the information:). It's very helpful.
Tian

 

[tianyaobu@gmail_com]

I would like to update the post. Students and researchers in US universities can request for a small allocation in Yellowstone, and it's a quick solution for this problem.