lbirch@whrc_org
New Member
Hi All,I had significant problems getting restarts to read the correct time so I thought I would post my solution, as it was not obvious to me what all is necessary to do restart in clm. Set Up: Transiet historical simulation starting in 1850 with atmosphere cycled from 1901 to 1920.Problem: After the simulation reached 1900, I wanted to stop cycling and read forcing data from 1901 to 2010. I did this by restarting the model in 1901 and updating the stream files to contain the years 1901-2010, but the model just stopped reading data when it reached 1920. No error message was thrown, but the output was obviously wrong (January temperature and GPP pattern in August).Basically I wanted to change align, start, and end from 1901,1901,1920 to 1901,1901,2010, but the newly created stream files were not being read.Solution: Restarts create a number of files and *datm.rs1.yyyy-mm-dd-0000.bin is not necessary for a restart. It just hard codes the stream file dates, overwriting stream files and env_run.xml. If you move *datm.rs1.yyyy-mm-dd-0000.bin out of the run directory, the simulation correctly reads the forcing data prescribed in the stream files and env_run.xml.I think the point of this *datm.rs1.yyyy-mm-dd-0000.bin is to speed up reading data upon a restart, but it is a problem if you want to change your stream file at some point in your simulation. Both clm4.5 and clm5 have this issue.If anyone disagrees with me about the importance of *datm.rs1.yyyy-mm-dd-0000.bin, please let me know because as far as I can tell, my simulations have the correct values now that I remove this file when I stop cycling the data. Best,Leah