CESM1.0: POP2 Port Validation
In a successful port, CESM POP2 ocean-model solutions are expected to be the same "to roundoff level" as solutions generated on a trusted machine, although determining what is and what is not roundoff level is somewhat difficult and subjective. Presently no automated testing tools exist for validating the successful port of the CESM POP2 model to a new system.
In order to facilitate the evaluation of a successful port, we suggest that you set up and run on your local machine the cases described in the Port Validation Procedure document below.
When time permits, we plan to create scripts to automatically set up and run those cases, but until then, you will need to apply these steps manually.
Following the successful execution of these cases, you can use the ncl SSH RMS difference script to create plots and compare them with the results linked below in the "Port-Validation Plots" section in order to validate your port.
The following plots illustrate a successful port to the NCCS Cray XT jaguar. The plots display timeseries of RMS differences of the SSH field between the jaguar solutions and those generated on a "trusted machine," the NCAR IBM bluefire.
Note that the time axis on the following plots is in the range [366.0,370.], because the cases start from year 1; thus the plots show results for the first five days of a startup run.
The following plots were produced on 21 October 2010 with a cesm1_0 tag. Prior to this date, the plots shown on this page were produced with the ccsm4_0_rel04 tag and were not valid for the cesm1 tag.
- CASE1 bluefire vs jaguar
- CASE2 bluefire vs jaguar
- CASE3 bluefire vs jaguar
- CASE4 bluefire vs jaguar
- CASE5 bluefire vs jaguar
The following plots illustrate acceptable RMS SSH differences in POP2 solutions on trusted machines (bluefire and jaguar) due to differences in pe count and convergence criteria.
- CASE1 vs CASE2 bluefire
- CASE1 vs CASE3 bluefire
- CASE4 vs CASE5 bluefire
- CASE1 vs CASE2 jaguar
- CASE1 vs CASE3 jaguar
- CASE4 vs CASE5 jaguar
Port validation datasets generated on bluefire and jaguar are available on the NCAR mss in the following two tar files, and the bluefire dataset is available on the web via the link below. The datasets were generated using the cesm1_0_rel02 tag in October 2010.
- bluefire.cesm1_0_rel02.tar.gz (~320 MB)
On 29 November 2012, the port-validation procedure was repeated using the cesm1_0_4 tag (CESM1.0.4 was released in February 2012). For this case, the known good data sets were generated on bluefire but no comparisons with another machine are made. Below are links to plots of the 1v2, 1v3, and 4v5 comparisons as well as a tarball of the known-good data sets.