Help needed-Setting up HAFS DOCN workflow on Orion #88
-
|
I have submitted cronjob_hafs_docn.sh many times according to instruction. Appreciate your attention and help. ================================================================================================================================ |
Beta Was this translation helpful? Give feedback.
Replies: 5 comments 28 replies
-
|
second attempt succeeded forecast it seems lilke it! ================================================================================================================================ |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 Also, regarding your question: "One last question, I promise-- How to switch to lower resolution gfs?" The resolution of the atmospheric lateral boundary conditions is controlled by chgres_cube. However, I think we can modify the workflow to obtain lower-resolution LBCs. What resolution are you hoping for? |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 Thanks! I'm trying to figure out how to generate the gfs_bndy.tile7 file with these dimensions. What does the grid look like in your experiment? Is it the regional HAFS grid? If not, can you point me to your model output? |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina My question is how to output the lower resolution: I do not think this is related to regional ocean resolution, I checked both low and high gfs bndy experiments, regional resolution has alwasy been: |
Beta Was this translation helpful? Give feedback.
-
|
Hi @lliu1220, thanks! But do you have model output from a run in which the gfs_bndy.tile7.0*.nc has the desired lower resolution: If I could see the settings used in chgres_cube to generate the lower-resolution gfs_bndy file, I could implement them in the HAFS workflow. |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina and @lliu1220, if your test is based on @uturuncoglu's ufs-weather-model level regression test, then most likely, this is the HAFS application level configuration I used to generate the atmospheric model needed grid, IC/BC. ${PYTHON3} ./run_hafs.py -t ${dev} 2019082900 00L HISTORY |
Beta Was this translation helpful? Give feedback.
-
|
I made copy of /work/noaa/hwrf/save/bliu/hafs_develop_202102 I wonder if this is the right way of making this setup work, Thanks. |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220, it's better to checkout the latest HAFS develop branch or the feature/hafs_sync_202107 branch. But, anyway, if you copied over my directory, you need to change this 'ENTITY ACCOUNT "aoml-hafs1"' into 'ENTITY ACCOUNT "&CPU_ACCOUNT;"' in rocoto/sites/orion.ent. And then follow @evankalina's instruction above to specify the cpu_account properly (default is hurricane on Orion). BTW, if you want to run the HAFS CDEPS workflow, you probably want to checkout the feature/hafs_couplehycom_cdeps branch. Also @evankalina it will be nice if you sync this feature/hafs_couplehycom_cdeps with the latest HAFS develop branch when you get a chance. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
@BinLiu-NOAA I did not find any 'aoml' keywords, and it seemed already specified as CPU_ACCOUNT. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 it's in this file: rocoto/sites/orion.ent |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 and @BinLiu-NOAA As of this morning, the feature/hafs_couplehycom_cdeps branch is synced with the latest HAFS develop branch, and we have issued a PR to pull it into develop. |
Beta Was this translation helpful? Give feedback.
-
|
I'm currently using 20200825 hurricane data under your this directory- I wonder how likely to produce these hurricane data for 20200720, Hurricane Isaias Initial condition, can you please instruct? |
Beta Was this translation helpful? Give feedback.
-
|
Thanks! @evankalina , /work/noaa/dtc-hwrf/ekalina/hafs_couplehycom_cdeps-20210813 Thank you! [config] [forecast] |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 The permissions should be correct now. The downloaded OISST dataset has global coverage, but the data are being used in a regional HAFS DOCN experiment. |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina |
Beta Was this translation helpful? Give feedback.
-
|
@uturuncoglu @BinLiu-NOAA there is a script under /work/noaa/hwrf//save/lliu/hafs_couplehycom_cdeps-20210813/ush/hafs_oisst_download.py |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 I am not sure if I fully understand your question. OISST is an observations-based analysis of SST that uses data from many platforms (see link), while HYCOM data are an SST analysis (at t=0h) or forecast (at t>0h) generated by that modeling system. So they are different. If you are asking if it is possible to prepare DOCN runs that use the output from HYCOM, it can likely be done, but it isn't supported by the workflow itself. You would need to follow the steps in Appendix B ("Considerations for adding a new dataset") in the HAFS-CDEPS documentation. |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina So I took it as no, for now, correct me if i'm wrong. |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 OK, thanks for clarifying. Would you please open read permissions? |
Beta Was this translation helpful? Give feedback.
-
|
@lliu1220 Since you've already successfully run this configuration using HYCOM data, I see no reason why you couldn't use the code from the feature/hafs_couplehycom_cdeps branch to do it. The branch is up to date with all of the recent changes to ufs-weather-model. |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina |
Beta Was this translation helpful? Give feedback.
-
|
Closing the loop on this discussion, since it was addressed offline. In short, the answer to Ling's question is: "The hurricane track forecast is produced by the "product" task in the HAFS workflow, which requires inputs from the "atm_post" task. If you want the hurricane track forecast, you will need to run the full workflow, rather than the offline regression test." Ling wanted to create a data ocean using output from MOM6, which isn't officially supported in the workflow. But since she already prepared the MOM6 input data and the mesh file, it was straightforward to adapt the workflow to her needs: "Basically, you can pretend that you are using OISST data, and the workflow won't know the difference. Here are the steps:
|
Beta Was this translation helpful? Give feedback.
-
|
@evankalina |
Beta Was this translation helpful? Give feedback.
-
|
@evankalina |
Beta Was this translation helpful? Give feedback.
@evankalina and @lliu1220, if your test is based on @uturuncoglu's ufs-weather-model level regression test, then most likely, this is the HAFS application level configuration I used to generate the atmospheric model needed grid, IC/BC.
Orion:
vi /work/noaa/hwrf/save/bliu/hafs_develop_202102/rocoto/cronjob_hafs_rt_test.sh
${PYTHON3} ./run_hafs.py -t ${dev} 2019082900 00L HISTORY
config.EXPT=${EXPT} config.SUBEXPT=${EXPT}_rt_regional_C192_1n4_static_new
config.NHRS=126 ${scrubopt}
../parm/hafsC192_1n4_L91_AL.conf
../parm/hafs_hycom.conf
${PYTHON3} ./run_hafs.py -t ${dev} 2019082900 00L HISTORY
config.EXPT=${EXPT} config.SUBEXPT=${EXPT}_rt_regional_C192_1n4_static_atmonly_new
config.NH…