Activities for week of October 10, 2016
- Pull requests:
- #14: Fix problem reading radiance_info files for some compilers https://github.com/wrf-model/WRF/pull/14
- From Jamie: instructions for hybrid were incorrect; ensemble perturbations should be in directory named "ep/" within working directory, not linked into the working directory
- Modified WRFDA regtest
- MAJOR BUG FIX: get_repo_revision returned wrong hash due to missing "chdir" into repository
- check for compatible netcdf/hdf5 builds on yellowstone
- Stop renaming da_wrfvar.exe to include compiler/parallelism in filename; this hasn't been necessary for a long time and is cumbersome
- Move addition of HDF5 library path to LD_LIBRARY_PATH inside the job script, so tests can easily be manually re-run
- From Jamie: Fix hybrid tests as above
- Add commit date to regression test online summary (not finished)
- Consolidate some RTTOV code and allow command line option to specify whether or not to compile with RTTOV
- WRFHELP emails
- User (University of Chicago) seeking help running stand-alone adjoint in WRFPLUS; replied with guidance from Jamie
- Replied to Thomas Schwitalla about getting old SEVIRI data
- Helped NCAR visitor Chester Huang with problems running jobs on Yellowstone
- Helping user William Miller from University of Maryland with various questions and a 4DVAR issue. Ongoing.
- Updated website
- Updated free_data site with new NOMADS archive link
- Updated FAQ with question on how to run 4DVAR with nesting (not finished)
- Updated user's guide
- From Jamie: Fix hybrid instructions (not put on web yet)
- Preparing external user's code (radar null echo assimilation) for review.
- Put on Github branch in my fork (https://github.com/mkavulich/WRF/tree/null-echo-assimilation), made simple modification so it compiles for Linux (still needs updating)
- Made To-do list on the wiki: To do list
- Deleted some old branches in main repository; sent emails to some people asking about some more that might be deleted
Activities for week of October 17, 2016
- Ran some tests for Jamie regarding two bug fixes
- Ran regtest for a radar assimilation bug; this lead to a few regtest modifications to catch problems with this option in the future
- Ran regtest for a bug related to WRFPLUS with cu_physics=0 (confirmed temp fix works for 4DVAR). Also running stand-alone TL and AD models to try to diagnose the source of the problem
- Discovered bug in rsl printout for long filenames in WRF and WRFPLUS (found in above WRFPLUS testing) and verified a fix (not committed yet)
- Modified WRFDA regtest
- Allow command-line specification of a github fork to check out the code from
- Allow command-line specification of which tests to run, and only compile the necessary code
- WRFHELP emails
- Answered two questions from UMD user, one about VARBC, the other about satellite thinning
- User from University of Hawaii trying GPS_ZTD assimilation sent correction to help page (http://www2.mmm.ucar.edu/wrf/users/wrfda/OnlineTutorial/Help/littler.html#special); I verified the correction was correct and updated the page
- User from University of Naples "Parthenope" was asking about reducing memory usage in 4DVAR
- Sent some guidance to user looking for info on the PREPBUFR format
- Developer committee tasks:
- Gave WPS repository access to two developers
- Took half-day of PTO
Activities for week of October 24, 2016
- Developer committee tasks:
- Helped 3 developers with WRF repository access
- Updated wiki page with more about creating and using forks: SVN to Git guide for WRF developers
- Looking to reduce repository size by eliminating old, unused files from repo history
- Ran WRFDA regtests
- gfortran 4.8.2: pass
- gfortran 5.3.0 and 6.1.0: 4DVAR errors! Investigating...
- Modified WRFDA regtest
- Put test namelists under version control, so changes to tests are tracked
- Modified some tests to cover some more cases:
- Modified the WRF namelist for cycle_sene_hires so that file names are long
- Modified radar_4dvar_cv7 test so cu_physics=0; Jamie found a bug for this case in WRFPLUS
- Pull requests:
- #26: Allow filenames greater than 30 characters to be correctly printed to RSL log files https://github.com/wrf-model/WRF/pull/26
- WRFHELP emails
- Helping one user with radiance diagnostics
- Answered one user from private company in Hong Kong looking to use scale-dependent blending scheme
- User from University of Montana having compilation problems: due to HDF5 env variable
- Answered a user's questions about the obs thinning method
- WRFPLUS:
- Investigating easy method for keeping up-to-date with master using "git cherry-pick"
- Attended Release Committee meeting
- Some questions about upcoming release
- Are we ready for theta_m on by default?
- What needs to be changed for compatibility with new hybrid vertical coordinate?
- Some questions about upcoming release
Activities for week of October 31, 2016
- Ran WRFDA regtests
- Still investigating 4DVAR failures. Seems to be due to MPI issue with a recent commit by Jamie; sent her more information so she can take a look
- Jamie sent fix, all tests pass!
- Now getting random failures on Yellowstone for all REGTEST jobs. CISL told me they are due to memory problems
- Still investigating 4DVAR failures. Seems to be due to MPI issue with a recent commit by Jamie; sent her more information so she can take a look
- WRFHELP emails
- Still having a back-and-forth with Maryland user about VARBC and other radiance-related questions
- Answered user (Chinese Academy of Meteorological Sciences) asking about BE length scale and variance values
- Giving a Russian user guidance on cycling frequency
- Working on stand-alone GENBE
- Made a bunch of cosmetic changes, mostly eliminating unnecessary code copy-pasted from the WRF build system
- Contacted Gael again with a few questions