Checking intent (inout) vs (out) for shared version of WRF-MPAS Noah LSM by davegill · Pull Request #5 · davegill/WRF

davegill · 2019-03-15T20:20:27Z

/gpfs/fs1/scratch/gill/WRF_SHARED_PHYSICS/WRF/

After bug fix, ran a test with previous vs instrumented code. The instrumented code had every variables that was modified between WRF and MPAS initialized as HUGE. The only two fields that are
different between the two short runs are two diagnostic fields: SMCREL and NOAHRES.

> ../../external/io_netcdf/diffwrf INIT_HUGE_intent/wrfout_d01_2000-01-24_12:00:00 INIT_develop/wrfout_d01_2000-01-24_12:00:00 
 Just plot  F
Diffing INIT_HUGE_intent/wrfout_d01_2000-01-24_12:00:00 INIT_develop/wrfout_d01_2000-01-24_12:00:00
 Next Time 2000-01-24_12:00:00
     Field   Ndifs    Dims       RMS (1)            RMS (2)     DIGITS    RMSE     pntwise max
 Next Time 2000-01-24_12:03:00
     Field   Ndifs    Dims       RMS (1)            RMS (2)     DIGITS    RMSE     pntwise max
   NOAHRES      2031    2   0.2317166950E+39   0.1723680767E+00 -39   0.2317E+39   0.3146E+39
 Next Time 2000-01-24_12:06:00
     Field   Ndifs    Dims       RMS (1)            RMS (2)     DIGITS    RMSE     pntwise max
    SMCREL      8124    3   0.2317166950E+39   0.8362615464E+00 -38   0.2317E+39   0.7161E+38
   NOAHRES      2031    2   0.2317166950E+39   0.2442041814E+00 -38   0.2317E+39   0.1825E+39
 Next Time 2000-01-24_12:09:00
     Field   Ndifs    Dims       RMS (1)            RMS (2)     DIGITS    RMSE     pntwise max
    SMCREL      8124    3   0.2317166950E+39   0.8358389944E+00 -38   0.2317E+39   0.7161E+38
   NOAHRES      2031    2   0.2317166950E+39   0.2705677661E+00 -38   0.2317E+39   0.1619E+39
 Next Time 2000-01-24_12:12:00
     Field   Ndifs    Dims       RMS (1)            RMS (2)     DIGITS    RMSE     pntwise max
    SMCREL      8124    3   0.2317166950E+39   0.8355161195E+00 -38   0.2317E+39   0.7161E+38
   NOAHRES      2031    2   0.2317166950E+39   0.2709646415E+00 -38   0.2317E+39   0.1617E+39

…lc (had nan assignment after init value)

…zap i,j print

davegill · 2019-03-21T18:28:37Z

@mgduda @weiwangncar @dudhia @kkeene44 @smileMchen
Folks,
Please take a look at these diffs. I don't know Laura's github ID.

mgduda · 2019-03-21T18:30:41Z

@ldfowler58 Just bringing you into the conversation...

weiwangncar · 2019-03-21T23:06:34Z

@davegill Do you mean SH2OIN becomes 10^38?

davegill · 2019-03-21T23:14:51Z

Do you mean SH2OIN becomes 10^38?
@weiwangncar
Wei,
In the computation (I will replicate it for posterity), the value of SH2OIN is 10^38 (probably that is HUGE). That variable is used, and should not be.

weiwangncar · 2019-03-22T00:06:29Z

@davegill I will rephrase my question: did you set this field, SH2OIN, to a huge number, or this field contains huge numbers as a result of changing other fields to huge numbers?

TYPE: bug fix KEYWORDS: Noah LSM sh2o unified MPAS SOURCE: internal DESCRIPTION OF CHANGES: Problem: A single variable was used as two calling arguments in a subroutine. Inside the subroutine, the separate names were pointing to the same memory and were treated differently. Solution: Since the two instances of the array were treated as IN and OUT, a copy of the array was made and passed in as separately named input field. There were two locations to fix. LIST OF MODIFIED FILES: M phys/module_sf_noahlsm.F TESTS CONDUCTED: 1. Comparing the results before vs after show no bit-for-bit differences (what we wanted). ``` ../../external/io_netcdf/diffwrf INIT_develop/wrfout_d01_2000-01-24_12:00:00 INIT_FIX_dev/wrfout_d01_2000-01-24_12:00:00 Just plot F Diffing INIT_develop/wrfout_d01_2000-01-24_12:00:00 INIT_FIX_dev/wrfout_d01_2000-01-24_12:00:00 Next Time 2000-01-24_12:00:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:03:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:06:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:09:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:12:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max ``` 2. When the code was instrumented, the only differences were diagnostic only. No differences in any other WRF output fields. See davegill#5 (github.com/davegill/WRF PR 5).

TYPE: bug fix KEYWORDS: Noah LSM sh2o unified MPAS SOURCE: internal DESCRIPTION OF CHANGES: Problem: A single variable was used as two calling arguments in a subroutine. Inside the subroutine, the separate names were pointing to the same memory and were treated differently. Solution: Since the two instances of the array were treated as IN and OUT, a copy of the array was made and passed in as separately named input field. There were two locations to fix. LIST OF MODIFIED FILES: M phys/module_sf_noahlsm.F TESTS CONDUCTED: 1. Comparing the results before vs after show no bit-for-bit differences (what we wanted). ``` ../../external/io_netcdf/diffwrf INIT_develop/wrfout_d01_2000-01-24_12:00:00 INIT_FIX_dev/wrfout_d01_2000-01-24_12:00:00 Just plot F Diffing INIT_develop/wrfout_d01_2000-01-24_12:00:00 INIT_FIX_dev/wrfout_d01_2000-01-24_12:00:00 Next Time 2000-01-24_12:00:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:03:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:06:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:09:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max Next Time 2000-01-24_12:12:00 Field Ndifs Dims RMS (1) RMS (2) DIGITS RMSE pntwise max ``` 2. When the code was instrumented, the only differences were diagnostic only. No differences in any other WRF output fields. See davegill/WRF#5 (github.com/davegill/WRF PR 5).

davegill added 5 commits March 15, 2019 14:14

Checking intent in out for WRF MPAS Noah LSM swap

a49f7b7

extra lines

0887f7c

extra lines

b997d74

extra lines

a2e0664

extra lines

2c6e1bc

davegill changed the title ~~Checking intent in out for WRF MPAS Noah LSM swap~~ Checking intent (inout) vs (out) for shared version of WRF-MPAS Noah LSM Mar 15, 2019

davegill added 7 commits March 15, 2019 19:16

Add variable names to printout

e760bfb

Restrict NaN assignment to land-only, fix BIG oops in subtoutine alca…

d681d20

…lc (had nan assignment after init value)

Fix inits in noahlsm: rdtis, dew, and if(ua_phys) for etpn; noahdrv: …

9223ce3

…zap i,j print

Zap piles of print statements

e2cbd41

Remove "fixes" to character MMINLU etc, just be basic

7756072

Allow HUGE initialization, not NaN

63d4102

nan -> huge -> 0.

d56c03e

Found bug, tada! sh2o is two args in sstep

9b67d1b

davegill mentioned this pull request Mar 22, 2019

SH2O had two appearances in call to ssttep routine wrf-model/WRF#833

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Checking intent (inout) vs (out) for shared version of WRF-MPAS Noah LSM#5

Checking intent (inout) vs (out) for shared version of WRF-MPAS Noah LSM#5
davegill wants to merge 13 commits intodevelopfrom
intent

davegill commented Mar 15, 2019 •

edited

Loading

Uh oh!

davegill commented Mar 21, 2019

Uh oh!

mgduda commented Mar 21, 2019

Uh oh!

weiwangncar commented Mar 21, 2019

Uh oh!

davegill commented Mar 21, 2019

Uh oh!

weiwangncar commented Mar 22, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

davegill commented Mar 15, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

davegill commented Mar 21, 2019

Uh oh!

mgduda commented Mar 21, 2019

Uh oh!

weiwangncar commented Mar 21, 2019

Uh oh!

davegill commented Mar 21, 2019

Uh oh!

weiwangncar commented Mar 22, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

davegill commented Mar 15, 2019 •

edited

Loading