Skip to main content

Browse the archive

https://github.com/E3SM-Project/E3SM
06 April 2024, 03:10:57 UTC
sort by:
Revision Author Date Message Commit Date
8edb7cd Balwinder Singh 11 June 2020, 18:22:28 UTC Merge branch 'yfeng/eam/dust2updatesRRTMGP' (PR #3610) Update dust shortwave optics for RRTMGP and retuned emis factor for V2 Update dust shortwave optics file for RRTMGP Update dust emission factor retuned for V2 for all compsets (from 0.95 to 1.38) [Non-BFB] - Non Bit-For-Bit [NML] - Namelist Changing * yfeng/eam/dust2updatesRRTMGP: Update dust shortwave optics for RRTMGP and retuned emis factor for V2 11 June 2020, 18:22:28 UTC
bb2b27a Jon Wolfe 11 June 2020, 16:34:42 UTC Merge branch 'mark-petersen/ocean/update_mpas_200610' (PR #3631) mpas-source: Move daysSinceStartOfSim calculation to master when threaded This PR brings in a new mpas-source submodule, with changes only to the ocean core. It moves the calculation of daysSinceStartOfSim, which is done in three places, to inside a loop with "omp master" pragmas around it. Previously, one was inside an "omp single" pragma, but that did not guarantee that the master thread would get the correct result. While this does not impact the results at all, it was causing random problems in testing using threaded configurations where cprnc output from globalStats files would find an error, just in the daysSinceStartOfSim field. With thanks to @amametjanov See MPAS-Dev/MPAS-Model#601 [BFB] 11 June 2020, 16:34:42 UTC
320720a Wuyin Lin 10 June 2020, 20:54:29 UTC Merge tangq/atm/O3_hole_diags (PR #3625) Add total column ozone output for diagnosing ozone hole Ozone hole diagnostics require daily output. This commit adds the inline calculation for total column ozone and the corresponding output field "TOZ" and thus reduces the cost of saving daily 3D "O3" to 2D "TOZ". Additionally, the inline calculation is more accurate, because it uses instantaneous values instead of the averages, which is potentially important for ozone hole areas where large variances may occur. [BFB] - Bit-For-Bit 10 June 2020, 20:54:29 UTC
678123a mark-petersen 10 June 2020, 19:54:23 UTC mpas-source: Move daysSinceStartOfSim calculation to master when threaded 10 June 2020, 19:54:23 UTC
85c6db0 noel 10 June 2020, 18:40:24 UTC Merge branch E3SM-Project/ndk/machinefiles/gnu-remove-Og (PR #3629) Remove the -Og flag from all DEBUG GNU builds and add C++ flags for GNU 10 June 2020, 18:40:24 UTC
958d7bf Balwinder Singh 10 June 2020, 16:38:07 UTC Merge branch 'mingxuanwupnnl/atm/trop_strat_mam' (PR #3541) Fix bugs for running E3SM with trop_strat_mam3/mam7 Current model can't run with MOZART chemistry (trop_strat_mam3/mam7) as well as MAM3/MAM7 (trop_mam3/trop_mam7). These modifications don't change the results for default configurations (linoz_mam4_resus_mom_soag). To run E3SM with trop_strat_mam7, one also needs to modify namelist setting for now. linoz_data_file = '' linoz_data_path = '' demott_ice_nuc = .false. gas_wetdep_method = 'NEU' For running trop_strat_mam3, one also needs to either modify chem_mech.in or resetting CAM_CONFIG_OPTS (remove resus). Fixes #3542 [BFB] * mingxuanwupnnl/atm/trop_strat_mam: Fix bugs for running E3SM with trop_strat_mam3/mam7 10 June 2020, 16:38:07 UTC
ee648f0 Jon Wolfe 10 June 2020, 15:41:52 UTC Merge branch 'mark-petersen/ocn/update_mpas_200608' (PR #3627) Update mpas-source: cvmix driver area weighted average bug This PR brings in a new mpas-source submodule with changes only in the ocean core. Specifically, it fixes a bug in the area weighted averaging for 3 CVMix quantities. See MPAS-Dev/MPAS-Model#556 [non-BFB] 10 June 2020, 15:41:52 UTC
af3d238 Aaron Donahue 10 June 2020, 15:28:08 UTC Merge branch 'jayeshkrishna/scorpio_support_homme_standalone' (PR #3615) Adding support for standalone HOMME builds with Scorpio. The HOMME standalone builds can now use E3SM Scorpio by turning on the HOMME_USE_SCORPIO cmake option. By default the standalone builds were using the PIO library available with CIME. Now by default the HOMME standalone builds use the E3SM Scorpio classic library. This PR also updates the Scorpio submodule to bring in some fixes required for standalone HOMME builds [BFB] 10 June 2020, 15:28:08 UTC
e45232c noel 10 June 2020, 00:39:44 UTC Merge branch E3SM-Project/ndk/machinefiles/cori-update-mpich-netcdf (PR #3626) Update Cori modules for mpich and netcdf/hdf 10 June 2020, 00:39:44 UTC
b16eae1 Balwinder Singh 09 June 2020, 18:19:21 UTC Merge branch 'worleyph/cam/runtime_pcols_part1' (PR #3620) Implement pcols as a runtime option in EAM In the earlier 2000's, empirical experiments indicated that setting the first dimension of the chunk data structure (pcols) at compile-time allowed the compiler to generate more performance-efficient code, and this motivated this design choice in CAM. However, CAM/EAM has evolved significantly since then, and, for example, physics_types now uses psetcols to define this first dimension, and the compiler never sees a compile-time parameter. So the compile-time setting of pcols may no longer serve a useful purpose. Extensive experiments indicate that there is still a small performance advantage from setting pcols at compile-time, but also that being able to set pcols at runtime greatly simplifies the process of optimizing over the pcols parameter. Here pcols is implemented as a runtime option, currently specified via the user_nl_cam namelist variable phys_chnk_fdim. phys_chnk_fdim must be at least as large as 1, but there is no specified upper bound. The current compile-time value of the CPP variable PCOLS is used to define the default value for phys_chunk_fdim (and pcols). PCOLS is now represented in EAM as the parameter ppcols, but is only referenced in defining the default for phys_chunk_fdim. To restore the compile-time setting of pcols, and also to specify a pcols value of PCOLS, add -pcols to CAM_CONFIG_OPTS in env_build.xml . For example, to specify 32 as the compile-time value of pcols, add '-pcols 32' . This allows the recovery of any lost performance from the runtime-specification of pcols. Note that this is the first part of a two-part introduction of runtime pcols. This PR sets the runtime pcols during the call to read_namelist. To calculate a better default than the compile-time PCOLS CPP variable requires that it be calculated in phys_grid_init, and this requires changing the order of certain steps in the initialization of EAM. The current PR, however, stands on its own, and is acceptable even if the second part PR is never submitted. [BFB] This addresses some aspects of Issue #3538 , but not all. The issue will be resolved by a second runtime pcols PR. If we decide not to submit this follow-on PR, the issue will also be closed. * worleyph/cam/runtime_pcols_part1: Back out change in diag_surf Modify variables names Enable compile-time setting of pcols via CAM_CONFIG_OPTS Eliminate compile-time dependence on pcols in crm_physics.F90 Implement pcols as a runtime option in EAM Eliminate compile-time dependence on pcols in diag_surf Eliminate compile-time dependence on pcols in zm_conv Eliminate compile-time dependence on pcols in phys_grid.F90 Eliminate compile-time dependence on pcols in unicon_cam Eliminate compile-time dependence on pcols in check_tracers logic Eliminate compile-time dependence on pcols in camsrfexch.F90 Eliminate compile-time dependence on pcols in chemistry 09 June 2020, 18:19:21 UTC
e9a4ccc Jason Sarich 09 June 2020, 17:54:34 UTC Merge branch 'ndk/run_e3sm/adjust-order-submodule-init' (PR #3530) Update the run_e3sm template to do the submodule command after checking out any branches. Update the github location to refer to git@github.com:E3SM-Project/E3SM.git Also remove reference to retired edison machine. For the example "customknl" layout, switch to using the correct variable: MAX_MPITASKS_PER_NODE [bfb] 09 June 2020, 17:54:34 UTC
a69b2b2 Jon Wolfe 09 June 2020, 17:09:36 UTC Merge branch 'mark-petersen/ocn/update_mpas_200605' (PR #3623) Update mpas-source, compass only. Add WC14 This PR brings in a new mpas-source submodule, with changes only to the ocean model. Specifically, this modifies COMPASS initial conditions only by adding the Water Cycle 14 km case. [BFB] 09 June 2020, 17:09:36 UTC
cde90b1 noel 09 June 2020, 16:12:20 UTC Add boilerplate CXXFLAGS for GNU compiler. [bfb] 09 June 2020, 16:12:20 UTC
ca9476f Patrick Worley 08 June 2020, 23:04:56 UTC Back out change in diag_surf After re-examination with the total package of the changes, the line call outfld('QFLX', cam_in%cflx(1,1), pcols, lchnk) in subroutine diag_surf in physics/cam/cam_diagnostics.F90 no longer fails to compile when pcols is not a parameter. As clfx is a 2D array (pcols,pcnst), the original form of the call to outfld is more self-documenting, and also matches the style of other calls to outfld accessing slices of multidimensional arrays. Here, the original code is restored. [BFB] 08 June 2020, 23:04:56 UTC
b676b30 noel 08 June 2020, 22:56:32 UTC Removing the -Og flag from all DEBUG GNU builds. The -g flag remains in place, of course. This allows us to run on Cori with GNU DEBUG. I also removed the "gnu7" compiler definition as it's no longer used. [bfb] 08 June 2020, 22:56:32 UTC
65db573 mark-petersen 08 June 2020, 16:54:30 UTC Update mpas-source: cvmix driver area weighted average bug 08 June 2020, 16:54:30 UTC
a3ef027 Wuyin Lin 08 June 2020, 16:15:53 UTC Merge tangq/atm/PSC_Temp (PR #3618) Make PSC T threshold a namelist variable This PR makes the hard-coded T threshold of PSC ozone loss a namelist variable to facilitate model tuning. The default T threshold is set to 193 K, the same as the hard-coded value, so it won't change the model results by default. [BFB] 08 June 2020, 16:15:53 UTC
f023cf0 Qi Tang 06 June 2020, 05:00:10 UTC Add total column ozone output for diagnosing ozone hole Ozone hole diagnostics require daily output. This commit adds the inline calculation for total column ozone and the corresponding output field "TOZ" and thus reduces the cost of saving daily 3D "O3" to 2D "TOZ". Additionally, the inline calculation is more accurate, because it uses instantaneous values instead of the averages, which is potentially important for ozone hole areas where large variances may occur. [BFB] - Bit-For-Bit 06 June 2020, 05:00:10 UTC
865793e mark-petersen 05 June 2020, 23:13:43 UTC Update mpas-source, compass only. Add WC14 05 June 2020, 23:13:43 UTC
5363c7b noel 05 June 2020, 17:28:23 UTC Update Cori (cori-knl and cori-haswell) modules to be current defaults for the following: cray-mpich/7.7.6 -> cray-mpich/7.7.10 cray-netcdf-hdf5parallel/4.6.1.3 -> cray-netcdf-hdf5parallel/4.6.3.2 cray-hdf5-parallel/1.10.2.0 -> cray-hdf5-parallel/1.10.5.2 cray-parallel-netcdf/1.8.1.4 -> cray-parallel-netcdf/1.11.1.1 gcc/8.2.0 -> gcc/8.3.0 cray-libsci/19.02.1 -> cray-libsci/19.06.1 impi/2019.up3 -> impi/2020 And only affecting builds with intel19 compiler, use the latest version of intel19 which is intel/19.1.1.217. Not updating craype or the default intel compiler (still intel18). Those will be in another PR. [bfb] 05 June 2020, 17:28:23 UTC
a206b22 AaronDonahue 05 June 2020, 16:53:58 UTC Merge branch 'ambrad/eam/v2-perf-and-mass' PR #3594) Improve vectorization and threading in SL transport, physgrid remap, and theta-l. Fix a mass conservation error due to finite precision effects. These code mods would cause no answer change in exact arithmetic but are non-BFB in finite precision. Summary: Physgrid remap: Switch indexing from (:nf,:nf) to (:nf2) to improve vectorization. SL transport: Speed up global CAAS by locally reducing what can be reduced decomp-BFB. Improve own_q, rmt_q thread load balance. Flip level, q indexing to improve memory locality. After a performance regression to calc_q a few months ago to work around a compiler bug, address the performance issue in a different way. I avoid using pragma ivdep but get the performance that annotation provides by explicitly blocking. The result is effective vectorization of a key computation in SL transport. Fix dp_fv so it agrees numerically with the pdel calculation in dp_coupling.F90: derived_physics. Add a C++ block to the cori-knl/intel19 block in config_compilers.xml. Work around an Intel 17 -O3 bug to fix a standalone-Homme test on Anvil. Fixes #3588. [non-BFB] 4 EAM tests and 3 standalone Homme tests 05 June 2020, 16:53:58 UTC
f898f39 Junqi Yin 04 June 2020, 23:59:47 UTC Merge branch 'bishtgautam/lnd/ne120np4' (PR #3484) Adds support for F20TRC5-CMIP6 compset at ne120 resolution The definition of ne120np4 is modified to use 0.125deg river grid instead of 0.5deg river grid. The missing landuse timeseries for ne120 resolution is added for F20TRC5-CMIP6 [non-BFB] 04 June 2020, 23:59:47 UTC
8f7abca Balwinder Singh 04 June 2020, 17:10:39 UTC Merge branch 'jiansunpnnl/atm/nudge_for_v2'(PR #3569) Implements revised nudging code We have implemented a revised version of nudging code in E3SM. Changes include: linear interpolation functionality for handling the nudging data flexible handling of nudging input data (now can read daily nudging input file with multiple time slices) switch to change the location where nudging tendency is calculated (in nudging.F90 and physpkg.F90). bug fixes for intermittent nudging (in nudging.F90) nudging tendency update (in physpkg.F90, affect FV dycore only) duplicated namelist definitions (in namelist_definition.xml) unit of Nudge_T (from cp*K/s to K/s, in nudging.F90) example of nudging settings for a typical E3SM run or RRM run Example settings for a nudged simulation using the newly added namelist variables: - Nudge_Method = 'Linear'! use linear-inter instead of !step-function nudging - Nudge_File_Ntime = 4! if there are four time slices per !nudging input data file - Nudge_Loc_PhysOut = .TRUE.! calculate the nudging tendency where ! model state are saved from a baseline simulation; this helps !when the model is nudged towards its baseline simulation More details about these changes could be found in the comments of namelist_definition.xml and nudging.F90 modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 modified: components/cam/src/physics/cam/physpkg.F90 Original development branch can be found here: https://github.com/E3SM-Project/E3SM/tree/jiansunpnnl/ndg_loc The revised nudging implementation has been evaluated in Sun et al. (2019). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019MS001831 [BFB] when nudging is not switched on [BFB] when nudging is switched on, but with the following namelist configuration: nudge_method = 'step' nudge_loc_PhysOut = .false. nudge_tau = -999 nudge_file_ntime = 1 !! one time slice per nudging input data file nudge_currentStep = .false. Fixes #3565 * jiansunpnnl/atm/nudge_for_v2: Minor change for an if condition Bug fix for free-running run with latest nudging code Update description in xml file modified: components/cam/bld/namelist_files/namelist_definition.xml Fix some bugs in the code and update words in xml file modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 Update the nudging codes based on Wuyin Lin's comments Add the missing codes accidently deleted by the previous commit This commit has not been tested yet. modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 modified: components/cam/src/physics/cam/physpkg.F90 Update the comments for CONUS RRM nudged simulation modified: components/cam/src/physics/cam/nudging.F90 Add example namelist settings for the typical global nudged simulation and CONUS RRM simulation, suggested by Kai Zhang and Qi Tang. modified: components/cam/src/physics/cam/nudging.F90 Update nudging code for nudge_file_ntime check modified: components/cam/src/physics/cam/nudging.F90 Some minor changes modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 Further clean up codes based on Kai's comments modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 modified: components/cam/src/physics/cam/physpkg.F90 Update the nudging code and xml descriptions for Num_Slice Confirm that the current nudging code only works for the nudging file with one-day data. modified: components/cam/bld/namelist_files/namelist_definition.xml modified: components/cam/src/physics/cam/nudging.F90 Update the code and namelist descriptions based on Kai's comments Remove the constraint for Num_Slice for testing Update the nudging related codes and xml file The major changes can be found in Sun et al. (2019, https://github.com/E3SM-Project/E3SM/tree/jiansunpnnl/ms2019) Some further improvements can be found at https://github.com/E3SM-Project/E3SM/tree/jiansunpnnl/ndg_loc 04 June 2020, 17:10:39 UTC
98d349e Jon Wolfe 04 June 2020, 14:27:30 UTC Merge branch 'mark-petersen/ocn/update_mpas_200601' (PR #3613) Update mpas-source: analysis member update, remove unused modules This PR brings in a new mpas-source submodule with changes only to the ocean, including several updates to analysis members and removing modules that are not used in E3SM simulations: * Adds normalVelocity arrays to eddy products (MPAS-Dev/MPAS-Model#544) * Adds Ocean Heat Content AM (MPAS-Dev/MPAS-Model#455) * Ocean/remove unused modules (MPAS-Dev/MPAS-Model#569) It includes corresponding changes to E3SM namelists and streams files for MPAS-Ocean. COMPASS changes associated with these PRs are also included. [NML] [BFB] 04 June 2020, 14:27:30 UTC
bdc1d9c Patrick Worley 04 June 2020, 00:13:07 UTC Modify variables names Modified names of variables referring to virtual SMPs in phys_grid.F90, for example, from smpx to vsmp, for improved self-documentation. [BFB] 04 June 2020, 00:13:07 UTC
493a78b Patrick Worley 03 June 2020, 23:53:49 UTC Enable compile-time setting of pcols via CAM_CONFIG_OPTS Currently, adding the -pcols option to CAM_CONFIG_OPTS in env_build.xml sets PCOLS, which is then used as the compile-time parameter if -DPPCOLS is also added to Macros.cmake and as the default value of phys_chnk_fdim otherwise. Here we change the behavior to force the addition of -DPPCOLS whenever -pcols is included in CAM_CONFIG_OPTS. This is a more convenient mechanism for restoring the compile-time setting of PCOLS. This convenience comes at the cost of removing the ability to change the default value for phys_chnk_fdim. This capability is unlikely to be missed, but we can revisit the decision in the future if it ends up that there is a demand for this capability. [BFB] 03 June 2020, 23:53:49 UTC
747cc60 Balwinder Singh 03 June 2020, 23:15:55 UTC Minor change for an if condition 03 June 2020, 23:15:55 UTC
28dabf8 Patrick Worley 03 June 2020, 22:55:44 UTC Eliminate compile-time dependence on pcols in crm_physics.F90 The module variables cldo_save, q_aero, qqcw_save, qqcw_all, and dgnumwet_save in module crm_physics in crm_physics.F90 include pcols as one of their dimensions. These variables are changed to allocatable, and then are allocated in crm_physics_init (also in crm_physics.F90). Also in crm_physics.F90, the crm_physics_register local variables dims_gcm_1D, dims_gcm_2D, dims_crm_2D, dims_crm_3D, dims_crm_rad, and dims_crm_aer are initialized in their declarations with the value pcols. This requires pcols to be available at compile-time. Here they are instead initialized during runtime at the beginning of the routine crm_phys_register. [BFB] 03 June 2020, 22:55:44 UTC
b1b5ea3 Patrick Worley 03 June 2020, 22:50:51 UTC Implement pcols as a runtime option in EAM In the earlier 2000's, empirical experiments indicated that setting the first dimension of the chunk data structure at compile-time allowed the compiler to generate more performance-efficient code, and this motivated this design choice in CAM. However, CAM/EAM has evolved significantly since then, and, for example, physics_types now uses psetcols to define this first dimension, and the compiler may not recognize that some fields have the first dimension set at compile-time. Based on extensive experiments, compile-time setting of PCOLS is no longer as big a performance advantage as previously (though it is still somewhat faster). Being able to set pcols at runtime increases ease of use, and increased flexibility for future developments where pcols might need to be set after runtime case specifications are read in. Here pcols is implemented as a runtime option, currently specified via the user_nl_cam namelist variable phys_chnk_fdim. phys_chnk_fdim must be at least as large as 1, but there is no specified upper bound. The current compile-time value of PCOLS is used to define the default value for phys_chnk_fdim (and pcols). PCOLS is now represented in EAM as the parameter ppcols, but is only referenced in defining the default for phys_chnk_fdim. The ability to set pcols at compile-time is retained as well. This allows continued comparisons of performance between compile-time and runtime pcols, and also to specify pcols at compile-time for maximum performance for production jobs. To enable compile-time setting of pcols, add '-DPPCOLS' to CPPFLAGS in Macros.cmake for a case before building. With this change, phys_chnk_fdim will be ignored (but a warning message will be output if phys_chnk_fdim differs from the compile-time setting of pcols. [BFB] 03 June 2020, 22:50:51 UTC
eb2017e Jian Sun 03 June 2020, 22:29:53 UTC Bug fix for free-running run with latest nudging code modified: nudging.F90 03 June 2020, 22:29:53 UTC
118771e Qi Tang 20 May 2020, 05:46:35 UTC make PSC T threshold a namelist variable (cherry picked from commit 17aa44dd9e5e05bd8491d3468dc45e32e9a2c0b9) 03 June 2020, 22:06:47 UTC
53f2392 Patrick Worley 03 June 2020, 21:13:04 UTC Eliminate compile-time dependence on pcols in diag_surf The line call outfld('QFLX', cam_in%cflx(1,1), pcols, lchnk) in subroutine diag_surf in physics/cam/cam_diagnostics.F90 fails to compile when pcols is not a parameter. However, call outfld('QFLX', cam_in%cflx, pcols, lchnk) compiles successfully when pcols is a global runtime variable, and appears to still do the right thing. This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 21:13:04 UTC
6984c67 Patrick Worley 03 June 2020, 21:06:43 UTC Eliminate compile-time dependence on pcols in zm_conv The module variable dcapemx in the zm_conv module, in physics/cam/zm_conv.F90, is dimensioned as size pcols. This is changed to allocatable, and is allocated in subroutine zm_convr. It is not subsequently deallocated as dcapemx is persistent throughout the execution of EAM. A test in zm_convr prevents it from being allocated more than once. This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 21:06:43 UTC
1b6512e Patrick Worley 03 June 2020, 21:02:22 UTC Eliminate compile-time dependence on pcols in phys_grid.F90 The data types chunk and lchunk in the phys_grid module, in physics/cam/phys_grid.F90, refer to pcols in dimensioning some of the fields in these types. These fields are changed to allocatable. Then these fields for variables of type chunk are allocated in the subroutine phys_grid_init and in the subroutine create_chunks, and these fields for variables of type lchunk are allocated in the subroutine phys_grid_init. They are never deallocated as the associated type chunk and lchunk variable are module variables persistent throughout the execution of EAM. Also, the local variable gcols in phys_grid_init is dimensioned as size pcols. As phys_grid_init is a potential location for optimizing the value of pcols, this is changed to be allocatable, and then allocated and deallocated in phys_grid_init. This removes dependencies on the compile-time specification of pcols, and also introduces the flexibility of computing an optimal choice of pcols within physi_grid_init. [BFB] 03 June 2020, 21:02:22 UTC
00b21cb Patrick Worley 03 June 2020, 20:28:12 UTC Eliminate compile-time dependence on pcols in unicon_cam The data type unicon_out_t in the unicon_cam module, in physics/cam/unicon_cam.F90, refers to pcols in dimensioning some of the fields in the type. These fields are changed to allocatable, and these fields for variables of type unicon_out_t are then allocated in unicon_cam_tend, and deallocated in a new subroutine unicon_cam_tend_free. A call to unicon_cam_tend_free is added to subroutine convect_shallow_tend in convect_shallow.F90. This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 20:28:12 UTC
b8b6a73 Patrick Worley 03 June 2020, 20:17:20 UTC Eliminate compile-time dependence on pcols in check_tracers logic The data type check_tracers_data in the check_energy module, in physics/cam/check_energy.F90, refers to pcols in dimensioning some of the fields in the type. These fields are changed to allocatable, and these fields for variables of type check_tracers_data are then allocated in check_tracers_init, and deallocated in a new subroutine check_tracers_fini. A call to check_tracers_fini is added to subroutine tphysac and to subroutine tphysbc, both in physics/cam/physpkg.F90 and in physics/crm/physpkg.F90 . This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 20:17:20 UTC
6ba7240 Bibi Mathew 03 June 2020, 20:15:20 UTC Merge branch 'azamat/compy/throttle-par-build' (PR #3504) Reduce parallel builds on Compy [BFB] 03 June 2020, 20:16:37 UTC
e710122 Patrick Worley 03 June 2020, 19:51:54 UTC Eliminate compile-time dependence on pcols in camsrfexch.F90 The data types cam_in_t and cam_out_t in the camsrfexch module in control/camsrfexch.F90 and in physics/crm/camsrfexch.F90 refer to pcols in dimensioning some of the fields in these types. These fields are changed to allocatable. Then these fields for variables of type cam_in_t are allocated in hub2atm_alloc and deallocated in hub2atm_dealloc, and these fields for variables of type cam_out_t are allocated in atm2hub_alloc and deallocated in atm2hub_dealloc. All of these routines reside in camsrfexch.F90. This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 19:51:54 UTC
d2432ca mark-petersen 03 June 2020, 18:33:41 UTC Update mpas-source to fix bugs 03 June 2020, 19:15:40 UTC
6fd0325 Patrick Worley 03 June 2020, 18:55:20 UTC Eliminate compile-time dependence on pcols in chemistry The data type wetdep_inputs_t in chemistry/aerosol/wetdep.F90 refers to pcols in dimensioning some of the fields in the type. These fields are changed to allocatable, and these fields for variables of type wetdep_inputs_t are then allocated in wetdep_inputs_set, and deallocated in a new subroutine wetdep_inputs_unset. A call to wetdep_inputs_unset is added to aero_model_wetdep in chemistry/bulk_aero/aero_model.F90 and to aero_model_wetdep in chemistry/modal_aero/aero_model.F90. This removes one dependency on the compile-time specification of pcols. [BFB] 03 June 2020, 18:55:20 UTC
a416ca4 Mark Petersen 01 June 2020, 22:30:27 UTC Add variables to monthly time average 03 June 2020, 17:34:46 UTC
968af37 Mark Petersen 01 June 2020, 22:14:46 UTC add ocean_heat_content to cmake 03 June 2020, 17:34:46 UTC
5edb8e8 Jon Wolfe 01 June 2020, 21:20:40 UTC Add new oceanHeatContent AM to streams generation script 03 June 2020, 17:34:46 UTC
d6305f7 Jon Wolfe 01 June 2020, 21:04:03 UTC Bld/script updates corresponding to Registry changes 03 June 2020, 17:34:45 UTC
6f1a259 mark-petersen 01 June 2020, 17:50:55 UTC Update mpas-source: analysis member update, remove unused modules 03 June 2020, 17:34:45 UTC
c8e85e9 jayeshkrishna 02 June 2020, 17:55:22 UTC Updating Scorpio to v 1.1.1 Updating the Scorpio and Scorpio classic to latest version 1.1.1 that includes fix to disable builds of Scorpio tools. 02 June 2020, 17:55:22 UTC
8b641ac Gautam Bisht 02 June 2020, 03:56:22 UTC Merge branch 'rgknox/lnd/fates-api-10' (PR #3563) This set of changes modifies the interface with fates to accommodate API 10. This includes an updated fates parameter file. FATES API 10 focused on changes to plant hydraulics and plant age. [non-BFB] 02 June 2020, 03:56:22 UTC
cc28394 James Foucar 29 May 2020, 18:28:46 UTC Update CIME to ESMCI cime5.8.23 (#3603) Update CIME to ESMCI cime5.8.23 Squash merge of jgfouca/branch-for-to-acme-2020-05-27 Features: * allow compiler to modify MAX_TASKS_PER_NODE and MAX_MPITASKS_PER_NODE * cprnc: extend allowed string length * cprnc: Change type to double for derived vars in cprnc * create_newcase: Allow user to specify an extra machines directory * use shrmap by reference instead of copy Bug fixes: * improve cime buildlib function debug logging * mct driver: fix cime_model to stdout issue * cprnc: Fix div0s in cprnc comparisons * atm grids: fix atm level spec so that it does not match 1x1_brazil *case.setup: fix case.setup --clean Other changes not relevant to E3SM * PIO: fix issue with mpi-serial io * first steps to ESMF data models - creation of separate ESMF data model share code Fixes #3573 [NML] 29 May 2020, 18:28:46 UTC
d6a1709 Ryan Knox 29 May 2020, 17:38:40 UTC removed fates hydro test until it passes 29 May 2020, 17:38:40 UTC
be8d6ea Wuyin Lin 29 May 2020, 15:40:43 UTC Merge singhbalwinder/atm/gust-wsubmin (PR #3526) Adds wsubmin namelist parameter to gustiness use case The gustiness parameterization is missing wsubmin in the use case file. This PR fixes that. It also changes the test mod directory to add a shell command to use the use case file for the gustiness test rather than mentioning namelist parameters explicitly. [Non-BFB] - Non Bit-For-Bit for cam-gust_param tests. [NML] - Namelist Changing 29 May 2020, 15:40:43 UTC
dba932a Jon Wolfe 29 May 2020, 15:30:48 UTC Merge branch 'jayeshkrishna/fix_cice_hist_sync_fail' (PR #3607) Ensure sync is called before freedecomp This PR modifies CICE sourcecode only, freeing decompositions in the CICE history write routine before the file close. Fixes #3605 [BFB] 29 May 2020, 15:30:48 UTC
c6f15b2 Jon Wolfe 29 May 2020, 15:21:52 UTC Merge branch 'jonbob/scripts/update-lanlic-200421' (PR #3554) Update machine settings for LANL machines grizzly and badger This PR updates machine and compiler settings for LANL IC machines grizzly and badger. It reflects the new modules made available by the LANL HPC group, with many thanks to Jennifer Green. No testing is performed on these machines, so the changes will not impact test results. However, it will likely change answers on cases run on these machines due to new compiler versions. [BFB] 29 May 2020, 15:21:52 UTC
aa5649e Jon Wolfe 28 May 2020, 19:34:26 UTC Remove -lmpi_cxx library from grizzly support 28 May 2020, 19:34:26 UTC
471916e AaronDonahue 28 May 2020, 19:00:41 UTC Merge branch aarondonahue/uninitialized_elem (PR #3571) This commit addresses an issue with passing a null pointer as input to a subroutine when the number of MPI ranks is greater than the number of dynamics elements. The element structure "elem" is unitialized on all MPI ranks that are not tagged to solve dynamics. Thus keeping the original declaration as a null pointer. In some cases with certain compiler flags active this can cause the simulation to halt with an error related to passing the null pointer "elem" as an input. This fix allocates the elem pointer to be empty. [BFB] 28 May 2020, 19:00:41 UTC
d8f3666 Yan Feng 28 May 2020, 17:30:53 UTC Update dust shortwave optics for RRTMGP and retuned emis factor for V2 * Update dust shortwave optics for RRTMGP * Update dust emission factor retuned for V2 [Non-BFB] - Non Bit-For-Bit [NML] - Namelist Changing 28 May 2020, 17:30:53 UTC
98e8084 Jon Wolfe 28 May 2020, 15:16:24 UTC Merge branch 'jonbob/ocean/streams-set-io_type' (PR #3525) Set io_type to ocn_pio_typename for all streams that were not set This PR modifies the script that builds the streams.ocean file. It changes the io_type to be ocn_pio_typename for all streams that were previously unset. This fixes problems on platforms that do not have pnetcdf built. [BFB] 28 May 2020, 15:16:24 UTC
382ce93 jayeshkrishna 28 May 2020, 14:21:45 UTC Merge pull request #3601 from E3SM-Project/jayeshkrishna/add_openmpi_anlworkstation Adding support for OpenMPI on anlworkstation. Also adding paths to I/O libraries that is required for static builds with Scorpio. [BFB] 28 May 2020, 14:21:45 UTC
d042ae5 jayeshkrishna 27 May 2020, 23:39:42 UTC Ensure sync is called before freedecomp Moving the freeing of decompositions before the file close. Without the change when freeing io descriptors Scorpio performs a syncfile on the File handle (that is no longer valid because it was closed) resulting in an abort (sync fails since it cannot find the associated file handle). 27 May 2020, 23:39:42 UTC
272b791 James Foucar 27 May 2020, 20:10:49 UTC Merge branch 'jgfouca/cime/toss3_cprnc' into master (PR #3602) sandiatoss3: cprnc should be the thing you run Fixes HOMME which assumes 'cprnc'. [BFB] * jgfouca/cime/toss3_cprnc: sandiatoss3: cprnc should be the thing you run 27 May 2020, 20:14:24 UTC
199ae33 James Foucar 27 May 2020, 20:08:00 UTC sandiatoss3: cprnc should be the thing you run Fixes HOMME which assumes 'cprnc'. [BFB] 27 May 2020, 20:08:00 UTC
803d501 James Foucar 26 May 2020, 22:00:52 UTC Merge branch 'jgfouca/cime/toss3_modules' into master (PR #3600) sandiatoss3: Update netcdf and hdf5 modules Will fix ERIO.ne30_g16_rx1.A.sandiatoss3_intel. Fixes #3591 [BFB] * jgfouca/cime/toss3_modules: sandiatoss3: Update netcdf and hdf5 modules 27 May 2020, 16:58:35 UTC
ebc0b9b Jon Wolfe 27 May 2020, 15:10:39 UTC Merge branch 'mark-petersen/ocean/gpumesh' (PR #3589) New ocean mesh structure with GPU replication This PR brings in a new mpas-source submodule with changes only to the ocean core. It adds a mesh structure with all current mesh pool data and replicates that data on a GPU (or other accelerator) for the duration of the simulation to reduce the need for data transfers of static mesh data. It is a critical piece of infrastructure needed for future GPU modifications being staged. In a later change, this structure was replaced by public module variables instead since a large user-defined type did not perform well on GPUs. Mesh variables can now be accessed just by "use"-ing this module. This module also has a revised index naming in which nCells, nCellsSolve and nCellsArray are replaced by nCellsAll, nCellsOwned, and nCellsHalo(n) where the latter differs from nCellsArray by only including the index of each halo depth and no longer includes nCellsOwned as the first entry. Similar changes were introduced for Edges, Vertices. NOTE: this change requires only one block per MPI task and will abort if more than one block per MPI task is attempted. Any multi-block tests will fail. This has been tested on summit in a QU240 test case and is bit-for-bit since the structure itself is not yet being used anywhere. However, I have verified the data is correctly transferred to the GPU accelerator on Summit. See MPAS-Dev/MPAS-Model#496 by @philipwjones [BFB] 27 May 2020, 15:10:39 UTC
a0d23ee James Foucar 26 May 2020, 21:45:32 UTC sandiatoss3: Update netcdf and hdf5 modules Will fix ERIO.ne30_g16_rx1.A.sandiatoss3_intel. [BFB] 26 May 2020, 21:45:32 UTC
f2a0868 jayeshkrishna 26 May 2020, 19:50:15 UTC Disable Scorpio tools build Since HOMME builds Scorpio within its own build system it causes issues with building the Scorpio tools. Disabling the Scorpio tools for now. 26 May 2020, 19:50:15 UTC
2880ac6 Ryan Knox 26 May 2020, 18:19:06 UTC fates-interface: cleaning up some deprecated comments 26 May 2020, 18:19:06 UTC
53127d3 Balwinder Singh 26 May 2020, 16:41:09 UTC Merge branch 'cameronsmith1/atm/super-fast_MAM4' (PR #3491) Implements superfast chemistry into the E3SMv1 model Implements superfast chemistry into the E3SMv1 model, including merging with MAM4. The following compsets have been added (with appropriate defaults): A_WCYCL1850_AR5emis_superfast A_WCYCL20TR_AR5emis_superfast F1850S-AR5-superfast F20TRS-AR5-superfast NOTE: these 20TR compsets default to starting in 1990, and fixed concentration fields will cycle 1995 conditions. NOTE: the 'S' in the name of the F compsets indicates that there won't be energy flux correction (ieflx_opt=0). [BFB] * cameronsmith1/atm/super-fast_MAM4: Change F-Compsets for super-fast chemistry so that ieflx_opt = 0 Implementation of 1990(20TR) defaults for superfast chemistry. Corrected a comment in the pre-processor input file for superfast_mam4 Fix for groupid change from acme to e3sm at NERSC for superfast branch Added superfast chemistry into E3SMv1 Conflicts: cime/config/e3sm/machines/config_machines.xml components/cam/cime_config/config_component.xml components/clm/cime_config/config_component.xml 26 May 2020, 16:41:09 UTC
0280478 Andrew M. Bradley 25 May 2020, 08:28:58 UTC Homme: Increase q mass output to 16 digits, from 15. 25 May 2020, 08:28:58 UTC
8f26a70 Andrew M. Bradley 17 May 2020, 03:21:23 UTC EAM: Add C++ flags to the cori-knl intel19 config_compilers entry. This entry does not inherit from the base intel entry, and it lacked C++ flags. Add these. 25 May 2020, 01:18:25 UTC
4c5762f Andrew M. Bradley 24 May 2020, 22:06:13 UTC Homme: Work around an Intel 17 -O3 bug. 25 May 2020, 01:18:25 UTC
8b21600 Andrew M. Bradley 22 May 2020, 07:05:38 UTC HOMME: In physgrid remap, change dp_fv calculation to numerically match EAM. HOMME standalone was conserving mass to machine precision. But the method of computing dp on the physics grid in gllfvremap_mod differed numerically from that in dp_coupling.F90: derived_physics. That caused a numerical source/sink at relative magnitude ~1e-10 in EAM runs. Fixes #3588 25 May 2020, 01:18:25 UTC
e973f4c Andrew M. Bradley 17 May 2020, 04:11:05 UTC HOMME: In physgrid remap: Convert (:nf,:nf) -> (:nf2) indexing. This change improves vectorization. 24 May 2020, 01:52:02 UTC
748fc4b Andrew M. Bradley 11 May 2020, 06:23:32 UTC HOMME: Improve performance of SL transport, particularly on KNL. * Better threading in CAAS * Better loop order in CEDR run_global and run_local * Better threading SL own_q and rmt_q * Better vectorization in calc_q 24 May 2020, 01:52:02 UTC
e01f59f jayeshkrishna 03 January 2020, 22:38:24 UTC Avoid accessing internal Scorpio structs Fix code that accesses the internal members (type) of Scorpio data structures (var_desc_t). The code now uses extra variables to keep track of the variable type. 22 May 2020, 23:38:32 UTC
a968c12 jayeshkrishna 12 December 2019, 22:20:38 UTC Add option to build HOMME with Scorpio Adding the necessary cmake options and code to allow building HOMME (standalone) with Scorpio. 22 May 2020, 23:38:32 UTC
6d63530 jayeshkrishna 12 December 2019, 22:19:10 UTC Fix directory for cprnc on anlworkstation CPRNC_DIR needs to point to the directory containing the cprnc tool (not point to the tool itself). Fixing the directory name on anlworkstation avoids building cprnc everytime. 22 May 2020, 23:38:32 UTC
bd7818e jayeshkrishna 12 December 2019, 18:51:14 UTC Using env settings for anlworkstation builds Using paths to NetCDF, HDF5 and PnetCDF from the environment for anlworkstation, instead of hardcoding them in the CMake scripts. 22 May 2020, 23:38:32 UTC
8aef41b jayeshkrishna 12 December 2019, 18:07:11 UTC Using Scorpio instead of PIO for HOMME Using Scorpio (available in <HOMME_SOURCE>/utils/externals) instead of PIO (available via CIME) to build standlone version of HOMME 22 May 2020, 23:38:32 UTC
ff54662 jayeshkrishna 12 December 2019, 17:44:51 UTC Adding soft link to E3SM externals for HOMME Adding soft link to E3SM externals (<E3SM_SOURCE_DIR>/externals) in HOMME utils directory so that standalone HOMME builds can use Scorpio (and Scorpio classic). Without this change standalone HOMME builds can only use the PIO library available in CIME 22 May 2020, 23:38:31 UTC
e4a74e1 mark-petersen 22 May 2020, 22:12:44 UTC mpas-source: New ocean mesh structure with GPU replication 22 May 2020, 22:12:44 UTC
ec1a137 jayeshkrishna 22 May 2020, 20:09:22 UTC Merge branch 'jayeshkrishna/switch_to_scorpio2' (PR #3515) Switching the default I/O library from Scorpio classic to Scorpio Also updating Scorpio to the latest version (1.1.0) [BFB] * jayeshkrishna/switch_to_scorpio2: Updating Scorpio to 1.1.0 Updating NetCDF module on melvin Moving a high mem CLM test to integration suite Upgrading some intel related modules on bebop Switching to Scorpio Updating to Scorpio v 1.0.4 Conflicts: cime/config/e3sm/tests.py 22 May 2020, 20:09:22 UTC
a562e77 Jon Wolfe 22 May 2020, 20:02:06 UTC Merge branch 'mark-petersen/ocean/200429_compass_only' (PR #3561) Update mpas-source: COMPASS only This PR brings in a new mpas-source submodule, with changes that only update COMPASS initialization scripts in the ocean. These changes do not affect E3SM simulations. [BFB] 22 May 2020, 20:02:06 UTC
3bc8d86 Benjamin Hillman 22 May 2020, 19:00:24 UTC Merge branch 'brhillman/mmf/fix-rrtmgp-cam-optics' into master (#3581) Fixes the MMF-specific radiation_tend to use the new optics interfaces, which commit a4eea9c modified to provide additional outputs to use with COSP. Commit 417e7cd then broke the MMF build because it relied on the old interface, so this fixes the broken build. Also switches the MMF test in the integration suite from an ERP test to an ERS test so that we have a test that works in master for MMF until we can track down the ERP fail. [BFB] 22 May 2020, 19:00:24 UTC
5489e83 rgknox 22 May 2020, 00:42:27 UTC Added a hydro test to fates, correction to column checking in fates hydro 22 May 2020, 00:42:27 UTC
8b25ebd Wuyin Lin 21 May 2020, 23:14:13 UTC Merge singhbalwinder/atm/campp-O2-to-O1 (PR #3578) Modifies chem pp optimization level from O2 to O1 for PGI compilers Chem pp was hanging on Compy when compiled with the PGI compiler. An easy fix is to decrease the optimization level from 2 to 1. It works fine on all other testing machines and it also works fine on Compy with the Intel compiler with O2 optimizations. [BFB] - Bit-For-Bit 21 May 2020, 23:14:13 UTC
5f82375 Jon Wolfe 21 May 2020, 17:55:17 UTC Merge branch 'mark-petersen/ocean/submodfix' (PR #3560) Update mpas-source: fix incorrect path for ocean cvmix When changing to submodules for the ocean cvmix package, the path to cvmix source had changed, but one instance of this path was not changed, leading to build failures with PGI on Summit. This PR fixes that bad path. See MPAS-Dev/MPAS-Model#494 [BFB] 21 May 2020, 17:55:17 UTC
fb3d085 Benjamin Hillman 21 May 2020, 17:51:29 UTC Change MMF ERP test to ERS test Change the MMF test in the integration suite from an ERP test to an ERS test. ERP tests are currently failing the base restart comparison on most machines, so in order to have a test in the integration suite that will pass so that we can regression test we are temporarily changing to ERS while we further investigate the ERP fail. Note that the ERP actually passes on cori-knl, which is why we did not catch this sooner as we had been doing all of our MMF testing there. 21 May 2020, 17:51:29 UTC
901d72e mark-petersen 21 May 2020, 13:31:33 UTC Update mpas-source: COMPASS only 21 May 2020, 13:31:33 UTC
89dc465 Oksana Guba 20 May 2020, 21:24:11 UTC Merge branch 'AMB/ambrad/homme/fix-3582' into master (PR #3583) When restarting with theta_hydrostatic_mode = true, init w_i to phinh_i to 0. This prevents a possible NaN dection in prim_printstate. This PR does not change answers because w_i and phinh_i are not actually used in hydrostatic mode. I detected a second source of potential NaNs in prim_printstate: When Qdp is written at restart, the unused time slot is uninitialized. This can then lead to NaNs in repro_sum. Fix this by 0'ing out all of Qdp before writing restart data to it. Fixes #3582 [BFB] 20 May 2020, 21:24:11 UTC
56931d1 Oksana Guba 20 May 2020, 21:23:29 UTC Merge branch 'origin/oksanaguba/eam/ftype1-nh-ver2' into master (PR #3575) Fixing ftype1 for NH runs, previously, NH variables were not updated in dp_coupling layer. [nonBFB] for cam-thetanh_ftype1 test. 20 May 2020, 21:23:29 UTC
6a2aebd Oksana Guba 20 May 2020, 21:22:09 UTC Merge branch 'origin/ndk/cam/remove-native_mapping' into master (PR #3548) Remove cam/src/dynamics/se/native_mapping.F90 and 3 calls to this module in dyn_comp.F90. These calls were causing some issues when run with more MPI's than elements (which Intel 19 caught in DEBUG) and rather than correcting this, it was decided this functionality is no longer needed. This code creates mapping files (SE grid to lat/lon grid) and is not used since we make mapping files with ncremap. Fixes #2899 [bfb] 20 May 2020, 21:22:09 UTC
0e70545 jayeshkrishna 20 May 2020, 04:08:31 UTC Updating Scorpio to 1.1.0 Updating Scorpio (and Scorpio classic) to version 1.1.0 This version includes, * Support for ADIOS2 * Misc bug fixes (including a fix required for NetCDF 4.7.4, the version of the NetCDF library used in melvin) 20 May 2020, 04:08:31 UTC
1c1422c Benjamin Hillman 19 May 2020, 22:46:14 UTC Fix zero-size sw_band_midpoints and lw_band_midpoints output Fix bug that was causing us to try to write sw_band_midpoints and lw_band_midpoints using zero-size arrays in the MMF version of the radiation_init routine. This was caused by these arrays being automatic arrays that relied on the nswbands and nlwbands dimension sizes having not been defined yet. Oddly enough, the switch to PIO2 was what caught this bug, and threw an error at runtime when trying to output these arrays. Making these arrays allocatable and allocating them with sizes that are defined earlier in the routine fixes the runtime fails when using PIO2. 19 May 2020, 22:46:14 UTC
663fa20 Andrew M. Bradley 19 May 2020, 07:04:23 UTC Homme: Squash another source of NaNs in prim_printstate. When Qdp is written in restarted, the unused time slot is uninitialized. This can cause NaNs to be detected in prim_printstate. Like the previous commit, this fix does not change answers; it only affects prim_printstate. [BFB] 19 May 2020, 07:04:23 UTC
da47312 Andrew M. Bradley 19 May 2020, 01:10:38 UTC Homme: Fix a restart bug. When restarting with theta_hydrostatic_mode = true, init w_i to phinh_i to 0. This prevents a possible NaN dection in prim_printstate. This commit does not change answers because w_i and phinh_i are not actually used in hydrostatic mode. Fixes #3582 [BFB] 19 May 2020, 01:13:13 UTC
081781b jayeshkrishna 18 May 2020, 22:59:51 UTC Updating NetCDF module on melvin Updating NetCDF module on melvin from NetCDF 4.4.1 (C) to 4.7.4. (Patch from Jim Foucar) Without this upgrade ERIO.ne30_g16_rx1.A.melvin_gnu hangs on melvin with Scorpio. 18 May 2020, 23:08:03 UTC
0c237dc Oksana Guba 18 May 2020, 21:51:02 UTC Merge 'AMB/ambrad/eam/physgrid-sl3d-cleanup' into master (PR #3567) Cleanups and quality improvements to physgrid and SL algorithms: * Remove fv_physics_coupling_mod and fv_physgrid_struct in components/cam/dynamics/se. * Improve FV area calculation. * Use cubed_sphere_map=2 cell center when computing metric tensors for vector physgrid remap algorithm. * Switch 3D level vertical reconstruction to a forward-in-time algorithm. That is computationally more efficient and also mathematically tidier. * Add a Homme theta-SL-pg2 test using DCMIP 2016 test 1. One major change: * Apply tendencies at every vertical remap step when running SL transport in 3D mode. Previously, they were applied only at the tracer time step. [non-BFB] in 3 Homme tests (2 old, 1 new) and 4 e3sm_developer tests (3 old, 1 new) 18 May 2020, 21:51:02 UTC
63e9b9f Benjamin Hillman 18 May 2020, 16:12:41 UTC Update radiation_tend for cam_optics changes Fixes the MMF-specific `radiation\_tend` to use the new optics interfaces, which PR #3562 modified to provide additional outputs to use with COSP. PR #3553 then broke the MMF build because it relied on the old interface, so this fixes the broken build. 18 May 2020, 16:21:36 UTC
417e7cd Benjamin Hillman 18 May 2020, 15:07:34 UTC Merge branch 'whannah/mmf/fix-mmf-rrtmgp' into master (#3553) This fixes a bug in RRTMGP with the MMF configuration. The radiative fluxes were not multiplied by the pressure thickness. This caused the radiative heating to have the wrong units. The fix simply mirrors how radiative tendencies are treated for the non-MMF configuration with RRTMGP. This PR also updates the MMF tests to ensure threading is disabled, and adds a test to the e3sm_integration suite. [BFB] (except for MMF tests, which were not in the integration suite) 18 May 2020, 15:07:34 UTC
a4eea9c Wuyin Lin 18 May 2020, 14:53:08 UTC Merge brhillman/atm/enable-rrtmgp-cosp (PR #3562) Enable COSP with RRTMGP Enable running the CFMIP Observation Simulator Package (COSP) with RRTMGP. This provides a number of additional diagnostics that are directly comparable to satellite retrievals of cloud properties from a number of different instruments. To build with COSP, simply add -cosp to CAM_CONFIG_OPTS in env_build.xml (i.e., ./xmlchange --append CAM_CONFIG_OPTS='-cosp'). No other action should need to be taken, and h0 output files should include a selection of COSP outputs. [BFB] 18 May 2020, 14:53:08 UTC
5910536 Jian Sun 15 May 2020, 19:45:51 UTC Update description in xml file modified: components/cam/bld/namelist_files/namelist_definition.xml 15 May 2020, 19:45:51 UTC
back to top