From ed947c379da14a22780c4d52cc094893dfaa941d Mon Sep 17 00:00:00 2001 From: Bill Sacks Date: Thu, 8 Aug 2024 19:43:15 -0600 Subject: [PATCH] Rework documentation of glc_renormalize_smb Rework documentation of glc_renormalize_smb to be consistent with the changes in https://github.com/ESCOMP/CMEPS/pull/495 --- doc/source/clm-cism-coupling.rst | 12 ++++-------- doc/source/t-compsets.rst | 14 ++++++-------- 2 files changed, 10 insertions(+), 16 deletions(-) diff --git a/doc/source/clm-cism-coupling.rst b/doc/source/clm-cism-coupling.rst index 93f5d119..5e1e01fc 100644 --- a/doc/source/clm-cism-coupling.rst +++ b/doc/source/clm-cism-coupling.rst @@ -43,14 +43,10 @@ coupling becomes important. There are still some unhandled edge cases in this re significant case being when CLM dictates an amount of glacial melt that exceeds the available ice in a CISM grid cell), but we do take pains to achieve conservation in most cases. An important mechanism to achieve this conservation is via a global renormalization -step done when mapping SMB from CLM to CISM, as described in :numref:`remapping_smb`. **By -default, this renormalization is done when running with a two-way-coupled ice sheet that -sends fluxes to other components (which is typically true in this case of an evolving ice -sheet), but is turned off in other cases. This leads to small differences in the remapped -SMB field in runs with an evolving vs. non-evolving ice sheet. This default can be -overridden via the driver namelist variable,** ``glc_renormalize_smb``, **keeping in mind -that setting this to** ``off`` **will break conservation for a configuration with an -evolving, two-way-coupled ice sheet.** +step done when mapping SMB from CLM to CISM, as described in :numref:`remapping_smb`. +(This renormalization is done by default, but can be turned off by setting the driver +namelist variable, ``glc_renormalize_smb`` to ``off``, keeping in mind that this will +break conservation in a fully-coupled configuration.) Evolving ice sheet with one-way (diagnostic) coupling ----------------------------------------------------- diff --git a/doc/source/t-compsets.rst b/doc/source/t-compsets.rst index 72da0576..247c8e48 100644 --- a/doc/source/t-compsets.rst +++ b/doc/source/t-compsets.rst @@ -100,14 +100,12 @@ averages will appear in the cpl/hist directory within your archive space, with n files are about 7 MB per year.) A T compset run that later uses these coupler history files as forcing should give -*nearly* identical CISM results as the original run, **as long as you ensure that SMB -renormalization is done (or not done) in both cases, as described in** -:numref:`clm-cism-coupling`. **(For example, if the coupler history files were generated -from a fully-coupled case with an evolving, two-way-coupled ice sheet, then in the T case -you should set** ``glc_renormalize_smb = 'on'`` **in** ``user_nl_cpl`` **in order to -reproduce the results of the fully-coupled case.)** Small differences arise because these -forcing files are written with single precision, leading to roundoff error on the order of -10\ :sup:`-7`. If you need more precision, you can add the following to ``user_nl_cpl``: +*nearly* identical CISM results as the original run, as long as you ensure that SMB +renormalization is done (or not done) in both cases, as described in +:numref:`clm-cism-coupling` (based on the setting of ``glc_renormalize_smb``). Small +differences arise because these forcing files are written with single precision, leading +to roundoff error on the order of 10\ :sup:`-7`. If you need more precision, you can add +the following to ``user_nl_cpl``: .. code-block:: console