diff --git a/.gitignore b/.gitignore index a335b7107e..eb05482fec 100644 --- a/.gitignore +++ b/.gitignore @@ -120,3 +120,5 @@ Depends _build*/ _publish*/ doc/.coverage +doc/ssh_key* +*.ipynb_checkpoints/ diff --git a/doc/source/tech_note/Glacier/CLM50_Tech_Note_Glacier.rst b/doc/source/tech_note/Glacier/CLM50_Tech_Note_Glacier.rst index e8a4b81298..513f24194b 100644 --- a/doc/source/tech_note/Glacier/CLM50_Tech_Note_Glacier.rst +++ b/doc/source/tech_note/Glacier/CLM50_Tech_Note_Glacier.rst @@ -3,138 +3,143 @@ Glaciers ======== -This chapter describes features of CLM that are specific to coupling to an ice sheet model (in the CESM context, this is the CISM model; see the `CISM documentation and user's guide _` for more information). General information about glacier land units can be found elsewhere in this document (see Chapter :numref:`rst_Surface Characterization, Vertical Discretization, and Model Input Requirements` for an overview). +In CLM, glaciers are represented using glacier land units (Chapter :numref:`rst_Surface Characterization, Vertical Discretization, and Model Input Requirements`) that simulate snow, ice, surface energy balance, and runoff processes over permanently glaciated surfaces. -.. _Glaciers summary of CLM5.0 updates relative to CLM4.5: +CLM distinguishes between two major glacier categories: -Summary of CLM5.0 updates relative to CLM4.5 --------------------------------------------- +#. Continental glaciers, i.e., the Greenland and Antarctic Ice Sheets. -Compared with CLM4.5 (:ref:`Oleson et al. 2013 `), CLM5.0 contains substantial improvements in its capabilities for land-ice science. This section summarizes these improvements, and the following sections provide more details. +#. Mountain (or alpine) glaciers, represented in the Randolph Glacier Inventory (including glaciers on the peripheries of the ice sheets). -- All runs include multiple glacier elevation classes over Greenland and Antarctica and compute ice sheet surface mass balance in those regions. +Some glacier processes and model infrastructure are shared between these glacier types e.g., glacier land units. However, the fully developed glacier capabilities in CLM currently focus on continental ice sheets through coupling with the Community Ice Sheet Model (CISM; https://escomp.github.io/cism-docs/). Refer to section :numref:`Mountain Glaciers` for ongoing and future developments focusing on glaciers outside the ice sheets. -- A number of namelist parameters offer fine-grained control over glacier behavior in different regions of the world (section :numref:`Glacier regions`). (The options used outside of Greenland and Antarctica reproduce the standard CLM4.5 glacier behavior.) -- CLM can now keep its glacier areas and elevations in sync with CISM when running with an evolving ice sheet. (However, in typical configurations, the ice sheet geometry still remains fixed throughout the run.) +.. _Glacier regions: -- The downscaling to elevation classes now includes downwelling longwave radiation and partitioning of precipitation into rain vs. snow (section :numref:`Multiple elevation class scheme`). +Glacier regions and their behaviors +----------------------------------- -- Other land units within the CISM domain undergo the same downscaling as the glacier land unit, and surface mass balance is computed for the natural vegetated land unit. This allows CLM to produce glacial inception when running with an evolving ice sheet model. +Within CLM, the world's glaciers and ice sheets are divided into three default glacier regions, each with distinct behaviors related to elevation classes, glacial meltwater treatment, and runoff generation. These default configurations are summarized in Table :numref:`Table Glacier region behaviors`. -- There have also been substantial improvements to CLM's snow physics, as described in other chapters of this document. +.. _Table Glacier region behaviors: -.. _Overview Glaciers: +.. table:: Glacier region behaviors -Overview --------- + +---------------+---------------+---------------+---------------+ + | Region | Elevation | Glacial melt | Ice runoff | + | | classes | | | + +===============+===============+===============+===============+ + | Greenland | Virtual | Replaced by | Remains ice | + | | | ice | | + +---------------+---------------+---------------+---------------+ + | Antarctica | Multiple | Replaced by | Remains ice | + | | | ice | | + +---------------+---------------+---------------+---------------+ + | Mountain | Single | Remains in | Melted | + | glaciers | | place | | + +---------------+---------------+---------------+---------------+ -CLM is responsible for computing two quantities that are passed to the ice sheet model: -#. Surface mass balance (SMB) - the net annual accumulation/ablation of mass at the upper surface (section :numref:`Computation of the surface mass balance`) +The glacier regions differ in three primary respects: -#. Ground surface temperature, which serves as an upper boundary condition for CISM's temperature calculation The ice sheet model is typically run at much higher resolution than CLM (e.g., :math:`\sim`\ 5 km rather than :math:`\sim`\ 100 km). To improve the downscaling from CLM's grid to the ice sheet grid, the glaciated portion of each grid cell is divided into multiple elevation classes (section :numref:`Multiple elevation class scheme`). The above quantities are computed separately in each elevation class. The CESM coupler then computes high-resolution quantities via horizontal and vertical interpolation, and passes these high-resolution quantities to CISM. +#. Elevation class configuration -There are several reasons for computing the SMB in CLM rather than in CISM: + a. Multiple elevation classes (section :numref:`Multiple elevation class scheme`) + b. Multiple elevation classes plus virtual elevation classes + c. A single elevation class whose elevation matches the atmospheric grid-cell topography, such that no atmospheric downscaling is applied -#. It is much cheaper to compute the SMB in CLM for :math:`\sim`\ 10 elevation classes than in CISM. For example, suppose we are running CLM at a resolution of :math:`\sim`\ 50 km and CISM at :math:`\sim`\ 5 km. Greenland has dimensions of about 1000 x 2000 km. For CLM we would have 20 x 40 x 10 = 8,000 columns, whereas for CISM we would have 200 x 400 = 80,000 columns. +#. Treatment of glacial melt water -#. We can use the sophisticated snow physics parameterization already in CLM instead of implementing a separate scheme for CISM. Any improvements to CLM are applied to ice sheets automatically. + a. Replaced by ice: Meltwater runs off and is immediately replaced by ice, maintaining a permanently frozen glacier column. In the absence of a dynamic ice sheet model, this treatment implicitly assumes an unlimited ice reservoir available for melting, with additional adjustments applied to maintain mass and energy conservation. This behavior is discussed further in section :numref:`Computation of the surface mass balance`. + b. Remains in place: Regions using this approach cannot compute surface mass balance (SMB), because negative SMB would be physically inconsistent in the presence of retained liquid water on top of the ice column. Although this treatment is less realistic physically, it avoids the persistent negative ice runoff required by the "replaced by ice" formulation to conserve mass and energy. This behavior is particularly useful for mountain glaciers, where atmospheric topographic smoothing can produce unrealistically warm conditions. In such cases, avoiding unrealistic negative runoff is often preferable to representing more realistic glacier physics. Section :numref:`Mountain Glaciers` provides an overview of ongoing/future work to address these limitations with mountain glaciers. -#. The atmosphere model can respond during runtime to ice-sheet surface changes (even in the absence of two-way feedbacks with CISM). As shown by :ref:`Pritchard et al. (2008)`, runtime albedo feedback from the ice sheet is critical for simulating ice-sheet retreat on paleoclimate time scales. Without this feedback the atmosphere warms much less, and the retreat is delayed. +#. Treatment of runoff from snow capping -#. The improved SMB is potentially available in CLM for all glaciated grid cells (e.g., in the Alps, Rockies, Andes, and Himalayas), not just those which are part of ice sheets. + a. Ice runoff from snow capping remains ice. This serves as a crude parameterization of iceberg calving, and is most appropriate for regions with substantial real-world calving. + b. Ice runoff from snow capping is melted, generating a negative sensible heat flux, and then routed as liquid runoff. This behavior matches that of non-glacier land units and is more appropriate in regions with little iceberg calving. It can also help avoid unrealistic ocean cooling and runaway sea ice growth. -In typical runs, CISM is not evolving; CLM computes the SMB and sends it to CISM, but CISM's ice sheet geometry remains fixed over the course of the run. In these runs, CISM serves two roles in the system: +Further detail on snow capping is provided in section :numref:`Runoff from glaciers and snow-capped surfaces`. Note that these runoff treatments are irrelevant when using an evolving, two-way-coupled ice sheet model, because the snow capping flux is transferred directly to CISM rather than routed as runoff. -#. Over the CISM domain (typically Greenland in CESM2), CISM dictates glacier areas and topographic elevations, overriding the values on CLM's surface dataset. CISM also dictates the elevation of non-glacier land units in its domain, and only in this domain are atmospheric fields downscaled to non-glacier land units. (So if you run with a stub glacier model - SGLC - then glacier areas and elevations will be taken entirely from CLM's surface dataset, and no downscaling will be done over non-glacier land units.) -#. CISM provides the grid onto which SMB is downscaled. (If you run with SGLC then SMB will still be computed in CLM, but it won't be downscaled to a high-resolution ice sheet grid.) +.. note:: + The combination of "Glacial melt = Replaced by ice" and "Ice runoff = Melted" produces strongly non-physical behavior. During glacier melt, the model generates negative ice runoff under the "Replaced by ice" treatment. Under the "Ice runoff = Melted" treatment, this negative ice runoff is converted into negative liquid runoff and a positive sensible heat flux. The resulting behavior produces zero net runoff but an artificial positive sensible heat flux associated with glacier melt. Because this behavior is physically unrealistic, CLM does not allow this combination of glacier-region settings. -It is also possible to run CESM with an evolving ice sheet. In this case, CLM responds to CISM's evolution by adjusting the areas of the glacier land unit and each elevation class within this land unit, as well as the mean topographic heights of each elevation class. Thus, CLM's glacier areas and elevations remain in sync with CISM's. Conservation of mass and energy is done as for other landcover change (see Chapter :numref:`rst_Transient Landcover Change`). +.. note:: + Note that the CLM Greenland region extends only to the Greenland Ice Sheet boundary defined by CISM. As a result, SMB is not computed by default for grid cells that lie within the broader CISM domain but outside the Greenland Ice Sheet itself (i.e., the peripheral glaciers). Presently, these non-Greenland portions of the CISM domain are treated using the same configuration as the mountain glacier regions, rather than using the Greenland ice-sheet configuration. This choice helps avoid unrealistic runoff fluxes from the Canadian Arctic Archipelago that could otherwise contribute to excessive sea ice growth in the surrounding ocean. -.. _Glacier regions: +.. note:: + Non-virtual, non-SMB-computing glacier regions can exist within the CISM domain, as is the case for portions of the Greenland CISM domain outside the Greenland Ice Sheet itself. However, these regions always provide zero SMB and cannot respond to CISM-driven changes in glacier extent. For this reason, it is generally preferable for as much of the CISM domain as possible to use virtual, SMB-computing glacier regions. -Glacier regions and their behaviors ------------------------------------ -The world's glaciers and ice sheets are broken down into a number of different regions (three by default) that differ in three respects: -#. Whether the gridcell's glacier land unit contains: +.. _Ice_Sheets: - a. Multiple elevation classes (section :numref:`Multiple elevation class scheme`) +Ice Sheets +---------- - b. Multiple elevation classes plus virtual elevation classes +CLM computes and provides two quantities that are passed to the ice sheet model: - c. Just a single elevation class whose elevation matches the atmosphere's topographic height (so there is no adjustment in atmospheric forcings due to downscaling). +#. Surface mass balance (SMB) - the net annual accumulation and ablation of mass at the upper surface (section :numref:`Computation of the surface mass balance`) -#. Treatment of glacial melt water: +#. Ground surface temperature, which serves as an upper boundary condition for CISM's temperature calculation. Ice sheet models are typically run at much higher spatial resolution than CLM (for example, :math:`\sim\ 5km` versus :math:`\sim\ 100km`). To improve the downscaling of atmospheric forcing from the CLM grid to the ice sheet grid, the glaciated portion of each CLM grid cell is divided into multiple elevation classes (section :numref:`Multiple elevation class scheme`). The CESM coupler then performs horizontal and vertical interpolation to generate high-resolution fields for CISM. + +**Static ice sheet configuration** - a. Glacial melt water runs off and is replaced by ice, thus keeping the column always frozen. In the absence of a dynamic ice sheet model, this behavior implicitly assumes an infinite store of glacial ice that can be melted (with appropriate adjustments made to ensure mass and energy conservation). This behavior is discussed in more detail in section :numref:`Computation of the surface mass balance`. +In typical simulations, CISM is run in a non-evolving configuration. In this mode, CLM computes SMB and passes it to CISM, but the ice-sheet geometry remains fixed throughout the simulation. Under this configuration, CISM serves two primary roles: - b. Glacial melt water remains in place until it refreezes - possibly remaining in place indefinitely if the glacier column is in a warm climate. With this behavior, ice melt does not result in any runoff. Regions with this behavior cannot compute SMB, because negative SMB would be meaningless (due to the liquid water on top of the ice column). This behavior produces less realistic glacier physics. However, it avoids the negative ice runoff that is needed for the "replaced by ice" behavior to conserve mass and energy (as described in section :numref:`Computation of the surface mass balance`). Thus, in regions where CLM has glaciers but the atmospheric forcings are too warm to sustain those glaciers, this behavior avoids persistent negative ice runoff. This situation can often occur for mountain glaciers, where topographic smoothing in the atmosphere results in a too-warm climate. There, avoiding persistent negative ice runoff can be more important than getting the right glacier ice physics. -#. Treatment of ice runoff from snow capping (as described in section :numref:`Runoff from glaciers and snow-capped surfaces`). Note that this is irrelevant in regions with an evolving, two-way-coupled ice sheet (where the snow capping term is sent to CISM rather than running off): +#. Defining glacier extent and topography: Within the CISM domain (typically Greenland in CESM2), CISM specifies glacier area and topographic elevation, overriding the corresponding values in the CLM surface dataset. CISM also defines the elevation of non-glacier land units within its domain. Atmospheric downscaling over non-glacier land units is applied only within the CISM domain. If the stub glacier model configuration (`SGLC `_) is used instead of CISM, glacier areas and elevations are taken entirely from the CLM surface dataset, and no atmospheric downscaling is applied over non-glacier land units. - a. Ice runoff from snow capping remains ice. This is a crude parameterization of iceberg calving, and so is appropriate in regions where there is substantial iceberg calving in reality. - b. Ice runoff from snow capping is melted (generating a negative sensible heat flux) and runs off as liquid. This matches the behavior for non-glacier columns. This is appropriate in regions that have little iceberg calving in reality. This can be important to avoid unrealistic cooling of the ocean and consequent runaway sea ice growth. +#. Providing the target grid for SMB downscaling: CISM provides the high-resolution grid onto which SMB fields are downscaled. When using SGLC, SMB is still computed in CLM, but it is not interpolated to a separate high-resolution ice sheet grid. -The default behaviors for the world's glacier and ice sheet regions are described in :numref:`Table Glacier region behaviors`. Note that the Greenland region stops at the edge of Greenland as defined by CISM. This means that, by default, SMB is not computed for grid cells outside Greenland but within the CISM domain. (This treatment of the non-Greenland portion of the CISM domain as being the same as the world's mountain glaciers rather than like Greenland itself is mainly for the sake of avoiding unrealistic fluxes from the Canadian archipelago that can potentially result in runaway sea ice growth in that region.) -.. _Table Glacier region behaviors: +**Evolving ice sheet configuration** -.. table:: Glacier region behaviors +CESM can also be run with an evolving ice sheet. In this mode, CLM responds dynamically to changes in ice-sheet geometry computed by CISM. As the ice sheet evolves, CLM updates the glacier land-unit area, elevation-class area fractions, and mean elevation of each elevation class. This ensures that glacier extent and surface topography remain consistent between CLM and CISM throughout the simulation. Conservation of mass and energy follows the same framework used for other land-cover transitions (Chapter :numref:`rst_Transient Landcover Change`). - +---------------+---------------+---------------+---------------+ - | Region | Elevation | Glacial melt | Ice runoff | - | | classes | | | - +===============+===============+===============+===============+ - | Greenland | Virtual | Replaced by | Remains ice | - | | | ice | | - +---------------+---------------+---------------+---------------+ - | Antarctica | Multiple | Replaced by | Remains ice | - | | | ice | | - +---------------+---------------+---------------+---------------+ - | All others | Single | Remains in | Melted | - | | | place | | - +---------------+---------------+---------------+---------------+ -.. note:: +.. _Multiple elevation class scheme: - It is possible to have non-virtual, non-SMB-computing areas within the CISM domain (as is the case for the portion of CISM's Greenland domain outside of Greenland itself). However, these areas will send 0 SMB and will not be able to adjust to CISM-dictated changes in glacier area. Therefore, it is best to set up the glacier regions and their behaviors so that as much of the CISM domain as possible is covered by virtual, SMB-computing areas. +Multiple elevation class scheme +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -.. note:: +The glacier land unit contains multiple columns based on surface elevation. These are known as elevation classes, and the land unit is referred to as *glacier\_mec*. As described in section :numref:`Glacier regions`, some regions have only a single elevation class, but they are still referred to as *glacier\_mec* land units. The default is to have 10 elevation classes whose lower limits are 0, 200, 400, 700, 1000, 1300, 1600, 2000, 2500, and 3000 m. Each column is characterized by a fractional area and surface elevation that are read in during model initialization, and then possibly overridden by CISM as the run progresses. Each *glacier\_mec* column within a grid cell has distinct ice and snow temperatures, snow water content, surface fluxes, and SMB. - The combination of the ``Glacial melt = Replaced by ice`` and the ``Ice runoff = Melted`` behaviors results in particularly non-physical behavior: During periods of glacial melt, a negative ice runoff is generated (due to the ``Glacial melt = Replaced by ice`` behavior); this negative ice runoff is converted to a negative liquid runoff plus a positive sensible heat flux (due to the ``Ice runoff = Melted`` behavior). The net result is zero runoff but a positive sensible heat flux generated from glacial melt. Because of how physically unrealistic this is, CLM does not allow this combination of behaviors. +The atmospheric surface temperature, potential temperature, specific humidity, density, and pressure are downscaled from the atmosphere's mean grid cell elevation to the *glacier\_mec* column elevation using a specified lapse rate (typically 6.0 deg/km) and an assumption of uniform relative humidity. Longwave radiation is downscaled by assuming a linear decrease in downwelling longwave radiation with increasing elevation (0.032 W m\ :sup:`-2` m\ :sup:`-1`, limited to 0.5 - 1.5 times the gridcell mean value, then normalized to conserve gridcell total energy) :ref:`(Van Tricht et al., 2016)`. Total precipitation is partitioned into rain vs. snow as described in Chapter :numref:`rst_Surface Characterization, Vertical Discretization, and Model Input Requirements`. The partitioning of precipitation is based on the downscaled temperature, allowing rain to fall at lower elevations while snow falls at higher elevations. -.. _Multiple elevation class scheme: +This downscaling allows lower-elevation columns to undergo surface melting while columns at higher elevations remain frozen. This gives a more accurate simulation of summer melting, which is a highly nonlinear function of air temperature. Within the CISM domain, this same downscaling procedure is also applied to all non-urban land units. The elevation of non-glacier land units is taken from the mean elevation of ice-free grid cells in CISM. This is done in order to keep the glaciated and non-glaciated portions of the CISM domain as consistent as possible. -Multiple elevation class scheme -------------------------------- +In contrast to most CLM subgrid units, *glacier\_mec* columns can be active (i.e., have model calculations run there) even if their area is zero. These are known as "virtual" columns. This is done because the ice sheet model may require a SMB for some grid cells where CLM has zero glacier area in that elevation range. Virtual columns also facilitate glacial advance and retreat in the two-way coupled case. Virtual columns do not affect energy exchange between the land and the atmosphere. -The glacier land unit contains multiple columns based on surface elevation. These are known as elevation classes, and the land unit is referred to as *glacier\_mec*. (As described in section :numref:`Glacier regions`, some regions have only a single elevation class, but they are still referred to as *glacier\_mec* land units.) The default is to have 10 elevation classes whose lower limits are 0, 200, 400, 700, 1000, 1300, 1600, 2000, 2500, and 3000 m. Each column is characterized by a fractional area and surface elevation that are read in during model initialization, and then possibly overridden by CISM as the run progresses. Each *glacier\_mec* column within a grid cell has distinct ice and snow temperatures, snow water content, surface fluxes, and SMB. In CLM6 users can optionally specify using :ref:`Sturm et al. (1997)` or :ref:`Jordan (1991)` parameterizations for snow thermal conductivity over glacier land units (see Chapter :numref:`rst_Soil and Snow Temperatures`), with Sturm (1997) set as the default. +.. _Computation of the surface mass balance: -The atmospheric surface temperature, potential temperature, specific humidity, density, and pressure are downscaled from the atmosphere's mean grid cell elevation to the *glacier\_mec* column elevation using a specified lapse rate (typically 6.0 deg/km) and an assumption of uniform relative humidity. Longwave radiation is downscaled by assuming a linear decrease in downwelling longwave radiation with increasing elevation (0.032 W m\ :sup:`-2` m\ :sup:`-1`, limited to 0.5 - 1.5 times the gridcell mean value, then normalized to conserve gridcell total energy) :ref:`(Van Tricht et al., 2016)`. Total precipitation is partitioned into rain vs. snow as described in Chapter :numref:`rst_Surface Characterization, Vertical Discretization, and Model Input Requirements`. The partitioning of precipitation is based on the downscaled temperature, allowing rain to fall at lower elevations while snow falls at higher elevations. +Surface mass balance +~~~~~~~~~~~~~~~~~~~~~ -This downscaling allows lower-elevation columns to undergo surface melting while columns at higher elevations remain frozen. This gives a more accurate simulation of summer melting, which is a highly nonlinear function of air temperature. +Computing SMB in CLM rather than directly within CISM provides several advantages: -Within the CISM domain, this same downscaling procedure is also applied to all non-urban land units. The elevation of non-glacier land units is taken from the mean elevation of ice-free grid cells in CISM. This is done in order to keep the glaciated and non-glaciated portions of the CISM domain as consistent as possible. +#. Computational efficiency: SMB can be computed much more efficiently in CLM using a limited number of elevation classes than on the full high-resolution CISM grid. For example, consider a simulation with CLM at 50 km resolution and CISM at 5 km resolution. The Greenland Ice Sheet spans roughly 1000 x 2000 km, corresponding to approximately: -In contrast to most CLM subgrid units, glacier\_mec columns can be active (i.e., have model calculations run there) even if their area is zero. These are known as "virtual" columns. This is done because the ice sheet model may require a SMB for some grid cells where CLM has zero glacier area in that elevation range. Virtual columns also facilitate glacial advance and retreat in the two-way coupled case. Virtual columns do not affect energy exchange between the land and the atmosphere. + - 20 x 40 x 10 (elevation classes) = 8,000 glacier elevation-class columns in CLM + - 200 x 400 = 80,000 grid cells in CISM -.. _Computation of the surface mass balance: +#. Shared snow and surface physics: CLM already contains a sophisticated snow and surface energy balance parameterization. Computing SMB within CLM avoids the need to implement and maintain a separate SMB scheme within CISM, while ensuring that improvements to CLM physics are automatically applied to ice sheets. + +#. Atmosphere - ice sheet feedbacks: Computing SMB in CLM allows the atmosphere model to respond interactively to changes in ice-sheet surface properties, even without fully evolving ice-sheet geometry. As shown by :ref:`Pritchard et al. (2008)`, interactive albedo feedbacks are critical for realistic simulations of long-term ice-sheet retreat. -Computation of the surface mass balance ---------------------------------------- +#. Consistency across glacier types: Improvements to SMB calculations in CLM are potentially applicable to all glaciated grid cells, including mountain glaciers, and not only to continental ice sheets. -This section describes the computation of surface mass balance and associated runoff terms. The description here only applies to regions where glacial melt runs off and is replaced by ice, not to regions where glacial melt remains in place. Thus, by default, this only applies to Greenland and Antarctica, not to mountain glaciers elsewhere in the world. (See also section :numref:`Glacier regions`.) -The SMB of a glacier or ice sheet is the net annual accumulation/ablation of mass at the upper surface. Ablation is defined as the mass of water that runs off to the ocean. Not all the surface meltwater runs off; some of the melt percolates into the snow and refreezes. Accumulation is primarily by snowfall and deposition, and ablation is primarily by melting and evaporation/sublimation. CLM uses a surface-energy-balance (SEB) scheme to compute the SMB. In this scheme, the melting depends on the sum of the radiative, turbulent, and conductive fluxes reaching the surface, as described elsewhere in this document. +**Computation of the surface mass balance** -Note that the SMB typically is defined as the total accumulation of ice and snow, minus the total ablation. The SMB flux passed to CISM is the mass balance for ice alone, not snow. We can think of CLM as owning the snow, whereas CISM owns the underlying ice. Fluctuations in snow depth between 0 and 10 m water equivalent are not reflected in the SMB passed to CISM. In transient runs, this can lead to delays of a few decades in the onset of accumulation or ablation in a given glacier column. +This section describes the computation of SMB and associated runoff terms. The discussion applies only to glacier regions where meltwater runs off and the lost ice is immediately replaced, rather than to regions where meltwater remains within the glacier column. By default, this treatment applies to the Greenland and Antarctic Ice Sheets, but not to mountain glaciers. -SMB is computed and sent to the CESM coupler regardless of whether and where CISM is operating. However, the effect of SMB terms on runoff fluxes differs depending on whether and where CISM is evolving in two-way-coupled mode. This is described by the variable *glc\_dyn\_runoff\_routing*. (This is real-valued in the code to handle the edge case where a CLM grid cell partially overlaps with the CISM grid, but we describe it as a logical variable here for simplicity.) In typical cases where CISM is not evolving, *glc\_dyn\_runoff\_routing* will be false everywhere; in these cases, CISM's mass is not considered to be part of the coupled system. In cases where CISM is evolving and sending its own calving flux to the coupler, *glc\_dyn\_runoff\_routing* will be true over the CISM domain and false elsewhere. +The SMB of a glacier or ice sheet is defined as the net annual mass gain or loss at the upper surface. Accumulation occurs primarily through snowfall and deposition, while ablation occurs primarily through melting and evaporation/sublimation. Ablation is defined here as the mass of water that ultimately runs off to the ocean. Not all surface meltwater contributes directly to runoff; some meltwater percolates into the snowpack and refreezes. + +CLM computes SMB using a surface energy balance (SEB) approach, in which melt depends on the combined radiative, turbulent, and conductive energy fluxes at the surface. In glaciology, SMB is typically defined as the net balance of both snow and ice accumulation and ablation. However, the SMB flux passed from CLM to CISM represents the mass balance of the underlying ice only, excluding transient changes in snow storage. Conceptually, CLM can be viewed as owning the snowpack, while CISM owns the underlying glacier ice. As a result, fluctuations in snow depth between 0 and 10 m water equivalent are not reflected in the SMB passed to CISM. In transient simulations, this treatment can delay the onset of accumulation or ablation signals in a glacier column by several decades. + +SMB is computed and sent to the CESM coupler regardless of whether and where CISM is operating. However, the effect of SMB terms on runoff fluxes differs depending on whether and where CISM is evolving in two-way-coupled mode. This is described by the variable *glc\_dyn\_runoff\_routing* (this is real-valued in the code to handle the edge case where a CLM grid cell partially overlaps with the CISM grid, but we describe it as a logical variable here for simplicity.) In typical cases where CISM is not evolving, *glc\_dyn\_runoff\_routing* will be false everywhere; in these cases, CISM's mass is not considered to be part of the coupled system. In cases where CISM is evolving and sending its own calving flux to the coupler, *glc\_dyn\_runoff\_routing* will be true over the CISM domain and false elsewhere. Any snow capping (section :numref:`Runoff from glaciers and snow-capped surfaces`) is added to :math:`q_{ice,frz}`. Any liquid water (i.e., melted ice) below the snow pack in the glacier column is added to :math:`q_{ice,melt}`, then is converted back to ice to maintain a pure-ice column. Then the total SMB is given by :math:`q_{ice,tot}`: @@ -145,7 +150,7 @@ Any snow capping (section :numref:`Runoff from glaciers and snow-capped surfaces CLM is responsible for generating glacial surface melt, even when running with an evolving ice sheet. Thus, :math:`q_{ice,melt}` is always added to liquid runoff (:math:`q_{rgwl}`), regardless of *glc\_dyn\_runoff\_routing*. However, the ice runoff flux depends on *glc\_dyn\_runoff\_routing*. If *glc\_dyn\_runoff\_routing* is true, then CISM controls the fate of the snow capping mass in :math:`q_{ice,frz}` (e.g., eventually transporting it to lower elevations where it can be melted or calved). Since CISM will now own this mass, the snow capping flux does *not* contribute to any runoff fluxes generated by CLM in this case. -If *glc\_dyn\_runoff\_routing* is false, then CLM sends the snow capping flux as runoff, as a crude representation of ice calving (see also sections :numref:`Runoff from glaciers and snow-capped surfaces` and :numref:`Glacier regions`). However, this ice runoff flux is reduced by :math:`q_{ice,melt}`. This reduction is needed for conservation; its need is subtle, but can be understood with either of these explanations: +If *glc\_dyn\_runoff\_routing* is false, then CLM sends the snow capping flux as runoff, as a crude representation of ice calving. However, this ice runoff flux is reduced by :math:`q_{ice,melt}`. This reduction is needed for conservation; its need is subtle, but can be understood with either of these explanations: - When ice melts, we let the liquid run off and replace it with new ice. That new ice needs to come from somewhere to keep the coupled system in water balance. We "request" the new ice from the ocean by generating a negative ice runoff equivalent to the amount we have melted. @@ -155,3 +160,15 @@ For a given point in space or time, this reduction can result in negative ice ru In regions where SMB is computed for glaciers, SMB is also computed for the natural vegetated land unit. Because there is no ice to melt in this land unit, it can only generate a zero or positive SMB. A positive SMB is generated once the snow pack reaches its maximum depth. When running with an evolving ice sheet, this condition triggers glacial inception. +.. _Mountain Glaciers: + +Mountain Glaciers +----------------- + +Beginning with CLM5.2, mountain glacier outlines are derived from version 6 of the Randolph Glacier Inventory (RGI). + +As discussed earlier in this chapter, the current representation of mountain glaciers in CLM remains relatively limited compared to the treatment of continental ice sheets. By default, mountain glaciers are treated using a simplified single-elevation-class configuration, and SMB is not computed for these regions. This limitation arises from the glacier-region configuration described in section :numref:`Glacier regions`, where meltwater is retained within the glacier column rather than running off and being replaced by ice. + +This treatment is primarily intended to avoid unrealistic runoff fluxes in regions where coarse atmospheric topography produces climates that are too warm to sustain glaciers realistically. Such issues are particularly important for mountain glaciers, where strong local elevation gradients are poorly resolved at typical climate-model resolutions. + +Ongoing development efforts are focused on extending CLM's mountain glacier capabilities. In particular, the hillslope hydrology configuration (`Chapter Hillslope_Hydrology `_) is being adapted to support SMB calculations for mountain glaciers . These developments are intended to enable coupling between CLM and CISM for mountain glacier applications in a manner similar to the current treatment of ice sheets, leveraging recent CISM capabilities for simulating mountain glacier dynamics :ref:`(Minallah and Lipscomb et al., 2025) `. diff --git a/doc/source/tech_note/References/CLM50_Tech_Note_References.rst b/doc/source/tech_note/References/CLM50_Tech_Note_References.rst index b8daafe939..0f53a83d85 100644 --- a/doc/source/tech_note/References/CLM50_Tech_Note_References.rst +++ b/doc/source/tech_note/References/CLM50_Tech_Note_References.rst @@ -1053,6 +1053,10 @@ Miller, J.R., Russell, G.L., and Caliri, G. 1994. Continental-scale river flow i Millington, R. and Quirk, J.P., 1961. Permeability of Porous Solids. Transactions of the Faraday Society 57:1200-1207. +.. _Minallah2025: + +Minallah, S., Lipscomb, W. H., Leguy, G., and Zekollari, H. 2025. A framework for three-dimensional dynamic modeling of mountain glaciers in the Community Ice Sheet Model (CISM v2.2), Geosci. Model Dev., 18, 5467–5486, https://doi.org/10.5194/gmd-18-5467-2025. + .. _Mironovetal2010: Mironov, D. et al., 2010. Implementation of the lake parameterisation scheme FLake into the numerical weather prediction model COSMO. Boreal Environment Research 15:218-230.