@@ -110,12 +111,15 @@ The length of the time step should be carefully considered in assembling the for
**Note** The meteorological forcing data can be either given to the model at the physics timestep (e.g. site-level forcing at 30 mins for a physics timestep of 30 mins), or at a coarser temporal resolution (e.g. 6 hourly for a physics timestep of 30 minutes). If the forcing is the same temporal resolution as the physics, then it is used as read-in. If it is coarser, CLASSIC will do disaggregation following the logic in @ref metModule.f90.
# Geophysical model input fields {#geophysData}
# Input Vegetation Data {#vegetationData}
CLASSIC has several mandatory input fields
## Input Vegetation Data {#vegetationData}
CLASSIC can be run with either dynamic vegetation (CTEM+CLASS) or a physics only simulation (CLASS). The model inputs differ between the two configurations with some input vegetation data for a CLASS-only run ignored when CTEM is turned on. As well CTEM requires some additional inputs not needed for CLASS-only runs as described below.
## Required vegetation data {#vegCLASSonly}
### Required vegetation data {#vegCLASSonly}
The typical main vegetation categories for the model physics (CLASS) are needleleaf trees, broadleaf trees, broadleaf shrubs, crops and grass (e.g. ican = 5). However this is adaptable and can include further PFTs depending on model configuration. There are checks in the model code that look for known PFTs (e.g. 'NdlTr', 'BdlTr', 'BdlSh', 'Crops', 'Grass'). If a PFT is introduced that is not known, the model will bail and let the user know where the PFT check failed. This is designed to prevent poorly considered additions and ensure developers are aware of where the different PFTs branch in the code. Urban areas are also treated as “vegetation” in the CLASS code, and have associated values for *FCAN*, *ALVC*, *ALIC* and *LNZ0* (see below). Thus these arrays have a larger dimension of `ican + 1` rather than `ican`.
@@ -132,7 +136,7 @@ For each of the CLASS PFTs the following data are required for each mosaic tile
8.**ROOT** Annual maximum rooting depth of vegetation category [m]
### Variables not presently read in
#### Variables not presently read in
These variables are not presently read-in by CLASSIC but could be if desired. Model code changes would be required.
@@ -167,7 +171,7 @@ For the non-required stomatal resistance parameters, typical values for the four
\end{array}
\f]
## Required vegetation data for a biogeochemical simulation (CLASS+CTEM) {#vegCTEMtoo}
### Required vegetation data for a biogeochemical simulation (CLASS+CTEM) {#vegCTEMtoo}
In addition to the CLASS variables described above, CTEM requires the following further information about the vegetation:
@@ -175,7 +179,7 @@ In addition to the CLASS variables described above, CTEM requires the following
If land use is being simulated this value will come from a land use change file (see @ref inputLUC) otherwise (if in the job options file: lnduseon = .false. and fixedYearLUC = -9999) it is taken from the model initialization file and kept constant thoughout a run (provided competition between PFTs is not turned on).
title="Impact of topography and meteorological forcing on snow simulation
in the Canadian Land Surface Scheme Including Biogeochemical
Cycles ({CLASSIC})",
author="Wang, Libo and Mudryk, Lawrence and Melton, Joe R and Mortimer,
Colleen and Cole, Jason and Meyer, Gesa and Bartlett, Paul and
Lalande, Micka{\"{e}}l",
journal="EGUsphere",
pages="1--37",
abstract="Abstract. Our study evaluates the impacts of an alternate snow
cover fraction (SCF) parameterization on snow simulation in the
Canadian Land Surface Scheme Including Biogeochemical Cycles
(CLASSIC). Three reanalysis-based meteorological datasets are used
to drive the model to account for uncertainties in the forcing
data. While the default parameterization assumes a simple linear
relationship between SCF and snow depth with no dependence on
topography, the alternate parameterization accounts for the
topographic effects of sub-grid terrain on SCF. We show that the
alternate parameterization improves SCF simulated in CLASSIC
during winter and spring in mountainous areas for all three
choices of meteorological datasets. Annual mean bias, unbiased
root mean squared area, and correlation improve by 75 \%, 32 \%,
and 7 \% when evaluated with MODIS SCF observations over the
Northern Hemisphere. We also demonstrate that the improvements to
simulated SCF lead to further improvements in variables related to
surface radiation, energy fluxes, and the water cycle. Finally, we
link relative biases in the meteorological forcing data to
differences in simulated snow water equivalent and SCF. Assessment
of simulations with different combinations of SCF
parameterizations and meteorological datasets reveals the large
impact of meteorological forcing on snow simulation in CLASSIC.
Two out of the three meteorological datasets were bias-adjusted
using observation-based datasets. However, simulations forced by
the dataset without bias correction outperform relative to
simulations forced by datasets with bias correction, suggesting
that there are large uncertainties in the observation-based
datasets and/or methods used for bias correction. This study
underscores the importance of accounting for topographic effects
of sub-grid terrain and accurate meteorological forcing on snow
simulation in land surface models.",
month=mar,
year=2025,
doi="10.5194/egusphere-2025-1264",
language="en"
}
@ARTICLE{Gauthier2024-nx,
@ARTICLE{Gauthier2025-vw,
title="Parameter optimization for global soil carbon simulations: Not a
simple problem",
author="Gauthier, Charles B and Melton, Joe R and Meyer, Gesa and S, Raj
Deepak and Sonnentag, Oliver",
journal="ESS Open Archive",
abstract="Accurate simulation of soil organic carbon (SOC) dynamics by
terrestrial biosphere models is hampered by poorly constrained
author="Gauthier, Charles B and Melton, Joe R and Meyer, Gesa and Raj
Deepak, S N and Sonnentag, Oliver",
journal="Journal of advances in modeling earth systems",
publisher="American Geophysical Union (AGU)",
volume=17,
number=8,
pages="e2024MS004577",
abstract="AbstractAccurate simulation of soil organic carbon (SOC) dynamics
by terrestrial biosphere models is hampered by poorly constrained
parameters and parameter equifinality, amongst other issues. To
address this, we use Bayesian optimization to constrain the 16
SOC-related parameters in the Canadian Land Surface Scheme
Including biogeochemical Cycles (CLASSIC). We employed a global
sensitivity analysis (Sobol') to develop four parameter sets based
upon different sensitivity criteria. We then optimized each set
against observed SOC (World Soil Information Service; WoSIS) and
soil respiration (Soil Respiration Database; SRDB). Using two
sensitivity analysis (Sobol') to develop four parameter sets
based upon different sensitivity criteria. We then optimized each
set against observed SOC (World Soil Information Service; WoSIS)
and soil respiration (Soil Respiration Database; SRDB). Using two
different loss functions; one focused on reproducing the
observational mean value, and the other explicitly accounting for
an estimated observational uncertainty. The best optimized
parameter sets from each loss function had an average relative
difference of 61\%. Thus the choice of loss function impacts what
parameter values are deemed optimal and should be considered
carefully. The final selected optimal parameter set saw a 12\%
improvement against WoSIS and SRDB, had global SOC totals in line
with literature estimates, and better simulated high-latitude SOC
stocks evaluated against the Northern Circumpolar Soil Carbon
Database (RMSD: 16.39 vs. 17.61; bias: -5.57 vs. -10.78 kg C m2)
compared to the default CLASSIC parameters. However, some
parameters were not well constrained, in particular those of
needle-leaf deciduous trees which dominate Siberian boreal
forests, a region relatively poorly observed in WoSIS and SRDB.
Future work should apply further constraints on the optimization
framework and address observational gaps.",
month=jul,
year=2024,
doi="10.22541/essoar.172167442.20510812/v1"
parameter sets for each loss function had an average relative
difference of 61\%. Thus, the choice of loss function impacts
what parameter values are deemed optimal and should be considered
carefully. The final set of selected optimal parameters saw a
12\% improvement against WoSIS and SRDB, had global SOC totals in
line with literature estimates, and better simulated
high-latitude SOC stocks evaluated against the Northern
Circumpolar Soil Carbon Database (RMSD: 16.39 vs. 17.61; bias:
-5.57 vs. -10.78 kg C ) compared to the default CLASSIC
parameters. However, some parameters were not well constrained,
in particular those of needle-leaf deciduous trees that dominate
the Siberian boreal forests, a region relatively poorly observed
in WoSIS and SRDB. Future work should apply further constraints
on the optimization framework and address observational gaps.",
month=aug,
year=2025,
keywords="soil organic carbon; land surface model; Bayesian optimization;
sensitivity analysis; soil respiration; CLASSIC",
doi="10.1029/2024ms004577",
issn="1942-2466",
language="en"
}
@ARTICLE{MacKay2022-do,
title="On the Discretization of Richards Equation in Canadian Land
Surface Models",
@@ -4500,24 +4555,6 @@
doi = "10.1029/2018MS001490"
}
@ARTICLE{cite Curasi2022-ss,
title = "Evaluating the performance of the Canadian Land Surface Scheme Including Biogeochemical Cycles (CLASSIC) tailored to the pan-Canadian domain",
author = "Curasi, S.R., Melton, J.R., Humphreys, E.R., Wang, L., Seiler, C., Cannon, A., Chan, E. and Qu, B.",
abstract = "Canada's boreal forests and tundra ecosystems are responding to unprecedented climate change with implications for the global carbon (C) cycle and
global climate. However, our ability to model the response of Canada's terrestrial ecosystems to climate change is limited and there has been no comprehensive,
process-based assessment of Canada's terrestrial C cycle. We tailor the Canadian Land Surface Scheme Including Biogeochemical Cycles (CLASSIC) to Canada and evaluate
its C cycling performance against independent reference data. We utilize skill scores to assess model performance against reference data alongside benchmark scores that
quantify the level of agreement between the reference data sets to aid in interpretation. Our results demonstrate CLASSIC's sensitivity to prescribed vegetation cover.
They also show that the addition of five region-specific PFTs improves CLASSIC's skill at simulating the Canadian C cycle. CLASSIC performs well when tailored to Canada,
falls within the range of the reference data sets, and meets or exceeds the benchmark scores for most C cycling processes. New region-specific land cover products, well-informed
plant functional type (PFT) parameterizations, and more detailed reference data sets will facilitate improvements to the representation of the terrestrial C cycle
in regional and global land surface models (LSMs). Incorporating a parameterization for boreal disturbance processes and explicitly representing peatlands and permafrost soils
will improve CLASSIC's future performance in Canada and other boreal regions. This is an important step toward a comprehensive process-based assessment of Canada's terrestrial
C cycle and evaluating Canada's net C balance under climate change.",
journal = "ESSOAr",
year = 2022,
doi = "https://doi.org/10.1002/essoar.10512727.1"
}
@ARTICLE{Canada_Fire_Danger_Group1992-bt,
title = "Development and structure of the Canadian forest fire behavior
