Aihui Wang, Kaiyuan Li, and Dennis P. Lettenmaier

1
Integration of the VIC model soil hydrology
scheme into CLM
Aihui Wang, Kaiyuan Li, and Dennis P.
Lettenmaier Department of Civil and
Environmental Engineering, University of
Washington
5. Model evaluations Comparison of model
simulated streamflow with observation over two
large river basins
Monthly averaged snow water equivalent (SWE)
over a grid cell (43.4?N, -110.2?W)
1. Introduction Land surface models
predict energy and moisture fluxes at the land
surface for coupled land-atmosphere models. Of
the various processes parameterized in such
models, representation of soil hydrology is
particularly important since it affects both the
surface water budget, and the surface energy
budget through the evaporation feedback. In order
to improve its land surface hydrology
representation, a new soil hydrology scheme for
use in the NCAR Community Land Model (CLM) is
proposed. The new scheme incorporates the
fundamental principles and concepts of the
three-layer Variable Infiltration Capacity (VIC)
soil moisture generation scheme, as well as its
surface runoff and base flow schemes. The
modified version of CLM makes direct use of VIC
soil parameters that have been developed for
off-line regional, continental, and global
simulations. The performance of the new model and
CLM are evaluated through a comparison with
observations. Overall, the results show that the
new model better reproduces observed soil
hydrological variability, in particular, the
seasonal evolution and amplitude of soil moisture
as compared to CLM. The river basin simulations
show that in the new model, total runoff is
generally less than in CLM, and better agrees
with observations. Due to the interaction of
runoff and soil moisture, the simulation of total
evapotranspiration (or latent heat) is also
improved in the new model.
Comparison of modeled water and energy fluxes at
tower flux sites
a. Abracos

• CLM-VIC captures the magnitude of soil moisture
  over the wet seasons, although the simulated soil
  is slightly wetter than observed in the dry
  season, while the soil in CLM3 is too dry, and
  the evolution of soil moisture has much less
  seasonal variation than the observations.
• CLM-VIC simulated ET is closer to observed
  whereas CLM3 in general underestimates ET. The
  timing of base flow in CLM-VIC is consistent with
  the timing of the maximum soil moisture storage,
  whereas in CLM3 it is lagged by about two months.
  
• The net radiation is well simulated in both
  models. CLM-VIC does a better job in the
  simulation of laten heat than the CLM, which also
  reflects CLM-VIC simulated ET better than CLM.

2. Soil hydrology scheme descriptions

• CLM
• Surface runoff scheme is based on TopModel and
  BATs (Beven and Kirkby, 1979 , Dickinson et al.
  1993),and is contributed by soil water from the
  top 3 layers.
• Baseflow is from the soil water and lateral
  runoff in 6-9 layers and the 10th layer drainage.
• VIC
• Surface runoff is controlled by the infiltration
  capacity formulation and its water comes from the
  top two layers soil water.
• Baseflow follows a nonlinear curve whose source
  is the third soil layer (Liang et al. 1994).

CLM3
VIC

• The RMSEs and relative biases from CLM-VIC are
  general smaller than that from standard CLM,
  which confirm the improved performance of the new
  model for the larger drainage areas.

• CLM-VIC modeled soil moisture is in overall
  agreement with observation of both seasonal
  variation and amplitude, beside from slight
  underestimates in the wet season, while the soil
  in CLM is too dry and soil moisture variations
  are too small.
• Both models simulate net radiation well, but
  overestimate the sensible heat and underestimate
  latent heat. CLM-VIC simulated latent heat is
  better than in CLM, which corresponds to better
  simulation of ET in CLM-VIC.
• Note, the poor simulation of soil heat fluxes
  could be the result of mischaracterization of the
  physics of soil heat in the model or the
  different representations of the measured and
  modeled soil heat.

b. Hapex-Mobilhy
20cm
Arkansas-Red River basin
1.5m

• 3. Methodology
• Implement VIC soil moisture generation scheme,
  as well as its surface runoff and base flow
  scheme into CLM3 remain all other processes
  (e.g., vegetation related process, soil thermal
  process) unchanged.
• The CLM3 10-soil layer is divided into 3 layers
  which match the VIC layer depths, and all VIC
  soil parameters are read into CLM (soil hydraulic
  conductivity, base flow parameters, etc).
• Compute soil moisture, runoff, and base flow in
  the VIC scheme. The simulated soil moisture were
  compared with flux tower measurement. The
  simulated runoff pluses base flow was routed to
  the location of naturalized flow records using a
  routing algorithm, and then compared with
  observation.

• CLM tends largely overestimates runoff. The
  simulated streamflow by CLM-VIC is closer
  observation than in CLM, even though the
  simulated streamflow peck is still somewhat
  higher than the observed.
• CLM-VIC performs better in some downstream areas
  such as Little Rock and Shreveport.

• Summary
• The Variable Infiltration Capacity model (VIC)
  soil hydrology scheme has been incorporated into
  the NCAR Community Land Model (CLM3). The new
  model makes direct use of VIC soil parameters
  that have been developed for off-line regional,
  continental, and global simulations. The new
  model was evaluated using streamflow data from
  two large river basins, as well as the surface
  flux over tower observations.
• The new model improves the soil hydrology
  representation and in turn surface moisture and
  energy fluxes, especially in the reproduction of
  streamflow and soil moisture.
• The variation and amplitude of soil moisture in
  the new model are generally in closer agreement
  with observations.
• The new model produced streamflows that are
  generally smaller than those from CLM3 and in
  closer agreement with observations. Due to the
  interaction of runoff and soil moisture, the
  simulation of total evapotranspiration (or latent
  heat) is also improved in the new model.

c. Valdai

Colorado River Basin

• 4. Data descriptions and experiment designs
• Two large river basins
• Red-Arkansas River Basin

• Three fluxe towers
• a. Abracos (10.1?S, 61.9?W) a low latitude
  pasture clearing in the Amazon rain forest site
• b. Hapex-mobilhy (43.7?N, 0.1?W) a
  mid-latitude agricultural site
• Valdai (57.6?N, 33.1?W) a high-latitude
  grassland site

• Similar to other sites, CLM-VIC simulated soil
  moisture has more variation and is closer to
  observations than simulated by CLM3.
• The CLM-VIC simulated ET is in closer agreement
  with the observations than that from CLM.
• Both model-simulated snow depths are in general
  agreement with observations, although the maximum
  snow depth is overestimated in some years by both
  models and the snow melts too early in spring in
  both models.

Acknowledgements The research reported herein
was supported by the U.S. Department of Energy
under DOE Agreement Number DE-FG02-04ER63873 to
the University of Washington.
Over two river basins, meteorology forcing data
are from North American Land Data Assimilation
System (NLDAS) with a resolution of 1/8? (Maurer
et al. 2002), soil and vegetation parameters are
also from Maurer et al. Using the Lohmann et al.
(1998) routing algorithm at a daily time step.
The simulated runoff and bas flow were routed
onto the gauge location and compared with
observed naturalized streamflow. Over the flux
tower sites, the forcings and model setup follows
the PILPS project. The simulated surface soil
hydrological variables and fluxes data are
compared with the available observation The CLM3
with the VIC soil hydrology scheme is referred as
CLM-VIC, and the standard CLM3 is referred as
CLM.
Colorado River Basin

• CLM-VIC performs reasonably well in reproducing
  observed streamflow, while CLM3 overestimates
  seasonal peak stream-flow during the entire
  comparison period for all stations.
• CLM-VIC performance is quite similar to the VIC
  model in its reproduction of streamflow.
• The runoff in the Colorado basin is mainly
  contributed by snowmelt water. A consistent phase
  shift exists in the CLM simulations, that is, the
  runoff peck appears about one month early in
  CLMs simulation. The reason is most likely a
  bias toward early snow melt in CLM relative to
  VIC. The below figure compares snow water
  equivalent (SWE) from both CLM and VIC over a
  grid cell (43.4?N, -110.2?W) in the upper portion
  of the basin, where most snowmelt originates.

References Wang A., K. Li, and D.P. Lettenmaier
(2007), Integration of the Variable Infiltration
Capacity (VIC) model soil hydrology scheme into
the Community Land Model (CLM), J. Geophys. Res.
(in review). Liang, X., D. P. Lettenmaier, E. F.
Wood, and S. J. Burges (1994), A simple
hydrologically based model of land surface water
and energy fluxes for general circulation models.
J. Geophys. Res., 99, 14,415-14,428. Oleson K.
W., and coauthors (2004), Technical description
of the community land model (CLM), NCAR Technical
Note (NCAR/TN-461STR), 174pp