LandSurface Schemes in Global Climate Models

1
Land-Surface Schemes in Global Climate Models

• David Lawrence
• NCAR / Climate and Global Dynamics Division
• Climate Change Research and Terrestrial Sciences
  Sections
• Boulder, CO, USA
• With thanks to Gordon Bonan, Keith Oleson, and
  Andrew Slater

2
Outline

• Roles of land-surface schemes (LSS) in Global
  Climate Models (GCMs)
• Features of a state-of-the-art LSS (historical
  context)
• NCARs Community Land Model (CLM3)
• Hadley Centres Met Office Surface Exchange
  Scheme (MOSES2)
• Priorities for further improvement
• Case study in model development
• Amazon hydrology

3
IPCC Multi-Model Dataset

• Simulations in support of the latest round of
  IPCC reports (AR4, 2007) completed in December
  2004 and submitted to PCMDI in March 2005
• 17 Participating models from 12 international
  research centers
• NCAR CCSM3 alone is providing 100 Terabytes of
  data at T85 resolution (1.4o x 1.4o)
• 8 ensemble members
• 20th C simulations, multiple IPCC scenario runs,
  etc (10,000 simulation years total)
• some daily and extremes data
• Open for anyone in the international community to
  analyze and utilize in their research

4
Roles of Land-Surface Scheme within a GCM

• Provides the boundary conditions at the
  land-atmosphere interface
• e.g. surface albedo, vegetation cover, canopy
  height
• Partitions rainfall into runoff and evaporation.
• Evaporation provides surface-atmosphere moisture
  flux
• River runoff provides freshwater input to the
  oceans
• Partitions available energy at the surface into
  sensible and latent heat flux components
• Updates state variables which affect these
  partitionings
• e.g. snow cover, soil moisture, soil temperature
  .vegetation cover
• LSS cost is actually not that high ( 10 of full
  coupled model)

5
Roles of Land-Surface Scheme in a GCM

• Another way of putting it is that the
  land-surface scheme solves (at each timestep)
• Surface energy balance (and other energy
  balances, e.g. in canopy, snow, across soil
  layers)
• S? L? S? L? ?E H G
• S?, S? are down(up)welling solar radiation,
• L?, L? are up(down)welling longwave radiation,
• ? is latent heat of vaporization, E is
  evaporation,
• H is sensible heat flux, and G is ground heat
  flux
• Surface water balance (and other water balances
  such as snow balance, sub-surface water balance)
• P ES ET EC RSurf RSub-Surf
  ?SM / ?t
• P is rainfall,
• ES is soil evaporation, ET is transpiration, EC
  is canopy evaporation,
• RSurf is surface runoff, RSub-Surf is
  sub-surface runoff, and
• ?SM / ?t is the change in soil moisture over a
  timestep

6
Land-Surface Scheme ... Schematic Water balance
P E R ?SM/?t
?, T, zo
7
Influence on Climate and Climate Change
Simulations

• The way in which the land-surface is represented
  by such schemes is known to affect both the
  projections of climate change for the 21st
  century and the accuracy of numerical weather
  forecasts.
• As a result a great deal of research effort over
  the last two decades has been devoted to
  improving land-surface schemes for both
  applications.

Gedney and Cox, 2003
8
Evolution of Land-Surface Schemes

• To understand current LSSs, it is useful to
  consider the history of their development
• Review papers
• Pitman, A.J., 2003 The evolution of, and
  revolution in, land surface schemes designed for
  climate models. Int. J. Climatology, 23,
  479-510.
• Sellers et al., 1997 Modeling the exchanges of
  energy, water and carbon between continents and
  the atmosphere. Science, 275, 502-509.

9
History of Land-Surface Schemes Bucket Model
Figure courtesy G. Bonan
10
History of Land-Surface Schemes
Vegetation and Hydrologic Cycle

• Dickinson et al. (1986) NCAR/TN-275STR
• Sellers et al. (1986) J. Atmos. Sci. 43505-531

Figure courtesy G. Bonan
11
History of Land-Surface Schemes
Stomatal Gas Exchange
Plant physiological controls on
evapotranspiration Function of solar radiation,
humidity deficit, soil moisture, CO2,
temperature
Stomatal Gas Exchange
CO2
H2O
Leaf cuticle
Guard cell
Guard cell
Photosynthetically active radiation
Chloroplast
Bonan (1995) JGR 1002817-2831 Denning et al.
(1995) Nature 376240-242 Denning et al. (1996)
Tellus 48B521-542, 543-567 Cox (1999)
Figure courtesy G. Bonan
12
Recent Advances (last 10 years or so)

• Subgrid-scale surface hetereogeneity
• Land cover
• Soil moisture
• Lateral flow (rivers)
• Vegetation / Ecosystem dynamics

13
Biome Representation of Land Cover (IGBP)
14
MOSES 2 - Surface Tiling Fractional Surface
Cover (derived from IGBP, 9 total surface types)
and Ice Lake Urban
Essery et al. 2003
15
CLM Surface Dataset
Fluxes calculated separately for each tile and
aggregated for delivery to atmosphere model
16
Plant Functional Type Parameters (CLM3)

• Optical properties (visible and near-infrared)
• Leaf angle
• Leaf reflectance
• Stem reflectance
• Leaf transmittance
• Stem transmittance
• Land-surface models are parameter heavy!!!

• Morphological properties
• Leaf area index (annual cycle)
• Stem area index (annual cycle)
• Leaf dimension
• Canopy height
• Root distribution
• Photosynthetic parameters
• Vmax25 (maximum carboxylation at 250C)
• quantum efficiency (mmol CO2 mmol photon-1)
• m (slope of conductance-photosynthesis
  relationship)

17
Soil Properties (CLM)

• Soil parameters are derived from sand / clay
  percentage which is specified geographically and
  by soil level
• Soil moisture concentration at saturation
• Soil moisture concentration at wilting point
• Hydraulic conductivity at saturation
• Saturated soil suction
• Thermal conductivity
• Thermal capacity

18
Tropical Devegetation Sensitivity to Soil
Parameterization (HadAM3)
19
Subgrid-Scale Soil Moisture Heterogeneity

• Soil wetness varies considerably over areas the
  size of a model gridbox. Hilly areas tend to be
  dry while valleys are wet.
• Evaporation and runoff have non-linear
  relationships with soil wetness.
• The ability of a land-surface scheme to model
  evaporation correctly depends crucially on its
  ability to model runoff correctly. The two
  fluxes are intricately related. (Koster and
  Milly, 1997)

20
Catchment-Based Approach to Representing
Soil Moisture Heterogeneity

• An elegant approach to the problem is to organize
  the LSS into a mosaic of catchments, and to
  dynamically identify the fraction of the
  catchment area that falls into distinct
  hydrologic regimes, where there is a topographic
  dependence on distribution (Koster et al 2000,
  Ducharne et al 2000)
• Saturated
• freely evaporating, high runoff
• Unsaturated but no transpiration stress
• minimal stress on evaporation, higher
  infiltration, lower runoff
• Unsaturated and below wilting point
• little to no evaporation, almost no runoff

21
TOPMODEL-Based Runoff Scheme
TOPMODEL (Bevin and Kirkby, 1979) assumes that
subgrid moisture distribution is related to
landscape topography

• Calculate water table depth (zw), function of
  soil wetness, (groundwater)
• Determine saturation fraction (Fsat) which is
  function of zw and topographic index
• Rsurf FsatQliq(1-Fsat)ws,14Qliq
• Rsurf is surface runoff
• Qliq is liquid water reaching surface
• ws,1 is surface soil wetness

Yang and Niu, 2003 Niu and Yang, 2003 Gedney
and Cox, 2003
22
River Transport Models (RTMs)

• Motivation for development of RTMs
• Provide freshwater input to ocean model
• Another means of validating land model (E P
  R ?SMstorage)
• Forecast large-scale water resource changes in
  future climates
• dS/dt Fin- Fout R
• S river water storage
• Fin flux of water into cell
• Fout water flow velocity S / distance
  between neighboring cells
• R total runoff from land model

23
River Transport Models (2)
.
20-yr average river flow (m3 s-1)
Raw GCM runoff Routed GCM runoff
Observed riverflow
CLM-RTM
24
Vegetation and Ecosystem Dynamics

• Dynamic Global Vegetation Model (DGVM)
• Permits plant community composition to change
  over time (e.g., grassland conversion to forest)

25
HadCM3-DGVM Climate Change Experiment
(Cox et al. 2000)
Amazon dieback in mid-21st century acts as
positive climate change feedback
Change in C (Gt C)
26
Hydrology biases and vegetation
The coupled CAM3/CLM3-DGVM cannot simulate a
forest in eastern U.S.
Uncoupled CLM3-DGVM simulations demonstrate the
sensitivity of vegetation to precipitation
Bonan Levis (2005) J. Climate, CCSM special
issue
27
Current and Future Development Priorities
(CLM3.x-4)

• Separate land model grid from atmospheric grid
• Groundwater reservoir and aquifers, bedrock
  dataset, hydraulic lift
• Prognostic canopy airspace
• More comprehensive validation, better use of
  existing validation data (tower sites) and
  emerging satellite data (e.g. soil moisture)
• Improved / updated surface datasets
• Interactive Carbon / Nitrogen cycling
• Interactive forest fire modeling
• Coupled ice sheet model
• Interactive crops
• Introduce isotope capabilities
• Keep everything working, especially when
  coupled!!!

28
Significant Outstanding Issues in
Land-Surface Scheme Development

• Current suite of LSSs simply do not agree with
  each other (PiLPS, GSWP, GLACE).
• Range of annual average results for HAPEX-MOBILHY
  (PiLPS)
• Evaporation 550 mm (65 of P) -
  800 mm (95 of P).
• Runoff 50 - 300 mm, Latent heat 45 - 60 W
  m-2, Sensible heat 20 - 35 W m-2
• Wide range of diagnosed land-atmosphere coupling
  strengths (GLACE)
• Implicit versus explicit coupling with planetary
  boundary layer (PBL) scheme (Polcher).
• Biases compared to (limited) observations in both
  offline and coupled simulations.

29
Lessons from Recent Model Development Effort
Correcting a Major Bias in CLM3 Amazon Hydrology

• Amazon hydrology problems in CLM3
• Dry soils in general, little interseasonal SM
  storage
• Low evapotranspiration and high temperatures
  during dry season
• Very high canopy interception
• Very low transpiration (photosynthesis)
• Amazon runoff/river flow lower than
  observational estimates (GRDC, gauge)
• Amazon rainforest not supported in DGVM mode

ET ES EC 69 4 26
ET ES EC 11 18 71
30
Global Hydrology Issues
These problems are actually global, they are just
particularly acute in the Amazon .
Indicative of low interseasonal SM storage
31
Proposed Modifications to Vegetation and
Hydrology Schemes and Parameters
Targets for improvement not well-defined, no
global-scale observations of partitioning of
evapotranspiration, soil moisture.

• Modifications to vegetation-related parameters,
  schemes, and datasets
• Reduce canopy interception, capacity
• Insert 2-leaf model (transpire from sunlit and
  shaded portions of canopy)
• Select alternative formulation of ?t (soil
  moisture stress function)
• Reduce soil conductance beneath canopy
• Introduce new surface dataset (PFTs, LAI) based
  on MODIS data

• Modifications to hydrology scheme
• Turn off base flow from layers 6-9
• Introduce infiltration enhancement tied to root
  density in upper soil layers
• Turn off vertical scaling of Kast

All modifications are relatively minor, no major
reworking of code
32
1-D Simulation Forced with ABRACOS
Tower Site Data (April-July 1992)
33
River Flow and Discharge
All modifications
Annual discharge into Global Oceans
Control
34
Impact of Modifications on Amazon Hydrology
Column Soil Moisture
Control
All modifications
All modifications
Control
Control
All modifications
Choudhury 1998
ET ES EC 69 4 26
ET ES EC 11 18 71
ET ES EC 55 14 31
35
Global Hydrology Issues
36
Amazonia Hydrology
CLM3 Offline and CAM3 AMIP
Simulations
37
Summary

• Land-surface schemes are an increasingly
  important component of Global Climate Models.
• LSSs have come a long way in the last 20 years in
  terms of their ability to represent large-scale
  processes that are important to climate and
  climate change.
• There is a lot of work left to do!!!

38
Extra Credit

• The 12 Stone Problem
• You have 12 stones that look exactly alike, but
  one of them weighs either less or more than the
  other 11.
• Develop a model that describes how to robustly
  identify the unique stone using three
  measurements with a weight balance.