CABLE: the Australian community land surface model Bernard Pak, Yingping Wang, Eva Kowalczyk CSIRO Marine and Atmospheric Research OzFlux08, Adelaide, 4-6 Feb 2008 CABLE = Community Atmosphere Biosphere Land Exchange model

Australian Community Climate Earth System Simulator (ACCESS) modelling program Diagram to right shows ‘scope’ Fundamentally conceived as a modelling ‘system’ that meets a variety of needs. Priority needs are: Numerical weather prediction Climate change simulation capability Collaboration between key institutions (Bureau, CSIRO, Australian Universities,….) CABLE will be the land-surface module of the global model ACCESS. Who is developing Earth Systems Modelling now? Hadley Centre (UK) IPSL (France) MPI (Germany) FRCGC (Japan) NCAR (USA) GFDL (USA)

Kowalczyk et al., CMAR Research Paper 013, 2006 The general structure of CABLE Interface to the GCM Canopy radiation; sunlit & shaded visible & near infra-red, albedo SEB & fluxes; for soil-vegetation system: Ef , Hf , Eg , Hg; evapotranspiration Carbon fluxes; GPP, NPP,NEP stomata transp. & photosynthesis soil temp. soil moisture soil respiration snow Meteorological forcing provided by GCM are passed through the interface to CABLE (radiation, wind, pressure, CO2 concentration, air and soil temperature, precipitation, soil moisture). SEB = surface energy balance CABLE modifies the soil temperature and moisture, calculates the carbon fluxes and passes back through interface to the GCM. In offline mode, CABLE can work on single point or multiple points so long as the met forcing at those points are provided. carbon pools; allocation & flow CASA-CNP vegetation dynamics/disturbance Kowalczyk et al., CMAR Research Paper 013, 2006

Vegetation parameters required for CABLE VEGETATION TYPE 1 broad-leaf evergreeen trees 2 broad-leaf deciduous trees 3 broad-leaf and needle-leaf trees 4 needle-leaf evergreen trees 5 needle-leaf deciduous trees 6 broad-leaf trees with ground cover /short-vegetation/C4 grass (savanna) 7 perennial grasslands 8 broad-leaf shrubs with grassland 9 broad-leaf shrubs with bare soil 10 tundra 11 bare soil and desert agricultural/c3 grassland 13 ice Geographically explicit data LAI – leaf area index fractional cover C3/C4 - fraction the model calculates: z0 – roughness length α – canopy albedo A grouping of species that show close similarities in their response to environmental control have common properties such as: - vegetation height - root distribution - max carboxylation rate - leaf dimension and angle, sheltering factor, - leaf interception capacity

Soil parameters required for CABLE Soil types: Coarse sand/Loamy sand Medium clay loam/silty clay loam/silt loam Fine clay Coarse-medium sandy loam/loam Coarse-fine sandy clay Medium-fine silty clay Coarse-medium-fine sandy clay loam Organic peat Permanent ice Soil Properties: - water balance: wilting point field capacity saturation point hydraulic conductivity at saturation matric potential at saturation - heat storage: albedo, specific heat, thermal conductivity density - soil depth Post, W., and L. Zobler, 2000 Global Soil Types

Nonlinear parameter estimation 19 FLUXNET sites, including all major veg types in the temperate and subtropical climate; Uniform parameter range ; Optimisation was applied to each year’s measurements separately; Each type of obs was weighted by the SD of measurements over a year. Published in Wang et al. (Global Change Biology, 2001 and 2007)

A temperate evergreen forest, Tumbarumbra, Australia

Black line – predictions with constant Vcmax (used in most LSM): Red line – predictions with seasonal varying Vcmax as function of soil temperature at 20 cm depth. So, this is adopted as the standard calculations within CABLE for deciduous forests.

Relative Vcmax: deciduous forests Leaf growth Leaf fall Similar trends for various deciduous forests. The relative vcmax increases during the beginning of growth season due to increase in enzymes; and then during senescence, it decreases due to degradation and translocation of leaf enzymes. HE = Hesse, HV = Harvard Forest, VI = Vielsalm, WB = Walker Branch, WL = Park Falls/WLEF

Changes to-date Major clean-up and rewriting of codes in FORTRAN90 standard, making it more modular and more flexible. A secured website has been set up for code distribution, installation help and exchange of ideas. Standard scripts introduced to help analyzing results. Monthly meeting

Future changes CASA-CNP: Nitrogen and phosphorus have been found to impact strongly on climate predictions. Such nutrient cycles will be implemented within the current year. It also necessitates a better phenology module. Dynamic vegetation: a UNSW post-doc (Dr. Jiafu Mao) have just started in January. He will implement LPJ into CABLE. Hydrology: MU and BoM have done some work, we will have at least one post-doc later this year. Switching to IGBP vegetation types as our research show better correlation with phenological dates from satellite data. Robert Pipunic (MU) and Adam Smith (BMRC) have applied CABLE on Kyeamba Creek and Murray Darling Basin respectively.

CABLE as the Australian community land surface model Source codes and documentation are available to all registered users online at https://teams.csiro.au/sites/cable/default.aspx (email bernard.pak@csiro.au to register) We are looking for collaboration to validate/improve CABLE

The End

Modelling Vcmax and Jmax

Parameter ranges

Observed and estimated vcmax per unit leaf area during growing season (mmol m-2 s-1) Site Estimated observed Harvard forest 61 56 Tumbarumbra 64 71 Hyytiala 74 Hesse 65 70