Water, carbon and nutrients on the Australian continent: effects of climate gradients and land use c

1
Water, carbon and nutrients on the Australian
continent effects of climate gradients and land
use changes

• Michael Raupach, Damian Barrett, Peter Briggs and
  Mac KirbyCSIRO Land and Water, Canberra,
  Australia
• michael.raupach_at_csiro.au
• Outline
• Models, data, constraints
• Results
• Uncertainty and synthesis
• Acknowledgments Helen Cleugh, John Finnigan,
  Roger Francey, Dean Graetz, Ray Leuning, Peter
  Rayner, Hilary Talbot
• IGBP Global Change Science Conference, Amsterdam,
  July 2001

2
Two points on the landscape
Savannah woodland Rainfall 800 mm
Old-growth forest Rainfall 600 mm
3
Linked terrestrial cyclesof water, C, N and P
Water flow
C flow
N flow
P flow
4
Modelling water, carbon and nutrient cycles
Framework the dynamical system

• Variables X Xr set of stores (r)
  including all water, C, N, P, stores F
  Frs set of fluxes (affecting store r by
  process s) M set of forcing climate and
  surface forcing variables P set of
  process parameters
• Stores obey mass balances (conservation
  equations) of form (for store r)
• Statistical steady state or quasi-equilibrium
  solutions
• Fluxes are described by scale-dependent
  phenomenological equations of form

Used here!
Problem find these for large scales!
5
Scaling a general viewStatistical averaging of
phenomenological equations

• Requirement for scale consistency
• X, F, M and P are all defined with the same space
  and time averaging
• Related to smaller-scale process descriptions by
  statistical averaging
• Statistical averaging process space or time
  averages of fluxes are

Coarse-scale average flux
Fine-scale model
PDF of (X,M,P) V
Fine-scale model with coarse data
Bias (co)variance second derivative of
Frs(V)
6
Evaporation and TranspirationSimplifying
infiltration models to 2-layer soil, daily time
step (Mac Kirby)
Rain Ksat2
Rain Ksat1
Duplex soil Ksat1 Ksat2
Rain 7
Evaporation and TranspirationA simple
statistical-steady-state model

• Evaporation is determined by (rainfall, energy)
  in (dry, wet) environments
• Energy-limited Evaporation
  Priestley-Taylor Evaporation constant
  Available Energy
• A single-parameter hyperbolic function
  interpolates between dry and wet limits
• Total Evaporation Plant Transpiration
  Soil Evaporation
• Time average of Soil Evaporation / Total
  Evaporation exp(-cLAI)
• Annual mean, catchment-scale water balance

8
Evaporation and TranspirationTests of
statistical-steady-state model

• Annual mean, catchment-scale water balance

9
Evaporation and energy forest sytemsRay Leuning
and Helen Cleugh (CLW), Tumbarumba flux site

• Daytime evaporation 1.1 equilibrium
  evaporation

10
Evaporation and energy cropping sytemsChris J
Smith and Frank Dunin, CSU Site, Wagga
60
Triticale, 1999
Priestley Taylor
Lysimeters
50
40
30
20
10
0
Jul-99
Jan-00
Mar-99
May-99
Nov-99
Sep-99
Evapotranspiration (mm/week)
60
Lupin, 2000
50
40
30
20
10
0
Jul-00
Jan-01
Mar-00
Sep-00
Nov-00
May-00
11
Quasi-steady surface energy balance in an
entraining convective boundary layer

• Why Priestley-Taylor evaporation is a good
  measure of potential evaporation over a moist
  region

parameter relative deficit of entrained air
12
Net Primary Productivity (NPP)

• NPP Photosynthetic Assimilation -
  Autotrophic Respiration
• A simple, linearised model for light and water
  limited NPP

13
Testing predictions of NPPVast dataset (Barrett
2001)

• linear axes logarithmic axes

14
Testing predictions of NPPVast dataset (Barrett
2001)

• NPP depends on saturation deficit, through water
  use efficiency

MeasurementsModel
15
Testing predictions of C storesVast dataset
(Barrett 2001)
16
Data requirements

• Climate
• Rainfall solar irradiance temperature
  humidity
• Land cover and land use
• Vegetation properties (leaf area index height)
• Land use (forest / rangeland / crop / pasture /
  horticulture)
• Land management
• Fertiliser application rate (N, P)
• N fixation by legumes
• Irrigation
• Soils
• Soil type (via pedotransfer functions)
• Soil depth soil texture hydraulic properties
  bulk density

17
C, N and P balances with present climate and
agricultural nutrient inputs Net Primary
Production

• NPP broadly follows rainfall, with additional
  modulation by saturation deficit (through water
  use efficiency). Hence there is less NPP per unit
  rainfall in north than in south.

18
Effect of agricultureExample ratio of (NPP with
agriculture) / (NPP without agriculture)

• NPP has increased locally (at scale of 5 km
  cells) by up to a factor of 2 in response to the
  nutrient inputs associated with European-style
  agriculture
• Largest regional-scale increases occur in the WA,
  SA, Victorian and NSW wheatbelts

19
Mineral N balance

• Without agriculture
• IN fixation, small deposition
• OUT leaching, volatilisation, disturbance
• With agriculture
• More fixation (x 2)
• More disturbance

20
Summary

• A formal dynamical-system framework
• rigorous treatment of scaling, uncertainty,
  synthesis
• Information flow evaporation - NPP -
  fluxes and stores of C,N,P
• Effects of agriculture on NPP, nitrogen and
  phosphorus
• Agricultural nutrient inputs (fertiliser,
  legumes) have led to regional-scale increases
  (relative to pre-agricultural conditions) of up
  to factor of 2 for NPP, and up to a factor of 5
  for mineral N, labile P
• Largest changes in N balance are fixation (sown
  legumes) and disturbance (herbivory)
• Continental aggregates
• Mean continental NPP without agriculture is 0.96
  GtC/year
• Continental changes induced by agriculture NPP
  4.8 mineral N 13 labile
  P 7.6 N budget (in, out) factor 2

21
SynthesisA multiple-constraint approach (1)

• Problem What is the space-time distribution of
  the sources and sinks of CO2 (water, CH4, N2O,
  dust ) across a large region?
• Available information from observations
• C(i) atmospheric concentrations provide
  budget constraint
• E(j) eddy fluxes provide accurate point
  checks
• R(k) remotely sensed data provide indirect
  continental coverage
• S(m) carbon stocks provide biological linkage
• Model
• Includes a (small) set of N parameters p which
  are poorly known
• Predicts flux distribution F with given
  parameters p
• Can also predict observable quantities (C, E, R,
  S)
• How can we use observations (of C, E, R, S) to
  constrain p?

22
SynthesisA multiple-constraint approach (2)

• Approach
• Use the model to predict the observed quantities
  C, E, R and S, and also the regional flux
  distribution F(p), using a consistent small set
  of parameters p
• Determine p by minimising a multiple objective
  function JmultJmult sum of several single
  objective functions (each a sum-of-squared
  errors)

• Use p to determine regional flux distribution
  F(p)
• Keys to approach
• Multiple (not necessarily direct) observations
• A model which predicts F and all observables with
  common parameters p
• Consistency check use single objective functions
  (for C, E, R, S) separately