Reference canopy conductance through space and time: Unifying properties and their conceptual basis

1
Reference canopy conductance through space and
timeUnifying properties and their conceptual
basis

• D. Scott Mackay1 Brent E. Ewers2 Eric L.
  Kruger3
• Jonathan Adelman2 Mike Loranty1 Sudeep
  Samanta3
• 1SUNY at Buffalo 2University of Wyoming
  3UW-Madison

NSF Hydrologic Sciences EAR-0405306 EAR-0405381
EAR-0405318
2
Problem

• Prediction of water resources from local to
  global scales requires an understanding of
  important hydrologic fluxes, including
  transpiration
• Current understanding of these fluxes relies on
  center-of-stand observations and
  paint-by-numbers scaling logic
• Spatial gradients are ignored, but this is an
  unnecessary simplification
• New scaling logic is needed that includes linear
  or nonlinear effects of spatial gradients on
  water fluxes

3
Why is canopy transpiration important to
hydrology?
Average annual precipitation 800 mm Growing
season precipitation 300-500 mm Growing season
evapotranspiration 350-450 mm Canopy
transpiration (forest) 150-200 mm Canopy
transpiration (aspen) 300 mm
Ewers et al., 2002 (WRR) Mackay et al., 2002 (GCB)
4
Assumptions

• Transpiration is too costly to measure
  everywhere, and so appropriate sampling
  strategies are needed
• The need for parameterization (e.g., sub-grid
  variability) will never go away
• Both forcing on and responses to transpiration
  are spatially related (or correlated), but this
  correlation is stronger in some places
• Human activities may increase or decrease this
  correlation

5
What if we increase edge effects?
Center-of-Stand Basis
Spatial Gradient Basis
Transpiration mm (30-min) 1
6
Why is Transpiration a Nonlinear Response?
Relative Response
Relative water demand
7
Conceptual Basis of Spatial Reference Conductance
Environmental Gradient
Canopy stomatal control of leaf water potential
Hydraulic Universal line
GS GSref mlnD m 0.6GSref
(Oren et al., 1999)
Mapping from spatial domain into a linear
parameter domain
8
Mackay et al., 2003 (Advances in Water Resources)
9
Hypothesis 1

• GSref varies in response to spatial gradients
  within forest stands, but the relationship
  between GSref and m remains linear
• Note that 1/D ? 1- 0.6ln(D) for 1 D 3 kPa
  error is maximum of 16 at 2 kPa
• Thus many empirical stomatal conductance models
  are applicable, but discrepancies will occur at
  moderate mid-day D when it is hydrologically most
  relevant

10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Some model realizations follow hydraulic theory
Agricultural and Forest Meteorology (in review)
14
These models preserve plant hydraulics and
represent the regional variability for Sugar maple
Agricultural and Forest Meteorology (in review)
15
Aspen flux study, northern Wisconsin
X sample point X - Aspen
Funded by NSF Hydrological Sciences
16
Aspen Restricted Simulations
Funded by NSF Hydrological Sciences
17
Lodgepole pine study, Wyoming
X sample point X Lodgepole pine
18
Lodgepole Pine Restricted Simulations
A1, riparian zone
Row 4, lower slope
Row 5, mid-slope
Row 6, mid-slope
Basal area crowding
Row 7, mid-slope
Row 8, upper slope
19
Summary of Ecohydrologic Constraints
xeric
mesic
high
Reference Canopy Conductance
low
low
high
high
low
Water availability Index
Hydraulic Constraint Index
20
Hypothesis 2

• Variation in leaf gS within and among species and
  environments is positively related with leaf
  nitrogen content and leaf-specific hydraulic
  conductance
• The relative response of gSmax to light intensity
  (Q) is governed in large part by ?leaf, and this
  dependence underlies stomatal sensitivity to D
• Corollary i gS will increase with increasing Q
  until it reaches a limit imposed ?leaf, which for
  a given leaf is mediated primarily by D
• Corollary ii The limit imposed on relative
  stomatal conductance (g/gSmax) by ?leaf (relative
  to the threshold linked to runaway cavitation,
  ?crit) is consistent within and among species

21
Hypothesis 3

• The model complexity needed to accurately predict
  transpiration is greater in areas of steep
  spatial gradients in species and environmental
  factors
• Model complexity (e.g. number of functions,
  non-linearity) should be increased when
  absolutely necessary, and it should subject to a
  penalty
• We should gain new knowledge whenever we are
  forced to increase a models complexity

22
(No Transcript)
23
(No Transcript)