4D modeling of canopy architecture for improved characterization of state and functionning

1
4D modeling of canopy architecture for improved
characterization of state and functionning

• F. Baret
• INRA-CSE Avignon

2
Introduction

• The description of vegetation architecture is one
  of the main limiting factor in the estimation of
  canopy characteristics such as LAI
• Importance of the temporal dimension that drives
  the generation of canopy architecture and that
  offers regularities to be exploited

Turbid medium
Geometric
Explicit
3
Requirements

• Good dynamic description of canopy architecture
• Low amount of parameters/variables (for better
  retrieval)
• Fast computation of the radiative transfer

Objectives of the study

• Illustrate how canopy structure evolution could
  be generated
• Present the corresponding variables and
  parameters used
• Describe how to compute the radiative transfer
• Conclude on the work to achieve

4
The context of high spatial and temporal
resolution observations

• High spatial resolution
• Generally pure pixels
• object observed could be identified in terms of
  species
• High temporal resolution
• Continuous monitoring to be exploited in the
  understanding of how the architecture builds up
  (or destroys down!)

Case illustrated here maize canopies with
relatively simple and well known architecture
5
Modeling maize canopies architecture

• Work derived from previous studies M. Espana,
  B. Andrieu, M. Chelle, B. Koetz, N. Rochdi
• Describing the time course of individual leaves
  and stems
• Based on a series of experiments
• Semi-mechanistic models
• Reduced number of variables
• Reasonable level of details in canopy
  architecture description

6
Level of canopy architecture details required for
reflectance simulation
T0
T1
T2
7
Leaf area time course

• Time of leaf of order n
• Apparition nDTc
• Disparition nDTcDTs
• Variables required
• N_max
• S_max
• To
• DTc
• DTs

8
Other architecture characteristics

• Canopy
• Plant density
• Distance between rows
• Row azimuth

• Plant
• Leaf insertion height
• Leaf shape/curvature
• Leaf azimuth
• Leaf zenith

Leaf insertion height
Leaf order
9
Properties of the 4D maize model

• Limited number of variables/parameters
• N_max
• S_max
• To
• DTc
• DTs
• H_max
• Density
• Leaf inclination
• Dynamics well described
• Improvements
• Leaf curvature (easy)
• Better senescence including keeping senescent
  leaves
• Variability between plants (size, position, )
• Flowers/ears
• Vertical gradients in chlorophyll

10
Regularities in Chlorophyll gradients
1999
2001
Distribution verticale du contenu en chlorophylle
mesurée à partir de linstrument SPAD502
11
From canopy architecture to reflectance
Parcinopy
Multispectral version now available (M. Chelle,
V. Rancier)
12
Decomposing radiative transfer
Terms required
n(level,way,direction,inter_sol,inter_veget)numbe
r of photons (radiance) level bbottom
ttop way - downward upward direction
qssun direction qvview directionhhemispheric
interaction order (inter_sol,
inter_veget) 0 no interaction 1 1
interaction only ?1 one or more interactions
tss a tsstoo.rso f rso de rdd c/(a.Rsb.Rs)
tsd b tdo g /(a.Rsb.Rs)
13
Vegetation contribution (rso)
rso f(rleaf,tleaf,P(LAI,ALA,S,D,?,?s?v))
Parameters 'P' are spectral invariants
14
Approach
Building a parametric model
LAI,ALA,S,D,? Distribution of input variables
Constructionof the 3D architecture
?s?vSun/view configuration
PARCINOPY
rl, tl leaf reflectance Transmittance
Canopyreflectance
rs Soil reflectance
RT components
Parametric model P(LAI,ALA,S,D,?,?s?v, rl, tl)
15
CONCLUSION

• A more mechanistic/realistic approach is proposed
• Based on a simple description of canopy
  architecture to use fewer variables
• No need for continuous description (discrete is
  enough)
• Needs sensitivity analysis to evaluate the
  influence of the variation of N_max, H_max,
• Needs full (or at least parametric for the
  spectral aspect) parametric model to be
  implemented to compute the reflectance fields
• Needs coupling to canopy functioning models

16
Coupling between structure and function models
Reflectance
LAI
4D Architecture Model
Initialization
Stress (H2O, N)
S T
Work in progress for exploitation within an
assimilation scheme,