Seeing the wood for the trees: forest canopy structure and EO data

1
Seeing the wood for the trees forest canopy
structure and EO data
Mathias Disney
 
2

• Why?
• How to interface EO data with models
• Assimilation
• at what level? EO measures scattered radiance
• Structure
• key for Optical, microwave, thermal.

 
3

• Aims
• Explore/understand relationship between
  first-order state variables and EO measurements
• e.g. LAI, biomass, structure, radiometric
  properties (chlorophyll, dry matter, water etc.),
  understory
• Requires canopy-radiation interaction models
• Develop methods to recover biophysical parameters
• Assess quality of EO-derived products
• Interface EO to ecological process models

 
4

• Strategy
• 3D structural models simulate radiation regime
• Compare with data from multiangle
  multi/hyperspectral sensors
• Airborne (ATM, CASI), CHRIS-PROBA, LIDAR, RADAR
  etc.
• Require
• Tree structure and development
• Needle scattering behaviour
• Field measurements of canopy properties for
  validation ? model development

 
5

• Modelling
• Lack of structural model for Sitka (Harwood),
  Maritime pine (San Rossore)
• Use parameterisations for Sitka developed by
  Cochrane Ford (Herriott-Watt, 1978)
• Adapt existing model, modify light extinction
  height to match
• Measurements made at Harwood San Rossore...

 
6

• Modelling
• Match measured and modelled height (and
  ?height(t))

7

• Modelling
• Needle shape and distribution (impact PAR)
• Use Fibonacci phyllotaxy
• Published data for many species
• Big impact potentially on multiple scattering
• Manuscript to submit

8

• Modelling
• Shape and distribution have BIG impact on
  multiple scattering...at shoot scale
• BUT gets smeared out canopy scale
• Cylinders NOT appropriate.....

9

• Where now?
• Simulating Harwood, San Rossore....
• Generate LUTs of scattering indexed by structure
• Retrieve age (height, dbh), LAI, chlorophyll
  content....
• Derive PRI/LUE compare with Carolines estimates

AM 8 views max.
PM 9 views max.
10
LIDAR
CASI (11 bands)
11
Field data
San Rossore CHRIS-PROBA 16/6/03
12

• Issues?
• Registration needs to be much better
  (sub-pixel)
• Atmospheric correction
• Use LIDAR to derive height/structure, and/or
  validate 3D models
• most direct EO measure of structure ( can
  simulate)
• Potential of laser measurements (Edinburgh,
  Durham...)

13

• Achievements
• Planning and acquisition of field campaigns at
  Harwood, San Rossore
• Collection, processing of structural and
  radiometric data to drive/validate modelling (all
  on web database....)
• EO data (air- spaceborne) acquired, being
  processed
• 3D modelling techniques developed
• better understanding of modelling issues,
  weaknesses and how to Brogress
• Publications arising

 
14

• Plans for 04
• Proposals for NERC ARSF/EPFS for 04 accepted
  (graded ?4)
• Including flights over San Rossore (end
  April-ish) in conjunction with CHRIS-PROBA
• Enable multi-scale study with Edinburgh, Menenti
  et al leaf level measurement ?? air ? space
• Parameter retrievals over Harwood
• Relate ground, model, high res. data to more
  moderate res. i.e. impact of scaling on
  parameters....
• PRI from space-borne (CHRIS) data over San
  Rossore (Harwood??), relate to leaf-scale and
  airborne data (Caroline)

 
15

• Plans for 04 cont.
• Assimilation of EO data into models SPA
  (structural terms)
• Build 3D forest structure model (validate/drive
  w/LIDAR etc.)
• Test ability to simulate EO data
• Convert representation to that used by SPA (hey
  presto!)
• Test SPA ability to model CO2 fluxes with this
  representation
• Assimilation
• Grow canopy assimilate CO2 data alone
  assimilate EO data alone assimilate both
  datasets.....