Scaling estimates of

1
Scaling estimates of net primary productivity
across spatial and temporal scales Frédéric
Raulier Pierre Bernier Dan McKenney
2
General context
10 of the world's forests are on Canadian
territory.
When planning forest use and conservation
strategies, Canada must accommodate the interests
of all the users while maintaining in the long
term
The Atlas of Canada

• the diversity and abundance of forest
  ecosystems
• the growth of the forest industry
• public access to forest territories.

Source http//www.cfl.scf.rncan.gc.ca/ecosys/issu
es/intro_issue_e.htm
3
Forest issues for Canada

• Achieving a balance between the need for fibre
  and the protection gives rise to many issues,
  including
• assessing the impacts of climate change
• maintaining the competitiveness of the forest
  industry
• assessing the forest sustainability with the
  current forest use
• monitoring forest health.

4
Within this context
5
ECOLEAP objectives
nobles (of course...) but a deep source of duality

• Improve our understanding of the environmental
  control on forest productivity
• Develop tools for predicting productivity over
  large areas, with a spatial scale pertinent to
  forest management applications

6
Duality of spatio-temporal scales
Processes affecting productivity
hour / leaves - canopy
decade / country
Processes affecting decision
7
Heart of this presentation
Set-up of an integration process to obtain a
decisional tool

• Simple (easy to communicate, apply and
  understand)
• Requiring data easily available at its
  operational scale
• Linked only to the influencial processes
• Honest (able to recognize its error)

8
To start, we worked with Net Primary Productivity
NPP is a measurement of plant growth obtained by
calculating the quantity of carbon absorbed and
stored by vegetation.
It is a major component of the forest carbon
cycle It encapsulates important determinants of
climate-growth interactions its a partial but
obligatory step
9
Boreal Shield Ecozone

• black - white spruces (Picea sp.),
• jack pine (Pinus sp.),
• balsam fir (Abies sp.)
• white birch (Betula sp.),
• trembling aspen (Populus sp.)

• Long, cold winters and short, warm summers
• Mean annual temperatures -4C to 5.5C
• Average annual precipitation 400 mm to 1 000 mm
  (up to 1600 mm)
• Acidic and poor soil (Canadian shield)

10
Scaling-up three-tiered procedure
Spatio-temporal domain
ForLEAP StandLEAP FineLEAP
Strategic model
Tactical models
Object of study
11
Why 2 spatial leaps ?
Stand dynamics (structure, species mixture)
Mixture of vegetation types
Basical physiological processes
Gross forest types
Stand types
Species
Objects of simulation
12
Why 2 temporal leaps ?
ecophysiological processes

• daily and seasonal variations of climate
• phenological processes

decisional processes
13
Leaf-to-plot modelling, multi-layer, half-hourly
time step, Farquhar equations.
Parameterization of StandLEAP (epsilon-type,
monthly time step) for the various tree species
Application of StandLEAP to pilot regions at the
30 x30 m resolution
30x30m pixels
4 km2 pixel
Parameterization of ForLEAP (epsilon-type, yearly
time step, 4 km2 pixel) using StandLEAP
simulations of all pilot regions.
20 parameters
14
How did we proceed ?
NPP r GPP GPP e APAR P(f)

• Screened for new explanatory variables defined at
  the upper scale
• Decided of a satisfaction criteria (DR2 5)

15
Inputs for FineLEAP

• Shoot and canopy light absorption properties
• Photosynthesis / transpiration measurements
• Hourly climatic values (temp, wind, radiation)

NASA / NSERC
BOREAS Northern Study Area
Canadian Forest Service
black spruce, jack pine, trembling aspen
Forêt Montmorency
balsam fir
16
Use of pilot regions to jump from the stand to
the forest
Five pilot regions
Abitibi north and south
Corner Brook
BOREAS northern study site
Forêt Montmorency
17
Inputs for StandLEAP
NASA (LANDSAT-TM)
provincial Ministries of Natural Resources /
Natural Ressources Canada
Agriculture Canada
Environment Canada / Canadian Forest Service
18
monthly Tmax
Inputs for ForLEAP
10C

• Long term mean climate grid for Canada were used
  to compute climatic indices
• growing season degree days
• growing season mean temperature
• growing season radiation

25C
Canadian Forest Service
annual precipitation
monthly Tmin
19

• 1995 landcover of Canada
• NDVI map (AVHRR - CCRS)

Inputs for ForLEAP
Canadian Centre for Remote Sensing / Canadian
Forest Service
20
Did we truly simplify ?
ForLEAP StandLEAP FineLEAP
3 cover type classes
20
15 forest types / 5 species
90
5 species
180
21
NPP map
Must account for the domain of the empirical fit
g m-2 y-1
Pilot regions
Boreal shield ecozone
22
Tracking the error down ! (FineLEAP)
23
(No Transcript)
24
Tracking the error down ! (StandLEAP)
Half confidence interval (?0.05) around the
predicted mean (g/m2/year)
Aboveground net primary productivity (g/m2/year)
25
Conclusions
spatializing productivity estimates

• Scaling up through an empirical procedure
  produces a very robust and simple tool
• Encapsulates processes and environmental
  variables from a variety of spatial and temporal
  scales
• Gives access to error analysis
• In our exercise, the number of necessary
  parameters goes from about 180 at the shoot/leaf
  level to 20 at the landscape level
• Model-to-model fit is good but degrades rapidly
  when new data sources are introduced at the
  different spatial scales.

26
Acknowledgments

• Joseph Cilahr (CCRS) (countrywide AVHRR NDVI
  map)
• Jacques Regnières, Rémi St-Amant (climate
  extrapolation)
• Jean Beaubien (land cover map)
• Ron Hall, Joan Luther, André Beaudoin, Richard
  Fournier, Luc Guindon, Eric Arsenault (setup of
  the pilot regions)
• Chhun-Huor Ung, Robert Boutin, Gilles
  Robitailles (setup of the intensive study site
  for balsam fir)
• BOREAS teams TE-09 (Margolis et al.), TE-06
  (Gower et al.) and RSS-07 (Chen et al.)
• Sébastien Dagnault, Adrien Forgues, Gérard
  Laroche and many summer students (field work)

27
Thank you !