Transcript 13 Henderson Halofsky

Integrating Climate Change into
Landscape Planning
Modeling climate and management
interactions within the ILAP framework
April 23, 2013
Jessica Halofsky
Emilie Henderson
Modeling
•
•
“Essentially, all models are wrong, but
some are useful.” – Box & Draper 1987
“It’s Only a Model.”
– Patsy, 1975
Projects behind today’s talks
• Integrated Landscape Assessment Project (ILAP)
– Climate Change Module
– Central Oregon Study Area
• Climate, Management and Habitat
How might climate and land management
interact to shape vegetation and habitat?
Coastal Washington
Northern Spotted Owl
Greater Sage Grouse
Southwest Oregon
Southeast Oregon
Scenarios
Management
Current
Management
Climate
Hadley
(Hot/Dry)
MIROC
(Hot/Wet)
CSIRO
(Warm/Moist)
Restoration
Natural
Resources
Economy
General Topics for today’s talk:
• Starting conditions
• STMs without climate
• Climate impacts modeling
Modeling Strata
The overall picture:
Not locally precise.
Useful for describing landscape-to regional
trends.
Current Vegetation
• GNN = Gradient Nearest Neighbor
– A spatial depiction of the FIA plots.
– Structured by an ordination model.
• Gives us information on current vegetation
within each modeling stratum.
Janet Ohmann, Matthew Gregory, Heather Roberts
General Topics
• Starting Conditions
• State Transition Modeling, without
accounting for climate
• Climate impacts modeling
State and Transition Modeling
Early
Successional
Young Forest
Mature
Forest
Old Growth
Forest
Growth
Fire
Regeneration
Harvest
State and Transition Modeling:
Dry Douglas-fir
Pole-stage, single-story, post-disturbance
Grass-Forb
Giant Trees
Moderate Canopy
Multi-Layered
ILAP Potential Vegetation Types
PROBLEM:
Basic framework assumes that
this map doesn’t change.
When climate shifts, so
will potential vegetation.
State and Transition Modeling
Early
Successional
Young Forest
Mature
Forest
Late
Successional
Old Growth
Growth
Fire
Regeneration
Harvest
Estimated harvests from the LANDSAT record in
Southwest Oregon
Acres Harvested
120000
Federal
Private Industrial
100000
80000
60000
40000
20000
year
Thanks to Robert Kennedy for the LandTrendr maps of disturbance history
2007
2005
2006
2004
2002
2003
2001
1999
2000
1998
1996
1997
1994
1995
1993
1991
1992
1990
1988
1989
1987
1985
1986
1984
0
Current Fire
PROBLEM:
500000
300000
Fire regimes are set to
resemble the recent past.
200000
year
Extracted from Monitoring Trends in Burn Severity dataset: mtbs.gov
2007
2005
2006
2004
2002
2003
2001
1999
2000
1998
1996
1997
1994
1995
1993
1991
1992
1990
1988
1989
1984
0
1987
100000
They will probably change with
shifting climate.
1985
1986
Acres burned
400000
Preliminary results for Southwestern Oregon:
Current management
Area with Large and Giant Trees
No Climate Change
General Topics
• Starting Conditions
• State Transition Modeling, without accounting
for climate
• Climate impacts modeling
– ILAP extension – Central Oregon Study Area
What about climate change?
• Climate controls
ecosystem processes,
including:
– Plant establishment,
growth, and mortality
– Disturbance
• Drought
• Fire
• Insect outbreaks
Dynamic Global Vegetation Models
(DGVMs):
• Link state-of-the-art knowledge of:
– plant physiology
– biogeography
– biogeochemistry
– biophysics
• Simulate changes in vegetation structure and
composition and ecosystem function through
time
The MC1 Dynamic Global
Vegetation Model
lifeform mixture
Biogeography
(MAPSS)
Biogeochemistry
(CENTURY)
live biomass
biomass mortality
fire
occurrence
nutrient loss and release
lifeform
mixture
Fire
(MCFire)
carbon pools
soil moisture
*adapted from: Bachelet, D., J. M. Lenihan, C. Daly, R. P. Neilson, D. S. Ojima, and W. J. Parton. 2001. MC1: A Dynamic Vegetation Model for Estimating the Distribution of
Vegetation and Associated Ecosystem Fluxes of Carbon, Nutrients, and Water. USDA Forest Service General Technical Report PNW-GTR-508.
A Linked Model Approach
STMs
MC1
Juniper woodland
Xeric Ponderosa Pine
Dry Mixed Conifer
Moist Mixed Conifer
Central Oregon Study Area
Historical vegetation in the study area
Vegetation type crosswalks
MC1 Functional Vegetation Type STM Potential Vegetation Type
Subalpine forest
Cool needle-leaved forest
Temperate needle-leaved forest
Temperate needle-leaved woodland
Mountain hemlock and subalpine fir
forests
Moist mixed conifer and white fir
forests
Ponderosa pine, lodgepole pine, and
dry mixed conifer forests
Mountain big sage – western juniper
woodland and shrubland
Temperate shrubland
Wyoming big sage shrubland
Xeromorphic shrubland
Salt desert shrubland
Temperate grassland
Bluebunch wheatgrass – Sandberg
bluegrass grassland
Warm-season grassland
Warm-season grassland
Climate Scenarios
MIROC
MC1
Functional
Vegetation
Type
Projections
Halofsky et al. in review
Hadley
CSIRO
MC1 fire projections
MIROC
CSIRO
Hadley
Halofsky et al.
in prep
MIROC
Linked
model
results
Hadley
Halofsky et al. in prep
CSIRO
Central Oregon Management Scenarios
• Fire suppression only
– Fire frequencies same as the last 25 years under
fire suppression
– No other active management
• Resilience
– Light to moderate levels of thinning and some
prescribed fire in dry forest types
Effects of management on dry forests
Fire
suppression
only
Mean
Min to max
Randomly
selected
simulations
Resilience
Halofsky et al. in prep
Effects of management on:
dry forests with large trees and open canopy
Fire suppression
only
Hectares
160,000
120,000
80,000
40,000
0
2010
2030
2050
Year
2070
2090
2010
2030
2050
Year
2070
2090
Resilience
Hectares
160,000
120,000
80,000
40,000
0
Halofsky et al. in prep
Resilience
Landscape proportion
Fire
suppression
only
Landscape proportion
Trends in dry forest structure
<12.7 cm
12.7-50.8 cm
>50.8 cm
Conclusions for Central Oregon
• Linked DGVM-STM output suggests greater
vegetation resilience than DGVM alone.
• Dry ponderosa pine and mixed conifer forests
will likely maintain dominance in the central
Oregon study area.
• In some cases, management may dampen the
magnitude of forest change under changing
climate.
Halofsky et al. in prep
Technical Thoughts
• All models are wrong, ours could be useful
• These models provide big-picture projections
• The linked model process is data-, labor-, and
software-intensive
Getting to Landscape Planning
• We haven’t described the planning process itself,
which involves conversations and people.
– Stakeholders
– Decision Makers
• Our models are useful storytelling tools.
– Enable the asking of questions.
– Realistic and plausible stories.
– Enhance the role of science in conversations about
planning.
• Half of science is asking the right questions.
-- Roger Bacon (c. 1214 – 1294)1
• Emilie, you really need to refine your questions!
-- Dr. David Mladenoff, numerous times throughout my
career as a PhD student in his lab.
1Wikiquotes
120000
Acres Harvested
What activities?
e.g., partial harvest
Groups
we have
regeneration
harvest
restoration harvest
•4 activities
prescribed
fire
•The Nature Conservancy
Federal
Private Industrial
100000
80000
60000
spoken with:
40000
20000
2007
2005
2006
2004
2002
2003
2001
1999
2000
1998
1996
1997
1994
1995
1993
1991
1992
1990
1988
1989
1987
1985
1986
1984
0
•Bureau of Land Management
• 24 ownership/allocation categories
At
what
rates? Service personnel – regional and local
•US
Forest
•Local chapters of the Society of American Foresters
• ∞ variations
in rates
•Washington
Department
of Natural Resources
•Oregon Department of Forestry
•Consulting foresters who serve nonindustrial
Where should they be applied?
private landowners
•County commissioners
year
Our Hope for our Work
Save the world!
• Tell informative stories that are grounded in
science about how different landscape
management policies and plans may lead to
different futures.
• Relevance and credibility beyond the science
community.
Research
Team:
Funding:
Dominique Bachelet
David Conklin
Megan Creutzburg
Jessica Halofsky
Joshua Halofsky
Miles Hemstrom
Emilie Henderson
James Kagan
Becky Kerns
Anita Morzillo
Janine Salwasser