constructing and evaluating forward models of the terrestrial carbon cycle peter thornton...

43
Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division NCAR

Post on 15-Jan-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Constructing and evaluating forward models of the terrestrial carbon cycle

Peter Thornton

Terrestrial Sciences Section

Climate and Global Dynamics Division

NCAR

Page 2: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Relevance of the terrestrial carbon cycle to studies of the global climate:

• Biophysical characteristics of the land surface depend on the interactions between surface weather, atmospheric composition and chemistry, and vegetation ecophysiology: leaf area, canopy height, Bowen ratio, relative abundance of different plant functional types.

• Atmospheric concentration of CO2 depends, in part, on the dynamics of the terrestrial carbon cycle, with important variation at seasonal, interannual, decadal, and longer timescales.

Page 3: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Research questions:

• What are the current dynamics of the terrestrial carbon cycle?

• What are the likely future trends in source/sink strengths under different emission and landuse change scenarios?

• How do current dynamics and likely future trends depend on coupling with climate and atmospheric chemistry?

• How much model detail is required to capture the dominant spatial and temporal modes of variation in the current and likely future terrestrial carbon cycles?

• How can coupled climate and carbon cycle models be evaluated?

Page 4: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

GEP

Re

NEE

MR

HR GR

GEP = gross ecosystemproduction

Re = total ecosystemrespiration

NEE = net ecosystemexchange

MR = maintenancerespiration

GR = growthrespiration

HR = heterotrophicrespiration

Page 5: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Coupled

Page 6: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Land-Atmosphere Nitrogen Cycle

Page 7: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division
Page 8: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Effects on NEE of natural and anthropogenic disturbances:

Simulation based on recorded disturbance history at Harvard Forest

NE

E (

gC m

-2 y

r-1)

Page 9: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Effects on NEE of anthropogenic changes in atmospheric composition and chemistry:

Page 10: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Terrestrial carbon cycle model evaluation over gradients in climate and disturbance

history

Evaluation based on observations of carbon and water budget components from seven evergreen needleleaf forest Fluxnet

sites

Fluxnet collaborators:

B.E. Law, H.L. Gholz, K.L. Clark, E. Falge, D.S. Ellsworth, A.H. Goldstein, R.K. Monson, D. Hollinger, M. Falk, J. Chen, J.P.

Sparks

Page 11: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Based on DeFries et al., 2000

Page 12: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Location of Fluxnet sites included in this study

Base map: average annual precipitation for

1980-1997

Page 13: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Location of seven ENF Fluxnet sites in climate space

Grey region: Complete climate space for the U.S. (lower 48)

Black region: Climate space occupied by evergreen needleleaf forest in the U.S (lower 48)

Page 14: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

SiteAge

(yrs)Disturbance History

Leaf long.

(yrs)

Historical Ndep

(gN m-2 yr-1)

Current Ndep

(gN m-2 yr-1)

Young Stands

BL 10 1990: clearcut, replanted 3 0.22 0.22

DU 171983: clearcut, slashburn, replanted

2 0.18 0.50

FL 101990: clearcut, slashburn, replanted

2 0.19 0.30

Mid-aged Stands

HL 90 1910: selective harvest 5 0.16 0.29

NR 95 1905: harvest 5 0.12 0.30

Old Stands (multiple age classes)

ME45-250 1750: stand-replacing fire,

1850: fire, 1950: fire4 0.18 0.18

WR 0-500 1550: stand-replacing fire 8 0.32 0.32

Summary of site characteristics

Page 15: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Blodgett Forest, CA

Howland Forest, ME

Duke Forest, NC

Niwot Ridge, CO

Metolius NRA, OR

Gainesville, FL

Wind River Crane, WA

Page 16: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Modeling Protocol

1. Using the Daymet surface weather database as the model driver at all sites.

2. Spinup simulation to bring C and N pools into steady state: repeated 18-year record of surface weather. Using assumed levels of [CO2]atm and Ndep circa 1795 AD.

3. Ensemble of site history simulations, with increasing [CO2]atm (IS92a), increasing Ndep from anthropogenic sources (following Holland et al., 1997), and disturbance history as recorded at each site. All simulations ending at 2000 AD.

4. Evaluation of ensemble statistics at current stand age, comparison against historical observations where possible.

Page 17: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Daymet database of daily surface weather drivers: climate summary

Page 18: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Disturbance history ensembling:

Each member consists of a disturbance history simulation initiated in a different year of the cyclic 18-year surface weather record.

Page 19: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Decadal and longer timescale variations dominated by disturbance history

Page 20: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Decadal and longer timescale variations dominated by disturbance history

Page 21: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Relationship between NEE and GEP depends on time since disturbance, and on changing [CO2]atm and Ndep

Page 22: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Relationship between NEE and GEP depends on time since disturbance (continued)

Page 23: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: NEE response to disturbance, changing [CO2]atm, and Ndep shows strong interaction effects

Page 24: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: NEE response to [CO2]atm for old sites near steady state depends on the rate of increase in CO2

Page 25: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model captures observed variability in leaf area index across the climate and disturbance history gradients

Page 26: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model captures observed variability in evapotranspiration across the climate and disturbance history gradients

Page 27: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model captures observed seasonality in evapotranspiration across the climate and disturbance history

gradients

Page 28: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model tends to underestimate annual NEE, with strong biases at the warmest sites (FL, DU) and the old sites (ME, WR).

Page 29: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model in generally poor agreement with observed seasonal cycle of NEE…

Page 30: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: … but the biases are not consistently attributed to either Re or GEP.

Page 31: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: … but the biases are not consistently attributed to either Re or GEP.

Page 32: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Component Observed Model

AG NPP 173 178

BG NPP 299 235

Foliage NPP 87 89

Fine root NPP 276 227

Wood NPP 109 98

Soil C flux 780 851

NEP28 (component budget)

167 ± 69 (Monte Carlo)

300 ± 170 (Eddy covariance)

60

Model evaluation at Metolius old-growth site: Comparison to biometric measurements of carbon budget components

(all units gC m-2 yr-1)

Page 33: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division
Page 34: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model in generally poor agreement with observed seasonal cycle of NEE…

Page 35: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model agrees well with chronosequence estimates of NEE at the one site where they are available (FL)

Page 36: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: Model in generally poor agreement with observed seasonal cycle of NEE…

Page 37: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Problem: How to get rid of the double peak in seasonal cycle of NEE? Focus on the Duke Forest site…

GEP

Re

NEE

MR

HR GR

Page 38: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

• Reducing the temperature sensitivity for MR and HR gives some improvement, but double peak persists.

• Excessive reduction of temperature sensitivity for HR results in low GPP due to limited N mineralization (reduced decomposition of soil organic matter).

• Have to turn to the temperature dependencies for GPP… and consider the RuBisCO enzyme.

Page 39: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Ribulose-1,5-bisphosphate carboxylase/oxygenase

(RuBisCO)

• Enzyme kinetics described by Woodrow and Berry, 1988, Ann. Rev. Plant Physiology and Plant Molecular Biology.

• Enzyme activity increases exponentially with temperature (eventual irreversible decrease in activity at high temperature, as proteins denature).

• RuBisCO catalyzes the fixation of CO2 (photosynthesis) and the fixation of O2 (leading to photorespiration).

• The carboxylase and oxygenase reactions have different Michaelis-Menten coefficients, and these coefficients have different temperature dependencies.

• These differences result in an optimal temperature for photosynthesis which does not depend directly on the temperature sensitivity for the enzyme activity.

Page 40: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

“The values listed… are calculated from measurements of the enzyme from spinach. There may, however, be small significant differences in the kinetic properties of Rubisco from different species of C3 plants.”

-Woodrow and Berry, 1988

GEP

Re

NEE

MR

HR GR

GEP

Re

NEE

MR

HR GR

Low Topt High Topt

Page 41: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Results: There is apparently a significant variation in the kinetic parameters for RuBisCO across the climate gradient.

Page 42: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Next steps in model parameterization and evaluation:

1. Find optimal kinetic parameters at each site, try to establish a dependency with simple climate indices that explains the variation in optimal temperature (underway).

2. Incorporate measurements from soil chambers and leaf-level gas exchange data as additional constraints. Test climate dependency of optimal temperature with these data (underway).

3. Expand the analysis to other plant functional types – deciduous broadleaf forest, evergreen broadleaf forest, and grasses (C3 and C4).

4. Bring in more extensive evaluation databases (inventory, remote sensing).

Page 43: Constructing and evaluating forward models of the terrestrial carbon cycle Peter Thornton Terrestrial Sciences Section Climate and Global Dynamics Division

Simulated Leaf Area Index for Evergreen Needleleaf Forest, NW U.S.

(Compute time approximately 70 hours running threaded code on lhotse)