fluxnet: measuring co 2 and water vapor fluxes across a global network dennis baldocchi...

FLUXNET: Measuring CO2 and Water Vapor Fluxes

Across a Global Network

Dennis Baldocchi

ESPM/Ecosystem Science Div.

University of California, Berkeley

IndoFlux, Chennai, India, July 2006

FLUXNET: From Sea to Shining Sea379 Sites, circa 2006

Global distribution of Flux Towers with Respect to Climate

Evolution of FLUXNET

• Measure Annual Cycle of NEE– Micromet issues of Detrending, Transfer Functions, Flux

Sampling and Measurements, Gap-filling, Error Assessment• Measure and Interpret Intra-annual Variation of NEE

– Flux partitioning (GPP & Reco); assessment of metadata,e.g. Vcmax, soil respiration, LAI, biomass inventories.

• Measure and Interpret Inter-annual variations of NEE• Measure NEE over multiple Land-Use Classes

– crops, grasslands, deciduous and evergreen broadleaf and conifer forests

– Disturbance: logging, biodiversity and fire• Manipulative Studies

– Nitrogen and H2O additions• Measure NEE over Representative Areas

– Scaling Flux Information of Footprint to MODIS pixel

Successes

• Mountains of data from a spectrum of canopy roughness conditions, functional types and climate spaces have been collected

• A Model for Data Sharing– FLUXNET Web Site, a venue for distributing Primary, Value-added and

Meta-Data products• Value-Added Products have been produced

– Development of Gap-Filling Techniques– Production of Gap-Filled Daily and Annual Sums

• Data for Validating and Improving SVAT models used for weather, climate, biogeochemistry and ecosystem dynamics

• Collaboration & Synthesis through Workshops and Hosting Visitors– Building a Collaborative, Cooperative, Multi-Disciplinary & International

Community of Researchers– Characterizing Annual C Fluxes– Environmental Controls on NEE

• Training New and Next Generation of Scientists, Postdocs, Students

‘Failures’/’Un-resolved’ Issues

• Not Measuring Night-time Fluxes Well• Not Measuring Fluxes over Complex terrain

and during Advection Well• ImPerfect U* correction

– New Gu Algorithm• ImPerfect Flux Partitioning

– Works Better on Longer Time Scales• ImPerfect Energy Balance Closure

– Could be ‘red-herring’• Need Better Outreach and Training• Needs Expansion into other Regions

– India– Africa

Visions with a Flux Measurement Network

• Processes– Canopy-Scale Response Functions

• Emergent Processes– Flux Partitioning, NEP=GPP-Reco

• Acclimation• Time

– Daily/Seasonal Dynamics– Pulses, Lags, Switches– Intra- + Interannual Variability– Stand Age/Disturbance

• Space– Climate/Structure/Function– Coherence/Gradients– Upscaling with Remote Sensing

• New Directions

Probability Statistics of NEE

Col 4 vs Col 5

NEE (gC m-2 y-1)

-1500 -1000 -500 0 500 1000

pdf

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

Published data (275 site years)Median: -229 gC m-2 y-1

Light and Photosynthesis:

Emergent Processes at Leaf and Canopy Scales

D208 Oak leaf, forest floorTleaf: 25o CCO2 : 360 ppm

Qpar (mol m-2 s-1)

0 200 400 600 800 1000 1200 1400 1600 1800

A (

mo

l m-2 s

-1)

0

2

4

6

8

10

12

data

model

model: clumped leaves

PPFD (mol m-2 s-1)

0 500 1000 1500 2000

Fc

( m

ol m

-2 s-1

)

-40

-30

-20

-10

0

10

measured

(a)

0 500 1000 1500 2000-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

absorbed PAR (µmol m-2 s-1)

WHEAT

Fc

(mg

m-2 s

-1)

Volcanoes, Aerosols + NEE

CO2 Flux and Diffuse Radiation

Niyogi et al., GRL 2004

Photosynthesis-Respiration

Processed by Falge

North America + Europe

Reco (gC m-2 yr-1)

0 200 400 600 800 1000 1200 1400 1600

GP

P (

gC m

-2 y

r-1)

0

200

400

600

800

1000

1200

1400

1600

b[1] 1.18r ² 0.893

Mean Summer Temperature (C)

5 10 15 20 25 30

Te

mp

era

ture

Op

tim

um

for

Ca

no

py C

O 2 u

pta

ke

(C

)

5

10

15

20

25

30

35

b[0] 3.192b[1] 0.923r ² 0.830

Analysis of E. Falge

NEE: Acclimation with TemperatureNEE: Acclimation with Temperature

GPP RH/Ca (mol m-2s-1)

0.00 0.01 0.02 0.03 0.04 0.05 0.06

gc (

mo

l m

-2s-1

)

0.0

0.2

0.4

0.6

0.8

Linking Water and Carbon:Potential to assess Gc with Remote Sensing

Xu + DDB

Temporal Dynamics of C Fluxes

• Hour• Day• Month• Season• Year• Multiple Years • Pulses

• Lags• Switches

Complicating Dynamical Factors

• Switches/Pulses– Rain

– Phenology/Length of Season

– Frost/Freezing

• Emergent Processes– Clouds & LUE

• Acclimation• Lags • Stand Age/Disturbance

Temperate Broadleaved Deciduous Forest

Day

0 50 100 150 200 250 300 350

NE

E (

gC

m-2

d-1

)

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

4

5

LAI=0GPP=0;Litterfall (+)Reco=f(litterfall)(+)

snow:

Tsoil(+)

GPP=0; Reco(+)

no snow

Tsoil (-)

Reco (-)

GP

P=f(LA

I, Vcm

ax )

late spring

early spring

Drought:(-)GPP(-); Re(-)

Clouds:PAR(-) GPP=f(PAR)(+)

Decadal Plus Time Series of NEE:Flux version of the Keeling’s Mauna Loa Graph

Harvard Forest

Year

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

CO

2 F

lux

De

nsity

(gC

m-2

d-1

)

-12

-8

-4

0

4

8 NEERecoGEE

Data of Wofsy, Munger, Goulden et al.

Re vs GPP

Interannual Variability in NEE

d GPP/dt

-300 -200 -100 0 100 200 300

d R

eco/

dt

-300

-200

-100

0

100

200

300

2002 2003 2004 2005

NE

E

[g C

m-2

wee

k-1 ]

-80

-70

-60

-50

-40

-30

-20

-10

0

10

20

HainichLeinefelde

Knohl et al Max Planck, Jena

Lag Effects Due to Drought/Heat Stress

Soroe, DenmarkBeech Forest1997

day

0 50 100 150 200 250 300 350

-10

-5

0

5

10

15

20

NEE, gC m-2 d-1

Tair, recursive filter, oC

Tsoil, oC

Data of Pilegaard et al.

Soil Temperature: An Objective Indicator of Phenology??

Data of: ddb, Wofsy, Pilegaard, Curtis, Black, Fuentes, Valentini, Knohl, Yamamoto. Granier, SchmidBaldocchi et al. Int J. Biomet, in press

Soil Temperature: An Objective Measure of Phenology, part 2

Temperate Deciduous Forests

Day, Tsoil >Tair

70 80 90 100 110 120 130 140 150 160

Da

y N

EE

=0

70

80

90

100

110

120

130

140

150

160

DenmarkTennesseeIndianaMichiganOntarioCaliforniaFranceMassachusettsGermanyItalyJapan

Spatial Variations in C Fluxes

Spatial Gradients:NEE and Length of Growing Season

Broad-Leaved Forests

Length of Growing Season

100 150 200 250

NE

E (

gC

m-2

yr-1

)

-800

-700

-600

-500

-400

-300

-200

-100

0

100Japan

Denmark

Italy

Massachusetts, USA

Belgium

Tennessee, USA

Prince Albert, CANADA

Ontario

Indiana, USA

Michigan, USA

Tower vs Satellite NDVI

0.00 0.25 0.50 0.75 1.000.00

0.25

0.50

0.75

1.00

NDVIMODIS

ND

VI

TO

WE

R

conifers deciduous

crops grassland

Falk et al., to be submitted

8 day means

Dai

ly g

ross

CO

2 fl

ux

(m

mol

m-2

day

-1)

0

200

400

600

800

1000

1200

1400

16008 day means

Dai

ly n

et C

O2 fl

ux

(mm

ol m

-2 d

ay-1

)

-400

-200

0

200

400

600

800

r2 = 0.92

8 day means

Dai

ly g

ross

LU

E

0.00

0.01

0.02

0.03

r2 = 0.65

Single clear days

AM net CO2 flux

(mmol m-2 hr-1)

0 20 40 60 80 100 120

Dai

ly n

et C

O2 fl

ux

(mm

ol m

-2 d

ay-1

)

-400

-200

0

200

400

600

800Single clear days

AM gross CO2 flux

(mmol m-2 hr-1)

0 20 40 60 80 100 120 140

Dai

ly g

ross

CO

2 fl

ux

(m

mol

m-2

day

-1)

0

200

400

600

800

1000

1200

1400

1600

r2 = 0.88

Single clear days

AM gross LUE

0.00 0.01 0.02 0.03

Dai

ly g

ross

LU

E

0.00

0.01

0.02

0.03

r2 = 0.73

r2 = 0.64

r2 = 0.56

Evergreen needleleaf forestDeciduous broadleaf forestGrassland and woody savanna

a b c

d e f

Sims et al 2005 AgForMet

Heinsch et al. IEEE 2006

Heinsch et al. IEEE 2006

Global MODIS Test

Limits to Landscape Classification by Functional Type

• Stand Age/Disturbance• Biodiversity• Fire• Logging• Insects/Pathogens• Management/Plantations• Kyoto Forests

Effects of Stand Age:After Logging

Law et al. 2003 Global Change Biology

Biodiversity and Evaporation

Temperate/Boreal Broadleaved ForestsSummer Growing Season

Number of Dominant Tree Species (> 5% of area or biomass survey)

1 2 3 4 5 6 7 8

E/

Ee

q

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

Baldocchi, 2004: Data from Black, Schmid, Wofsy, Baldocchi, Fuentes

Value of Flux Networks

• Documenting Change in Ecosystem Metabolism– Network acts as ‘canary in the mine’

• Produces Large and Long Data Sets– Reduced Sampling Error– Robust Dataset for Model Development

• Study Spectra of Time Scales– Capture Pulses and Lags

• Study Gradient of Climates, Structure and Function• Field of Dreams: ‘Build it and they will Come’

– Better Integrated Research Studies