Climate data

Tower Data
Surface Reflectance: MODIS
Validation &
Optimization

LSWI

EVI

GEE = ( Tscale Wscale Pscale ) FAPARPAV 1/(1+PAR/PAR 0) PAR

Respiration = x Ta +

Figure 1: Schematic diagram of the Vegetation Photosynthesis Respiration Model (VPRM). EVI denotes
Enhanced Vegetation Index, LSWI-Land Surface Water Index; FPARPAV-the fraction of incident light
absorbed by the photosynthetically active vegetation in the canopy; Tscale, Pscale and Wscale-scalars for
temperature, leaf phenology and canopy water content, respectively. GEE (Gross Ecosystem Exchange) is
the light-dependent part of Net Ecosystem Exchange, R (Respiration) is the light-independent part. MODIS
refers to the Moderate Resolution Imaging Spectroradiometer onboard the NASA Terra and Aqua satellites;
PAR0, , and are the four model parameters, one set per vegetation type.

Harvard Forest

20

Harvard Forest

10

(moles/m2/s )

0
10

OBS
VPRM
GEE
RESP

15

30

20

(moles/m2/s )

10

NEE Obs
NEE VPRM
GEE
RESP

182

184

186

188

2000

190

2001

2002

2003

2004

Date

Day of 2000
BOREAS

BOREAS

4
0
2

(moles/m2/s )

10

(moles/m2/s )

NEE Obs
NEE VPRM
GEE
RESP

174

175

176

177
Day of 2000

178

179

180

2000

2001

2002

2003

2004

2005

Date

Figure 2a: (left panels) Examples of hourly output from the VPRM for Harvard Forest (upper) and
NOBS (lower). Observations (points) show u*-filtered hourly data only. (right panels) Comparison
between the observed (solid-square) and VPRM (blue line) monthly mean NEE (mole m-2 s-1) over the
four-year period at these two calibration sites. The VPRM provides a consistent separation of the lightdependent part of NEE ("GEE") from the light-independent part (Respiration) across all sites.

Figure 2b: A comparison between the observed (solid-square) and VPRM (open-circle) mean diurnal
diurnal variation of NEE (mole m-2 s-1) during the peak growing season at Calibration sites.

10

12

10

12

10

12

0.30
0.25

HARVARD 2002 2003 ; rsq= 0.96 ( 0.96 )

HOWLAND 2001 2004 ; rsq= 0.33 ( 0.33 )

LUCKY1 2001 2004 ; rsq= 0.82 ( 0.36 )

TONZI 2001 2004 ; rsq= 0.65 ( 0.57 )

12

0.12

0.9

10

0.10
0.08
0.06

0.2

12

Month

VAIRA 2001 2004 ; rsq= 0.57 ( 0.44 )

PEATLAND 2001 2004 ; rsq= 0.54 ( 0.5 )

0.08

0.8

Index

0.4

0.4

0.04

Index

0.6

0.06

0.8

0.3
0.2

0.6
0.3

6
Month

0.00

0.2

0.5 0.0

0.0

0.2

0.02

0.2
0.0

0.0

6
Month

10

12

0.0
1.0

0.5

NEE (moles/m2/s )

1
3
4

3
4
0

NEE (moles/m2/s )

0
1
2

NEE (moles/m2/s )

5
10

NEE (moles/m2/s )

0.05

0.1

VPRM prefactor

0.6
0.4

Index

0.8

0.4
0.3
0.2
0.1

VPRM prefactor

0.0

0.5

1.0

0.5

1.0
0.8
0.6
Index
0.4
0.2
0.0

0.5
1.5

SOY 2001 2004 ; rsq= 0.65 ( 0.61 )

CORN 2001 2004 ; rsq= 0.93 ( 0.94 )

12

2.0

0.25

10

0.20

0.15

6
Month

0.10

1.0

1.0

NEE (moles/m2/s )

0.0

NEE (moles/m2/s )

0.5
1.0

0.6

12

VPRM prefactor

10

0.5

Index

0.04
0.02

0.2
0.5

0
1
2

NEE (moles/m2/s )

3
4
8

10

0.4

0.08
0.06

VPRM prefactor

0.6

Index

0.4

0.05
0.00

0.2
0.0

0.0

6
Month

12

0.7

0.8

0.20
0.15
0.10

VPRM prefactor

0.6
Index
0.4

0.1
0.2
2 0.0
6

10

0.8

0.10

0.25

1.0
0.8

0.3
0.2

VPRM prefactor

0.8
0.6
Index
0.4

1.0

Month

VPRM prefactor

1.0
1.5
2.0

NEE (moles/m2/s )

0
2

0.20

0.5

Month

NEE (moles/m2/s )

0.35

1.0
0.9
0.8
0.4

0.02

2.5

Month

NEEobs
NEEVPRM
Pscale
Wscale
Tscale
EVI
PreFac

0.7

Index

0.6

0.08
0.06
0.04

0.5

0.10

0.12

0.8
0.2
0
1

NEE (moles/m2/s )

2
3

6
Month

10

12

VPRM prefactor

Month

6
Month

10

12

10

12

Month

Figure 3: (upper sections) Seasonal dynamics of the pre-factors Tscale, Pscale, and Wscale in the VPRM equation
(Eq. 12), driven by satellite and meteorological data. All factors are significant, at various times and places.
(lower sections) Comparison of the seasonal dynamics between observed (black) and VPRM (red) monthly
mean NEE (mole m-2 s-1) at calibration sites, monthly means , averaged over all years. Values in the title
bar give the fraction of the total variance of the mean seasonal cycle of NEE captured by the model, [1
var( NEEobs NEEVPRM)/var( NEEobs)] (in parenthesis, the same quantity for the time series of individual
monthly means). Note that net uptake of CO2 corresponds to negative values of NEE.

VPRM prefactor

0.6

Index

0.4

0.10

VPRM prefactor

0.05
0.00

0.5
0.0
0.5

NEE (moles/m2/s )

1.0
1.5

0.14

0.15
0.8
0.6
Index
0.4
0.2
0.0

0.00

0.0
0.5

NEE (moles/m2/s )

1.0

1.0

DONALD 2001 2004 ; rsq= 1.72 ( 1.04 )

VPRM prefactor

1.0

0.10
0.06
0.02

0.04

VPRM prefactor

0.08

0.8
0.6
Index
0.4
0.2
0.5 0.0

OREGON 2001 2004 ; rsq= 0.72 ( 0.55 )

1.0

NIWOT 2001 2004 ; rsq= 0.39 ( 0.25 )

1.0

BOREAS 2001 2004 ; rsq= 0.87 ( 0.83 )

Figure 4: Comparison between the observed (solid-square) and predicted (open-circle) diurnal mean
NEE (mole m-2 s-1) over the peak photosynthetically active period, as in Fig. 2b, but at validation sites
without adjustment of parameters.

0
4

2
4
Month
6
8
10
12
0
2
4
Month
6
8
10
12

0
2
4
Month
6
8
10
12

0.15

0.20

0.25

0.30

NEE (moles/m2/s )

NEE (moles/m2/s )
0

0.2

0.2

0.6

Index

0.3

0.6

0.2

Index

0.6

Index

VPRM prefactor

0.4

VPRM prefactor

0.4

0.1

0.4

0.2

0.1

0.15

0.6

VPRM prefactor

0.10

Index

0.4

0.8

0.3

0.8

0.8

0.20

0.8

1.0

1.0

1.0

0.25

1.0

Month
Month
Month

4
Month

Month
6

Figure 5: A comparison of the seasonal dynamics of model and observed NEE, as in Fig. 3, but at
validation sites (no adjustment of VPRM parameters).
8

10

LCREEK 2001 2004 ; rsq= 0.77 ( 0.53 )

INDIANA 2003 2003 ; rsq= 0.75 ( 0.59 )
WCREEK 2001 2002 ; rsq= 0.86 ( 0.77 )
DUKEHW 2002 2004 ; rsq= 0.85 ( 0.64 )

Month
Month
Month
Month

10

SOYBOND 2001 2004 ; rsq= 0.83 ( 0.8 )

CORNBOND 2004 2004 ; rsq= 0.73 ( 0.76 )
ANLGRASS 2000 2003 ; rsq= 0.13 ( 0.4 )
WLEF 2001 2004 ; rsq= 9.55 ( 11.01 )

10
12

VPRM prefactor

0.05 0.10 0.15 0.20 0.25 0.30 0.35

0.0

0.0

0.0

0.05

0.00

0.4

2.0

1.5

0.5

0.0

NEE (moles/m2/s )

0.5

NEE (moles/m2/s )

1.0

0.0

0.0

NEE (moles/m2/s )

0.5

NEE (moles/m2/s )
1.0

1.0

1.5

0.5

0.5

0.5

0.00

0.0

0.00

0.0

0.00

0.0

0.1

0.4

0.02

0.2

0.02

0.2

0.02

0.2

Index

0.06

0.6

0.6

0.06

Index

0.08

0.6

0.2

0.3
VPRM prefactor

0.6

Index

VPRM prefactor

0.04

0.4

VPRM prefactor

0.06

Index

0.04

0.4

VPRM prefactor

0.04

0.4

0.8

0.10

0.8

0.08

0.8

0.08

0.8

0.4

1.0

0.12

1.0

0.10

1.0

0.10

1.0

EOBS 2001 2001 ; rsq= 0.71 ( 0.88 )

VPRM prefactor

12

0.10

1.0

0.2

1 0.0

12

0.05

0.8

10

0.6

10

Index

0.4

0.00

0.2

1 0.0

0.15

0.10

VPRM prefactor

12

0.05

0.8

NEE (moles/m2/s )

12

0.00

0.6

10

Index

NEE (moles/m2/s )
3

NEEobs
NEEVPRM
Pscale
Wscale
Tscale
EVI
PreFac

0.4

10

0.2

B1850 2000 2000 ; rsq= 0.82 ( 0.84 )

0.5

1.0

NEE (moles/m2/s )

0.4

0.8

0.3

0.6

2
4

Index

0
2

VPRM prefactor

0.2

0.4

0.4

1.0

0.3

0.8

12

0.1

0.2
VPRM prefactor

0.6

10
12

0.0

Index
0.4

8
10

0.2

0.1

0.2

6
8

0.0

0.0

0.0

4
6

NEE (moles/m2/s )

2
4

NEE (moles/m2/s )

0
2

10

NEE (moles/m2/s )
4

SOBS 2001 2004 ; rsq= 0.71 ( 0.81 )

DUKEPP 2000 2004 ; rsq= 1.19 ( 0.43 )

12

12

Growing Season Mean

NEE
2

Monthly NEE (moles/m2/s)

(0,1)

VPRM

VPRM

(moles/m /s )

10

(0,1)

(0.49, 1.08)

15

(0.03,1.1)

15

10

5
Obs

Obs

Figure 6: (left panel) Observed and predicted monthly mean NEE (mole m-2 s-1) for calibration sites (solid
symbols) and validation sites (open symbols) excluding WLEF. The regression line for all sites (dotted line)
is very similar to the regression for validation sites only (dashed line). (right panel) Mean NEE by site
(except WLEF) for the growing season.. The line labeled (0,1) has zero intercept and slope =1 ("1:1 line").
Regression lines are labeled similarly

a)

b)

Figure 7a-b: A comparison between GEE (mole m-2 s-2) obtained by fitting VPRM, MODEL-2,
MODEL-1 and MODEL-0 to tower data from 2000 to 2003 at Harvard Forest. (upper, 7a) and
NOBS/BOREAS (upper, 7a). The VPRM incorporates EVI, Pscaleand Wscale (driven by satellite data), plus
Tscale driven by meteorological data. MODEL-2 drops LSWI factors (Pscaleand Wscale) , MODEL-1 drops
all satellite indices (EVI, Pscaleand Wscale) and MODEL-0 drops all of these plus Tscale.