Using AmeriFlux Observations in the NACP Site-level Interim Synthesis

1
Using AmeriFlux Observations in the NACP
Site-level Interim Synthesis

• Kevin Schaefer
• NACP Site Synthesis Team
• Flux Tower PIs
• Modeling Teams

2
Do models match observations? If not, why?
30 Models
47 Flux Tower Sites
36 AmeriFlux 11 Fluxnet Canada
24 submitted output 10 runs per site
3
Analysis Projects
Published
Submitted
4
Products Derived from Flux Data

• Gap-filled observed weather (Ricciuto et al.)
• BADM files (everyone)
• Gap-filled fluxes Uncertainty (Barr et al.)
• Random
• U threshold
• Gap-filling Algorithm
• Partitioning Algorithm

5
Random Uncertainty (Barr et al.)
Needleleaf Forest Broadleaf Forest Mixedwood
Forest Juvenile Forest Wetland Grassland Shrublan
d Cropland
? USA ? Canada
Annual eNEP (g C m-2 y-1)
Annual Re (g C m-2 y-1)

• Random eNEP 4 Re
• Uth eNEP 1.3 Re

6
BADM Files

• Extremely useful to modelers
• Soil texture
• Site history
• Initial pools sizes
• Leaf Area Index
• We strongly encourage more submissions

7
Weather Uncertainty (Ricciuto et al.)

• Bias in radiation produces bias in GPP

8
Agriculture Sites (Lokupitiya et al.)
US-Ne3
Soybean
Corn
Corn
Soybean

• Need crop specific parameterizations

9
Wetland Sites (Desai et al.)

• Residuals correlate to water table depth
• Models should include water table dynamics

10
Spectral NEE Error (Dietze et al.)
Annual
Month
Diurnal
Synoptic
Not Significant

• Error peak at diurnal annual time scales
• Errors at synoptic monthly time scales

11
NEE Wavelet Coherence (Stoy et al.)
SiB at US-UMB
Annual
Significant
Month
Synoptic
Time Scale (hours)
Diurnal
Hour

• Models match observations only some of the time

12
NEE Seasonal Cycle (Schwalm et al.)
0.6
Add soil layers
0.5
Taylor Skill
0.4
0
11
15
1
2
3
7
9
10
Number Soil Layers
13
Phenology (Richardson et al.)

• Early/late uptake means positive GPP bias
• Models need better phenology

14
Regional vs. Site (Raczka et al.)
Flux Towers
Light Use Efficiency
Enzyme Kinetic

• Enzyme kinetic models biased high
• LUE models biased low

15
GPP Annual Bias (Schaefer et al.)
US-Me2 Light Use Efficiency Curve
Observed
Simulated
Daily Average GPP (mmol m-2 s-1)
Daily Average Shortwave Radiation (W m-2)

• Slope of LUE Curve drives Annual bias
• Models need better Vmax, leaf-to-canopy scaling,

16
Areas For Model Development

• Better Phenology
• More soil layers
• More vegetation pools
• Slopes to LUE curve
• Water table dynamics
• Crop parameterizations

17
Extra Slides
18
Annual GPP Bias due to Phenology
Evergreen sites
Deciduous sites
4080
160145
-565
75130
19
Multi-Model wavelet Coherence
Scale (hours)
20
NEE Seasonal Cycle (Schwalm et al.)
Our 1st published paper!
Perfect Model
Taylor Skill
Normalized Mean Absolute Error
Chi-squared
21
Uth vs. Random Uncertainty (Barr et al.)
Needleleaf Forest Broadleaf Forest Mixedwood
Forest Juvenile Forest Wetland Grassland Shrublan
d Cropland
? USA ? Canada
Uth Annual eNEP (g C m-2 y-1)
Random Annual eNEP (g C m-2 y-1)