Interannual Variability in the ChEAS Mesonet

1
Interannual Variability in the ChEAS Mesonet
ChEAS XI, 12 August 2008 UNDERC-East, Land O
Lakes, WI Ankur Desai Atmospheric Oceanic
Sciences, University of Wisconsin-Madison
2
Whats the Deal?

• Interannual variation (IAV) in carbon fluxes from
  land to atmosphere are significant at most flux
  sites
• Key to understanding how climate affects
  ecosystems comes from modeling IAV
• IAV (years-decade) is currently poorly modeled,
  while hourly, seasonal, and even successional
  (century) are better

3
Can we simulate this?
4
Sipnet

• A simplified model of ecosystem carbon / water
  and land-atmosphere interaction
• Minimal number of parameters
• Driven by meteorological forcing
• Still has gt60 parameters
• Braswell et al., 2005, GCB
• Sacks et al., 2006, GCB added snow
• Zobitz et al., 2008

5
Results
6
2 years 7 years
1997
1998
1999
2000
2001
2002
2003
2004
2005
7
Ricciuto et al.
8
Ricciuto et al.
9
Our region
10
Any coherence?
Desai et al, 2008, Ag For Met
11
Cross-site IAV

• Hypothesis IAV in flux towers in the same region
  are coherent in time
• Hypothesis Simple climate driven models can
  explain this IAV
• Growing season length
• Climate thresholds
• Mean annual precip

12
A whole bunch of data
13
Coherence?
14
Growing season and IAV

• Does growing season start explain IAV?
• Can a very simple model be constructed to explain
  IAV?
• Hypothesis growing season length explains IAV
• Can we make a cost function more attuned to IAV?
• Hypothesis MCMC overfits to hourly data

15
Hello again
16
The model

• Driven by PAR, Air and Soil T, VPD, (Precip)
• LUE based GPP model f(PAR,T,VPD)
• Three respiration pools f(T, GPP)
• Output NEE, ER, GPP, LAI
• Sigmoidal GDD function for leaf out
• Sigmoidal Soil T function for leaf off
• 17 parameters, 3 are fixed
• Desai et al., in prep (a)

17
The optimizer

• All flux towers with multiple years of data
• Estimate parameters with Markov Chain Monte Carlo
  (smart random walk)
• Written in IDL

18
MCMC

• MCMC is an optimizing method to minimize
  model-data mismatch
• Quasi-random walk through parameter space
  (Metropolis)
• Start at many random places (Chains) in prior
  parameter space
• Move downhill to minima in model-data RMS by
  randomly changing a parameter from current value
  to a nearby value
• Avoid local minima by occasionally performing
  uphill moves in proportion to maximum
  likelihood of accepted point
• Use simulated annealing to tune parameter space
  exploration
• Pick best chain and continue space exploration
• Requires 100,000-500,000 model iterations (chain
  exploration, spin-up, sampling)
• End result best parameter set and confidence
  intervals (from all the iterations)
• Cost function compared to observed NEE

19
New cost function

• Original log likelihood computes sum of squared
  difference at hourly timestep
• What if we also added monthly and annual squared
  differences to this likelihood?
• Have to scale these less frequent values
• Have to deal with missing data

20
I like likelihood
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Regional IAV

• How well do we know regional (scaled-up) IAV?
• Do top-down and bottom-up regional flux
  estimation techniques agree on IAV (if not
  magnitude)?
• What controls regional IAV?
• Wetland IAV vs Upland IAV
• Step 1 Scale the towers

27
(No Transcript)
28
Heterogeneous footprint
29
Scaling with towers

• NEP (-NEE) at 13 sites
• Stand age matters
• Ecosystem type matters
• Is interannual variability coherent?
• Are we sampling sufficient land cover types?

30
Desai et al., 2008, AFM

• Multi-tower synthesis aggregation
• parameter optimization with minimal 2 equation
  model

31
Tall tower downscaling

• Wang et al., 2006

32
Scaling evaluation

• Desai et al., 2008

33
Next step

• Use our IAV model with all 17 (19) flux towers -
  estimate parameters for each
• Use better landcover and better age distribution
  from NASA project
• Upscale again - this time over long time period
• This experiment for Northern Highlands 1989-2007
  (Buffam et al., in prep)

34
(No Transcript)
35
(No Transcript)
36
Many years of flux
37
Regional coherence?

• Desai et al., in prep

38
Regional coherence?
39
Conclusions

• There is some coherence in IAV across ChEAS
• Better statistical method to show this?
• A simple model with explicit phenology can
  capture the IAV across sites only with a better
  likelihood function
• Next step Simple model with fixed phenology
• Limited convergence on IAV from regional methods

40
Other things

• Sulman et al., in prep - the role of wetlands in
  regional carbon balance
• Lake Superior carbon balance from ABL budgets
  (Atilla, McKinley) - Urban et al, in prep
• Small lakes in the landscape (Buffam, Kratz)
• Successional trends and modeling (Dietze)
• Hyperspectral remote sensing (Townsend, Serbin,
  Cook)
• Top-down CO2 budgets in valeys and complex
  terrain (Stephens, Schimel, Bowling, deWekker)
• CH4 (pending), advection (pending - Yi), urban
  micromet and biogeochem (pending)
• NEON? (Schimel, UNDERC)

41
Thanks

• Desai lab http//flux.aos.wisc.edu
• Ben Sulman, Jonathan Thom, Shelly Knuth
• DOE NICCR, NSF, UW, DOE, NASA, USFS, Northern
  Research Station, Kemp NRS
• All the tower people

42
(No Transcript)