Towards Bayesian uncertainty quantification for forest models used - - PowerPoint PPT Presentation

▶
towards bayesian uncertainty quantification for forest
SMART_READER_LITE
LIVE PREVIEW

Towards Bayesian uncertainty quantification for forest models used - - PowerPoint PPT Presentation

Jun 26, 2023 49 likes •303 views

Towards Bayesian uncertainty quantification for forest models used in the U.K. GHG inventory for LULUCF Marcel van Oijen & Amanda Thomson Centre for Ecology and Hydrology Edinburgh, U.K. Contents 1. Current methodology used to make the

slide-1
SLIDE 1

Towards Bayesian uncertainty quantification for forest models used in the U.K. GHG inventory for LULUCF

Marcel van Oijen & Amanda Thomson Centre for Ecology and Hydrology Edinburgh, U.K.

slide-2
SLIDE 2

Contents

  • 1. Current methodology used to make

the LULUCF inventory

  • 2. Research on alternatives:
  • Forest models that include the effects of

climate & soil conditions

  • Bayesian uncertainty quantification
  • 3. Conclusions & outlook
slide-3
SLIDE 3

Primary information flows

LUC soil C stock changes (Tier 3) Forest C stock changes (Tier 3) GHG emissions / C stock changes (Tier 2)

slide-4
SLIDE 4

Primary information flows

Data CFLOW

LUC soil C stock changes (Tier 3) Forest C stock changes (Tier 3) GHG emissions / C stock changes (Tier 2)

Dynamic soil model Data

S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F

  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass

EF EF

slide-5
SLIDE 5

Uncertainty propagation

Data CFLOW

LUC soil C stock changes (Tier 3) Forest C stock changes (Tier 3) GHG emissions / C stock changes (Tier 2)

Dynamic soil model Data

S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F

  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass

EF EF

Uncertain inputs (pdf’s) Uncertain outputs

slide-6
SLIDE 6

CFLOW

Data CFLOW

Forest C stock changes (Tier 3)

  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Woody biomass Non-woody biomass Woody litter Non-woody litter Soil organic matter Wood products thinning and harvesting Transfer of residues to soil Natural mortality Thinnings Harvest debris

Yield tables & expansion factors

Woody biomass Non-woody biomass Woody litter Non-woody litter Soil organic matter Wood products thinning and harvesting Transfer of residues to soil Natural mortality Thinnings Harvest debris

Yield tables & expansion factors

slide-7
SLIDE 7

CFLOW

Data CFLOW

Forest C stock changes (Tier 3)

  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

CFLOW forest model :

  • Simple C-pools model with first order flows
  • Robust
  • Coarse (no effects climate & soil conditions)
  • Uncertainty quantification …
slide-8
SLIDE 8

CFLOW: Uncertainty quantification by MC

Data CFLOW

Forest C stock changes (Tier 3)

  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over
Initial Csoluble
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20 Initial Cstarch
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20 Initial Csoluble
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20 Initial Cstarch
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20 Initial Csoluble
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20 Initial Cstarch
  • 0.05
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.00 0.05 0.10 0.15 0.20
  • 1. Quantify pdf’s for all

inputs

  • 2. Take sample

from pdf’s

  • 3. Run model

1000 times

slide-9
SLIDE 9

BASFOR: process-based forest model

Data CFLOW

Forest C stock changes (Tier 3)

BASFOR

  • Parameters (many!)
  • Input variables (many!)

Atmosphere Tree Soil Subsoil

H2O H2O H2O H2O C C C N N N N

Atmosphere Tree Soil Subsoil

H2O H2O H2O H2O C C C N N N N

Wind speed Humidity Rain Temperature Radiation CO2 N-deposition BASFOR inputs:

slide-10
SLIDE 10

BASFOR: process-based forest model

Data CFLOW

Forest C stock changes (Tier 3)

BASFOR

  • Parameters (many!)
  • Input variables (many!)

BASFOR forest model :

  • Preprocessor for CFLOW (effects of climate,

N-deposition, [CO2])

  • … or replacement of CFLOW?
  • But is BASFOR robust enough ?
  • More inputs & parameters => more

demanding uncertainty quantification

slide-11
SLIDE 11

Process-based models

Data CFLOW

Forest C stock changes (Tier 3)

BASFOR

  • Parameters (many!)
  • Input variables (many!)

BASFOR forest model :

  • Preprocessor for CFLOW (effects of climate,

N-deposition, [CO2])

  • … or replacement of CFLOW?
  • But is BASFOR robust enough ?
  • More inputs & parameters => more

demanding uncertainty quantification Bayesian calibration & uncertainty quantification

slide-12
SLIDE 12

The Bayesian approach

Uncertainty propagation

Data Data CFLOW CFLOW

LUC soil C stock changes (Tier 3) Forest C stock changes (Tier 3) GHG emissions / C stock changes (Tier 2)

Dynamic soil model Dynamic soil model Data Data

S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F
  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass
  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass

EF EF EF EF

Uncertain inputs (pdf’s) Uncertain outputs

Bayesian calibration

slide-13
SLIDE 13

The Bayesian approach

Uncertainty propagation

Data Data CFLOW CFLOW

LUC soil C stock changes (Tier 3) Forest C stock changes (Tier 3) GHG emissions / C stock changes (Tier 2)

Dynamic soil model Dynamic soil model Data Data

S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F S G C F
  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass
  • Afforestation rates

(FC, NIDA)

  • Yield tables (FC)

Parameters:

  • Production
  • Turn-over

Parameters Land-use areas:

  • CS, MLC
  • NICS, AC & FS

Country-specific data on EFs & activities:

  • Deforestation
  • Liming
  • Lowland drainage
  • Peat extraction
  • Non-forest biomass

EF EF EF EF

Uncertain inputs (pdf’s) Uncertain outputs

Bayesian calibration

P(ϑ|D) = P(ϑ) P(D|ϑ) / P(D)

Posterior pdf for the parameters Prior pdf for the parameters Likelihood of the data Scaling constant ( = ∫ P(ϑ) P(D|ϑ) d ϑ )

slide-14
SLIDE 14

The Bayesian approach

P(ϑ|D) = P(ϑ) P(D|ϑ) / P(D)

Bayes’ Theorem implemented using MCMC (Metropolis algorithm)

slide-15
SLIDE 15

Using data in Bayesian calibration of BASFOR

1 2 3 x 10 4 500 1000 VolTot (m3 ha-1) 1 2 3 x 10 4 500 1000 Vol (m3 ha-1) 1 2 3 x 10 4 10 20 30 CtreeTot (kg m-2) 1 2 3 x 10 4 10 20 30 Ctree (kg m-2) 1 2 3 x 10 4 5 10 Cstem (kg m-2) 1 2 3 x 10 4 1 2 Cbranch (kg m-2) 1 2 3 x 10 4 0.5 1 Cleaf (kg m-2) 1 2 3 x 10 4 2 4 Croot (kg m-2) 1 2 3 x 10 4 10 20 30 h (m) 1 2 3 x 10 4 2 4 LAI (m2 m-2)

Time

1 2 3 x 10 4 5 10 15 Csoil (kg m-2)

Time

1 2 3 x 10 4 0.2 0.4 0.6 Nsoil (kg m-2)

Time

0.5 1 x 10
  • 3
1000 2000 CB0T 1 2 x 10
  • 3
1000 2000 CL0T 2 4 x 10
  • 3
1000 2000 CR0T Param eter m arginal probability distributions 1 2 x 10
  • 3
500 1000 CS0T 0.4 0.6 0.8 500 1000 BETA 300 350 400 500 1000 CO20 0.25 0.3 0.35 500 1000 FB 0.25 0.3 0.35 1000 2000 FLMAX 0.25 0.3 0.35 1000 2000 FS 0.4 0.6 0.8 500 1000 GAMMA 5 10 15 500 1000 KCA 0.35 0.4 0.45 500 1000 KCAEXP 2 4 x 10
  • 4
500 1000 KDBT 5 x 10
  • 4
1000 2000 KDRT 5 10 500 1000 KH 0.2 0.3 0.4 500 1000 KHEXP 1 2 x 10
  • 3
500 1000 KNMINT 1 2 x 10
  • 3
500 1000 KNUPTT 0.02 0.03 0.04 500 1000 KTA 10 20 30 500 1000 KTB 0.5 1 1000 2000 KEXTT 4 6 8 1000 2000 LAIMAXT 1 2 3 x 10
  • 3
500 1000 LUET 0.01 0.02 0.03 1000 2000 NCLMINT 0.02 0.04 0.06 500 1000 NCLMAXT 0.02 0.03 0.04 1000 2000 NCRT 0.5 1 1.5 x 10
  • 3
500 1000 NCW T 6 8 10 1000 2000 SLAT 4 6 8 500 1000 TRANCOT 150 200 250 1000 2000 W OODDENS 0.5 1 500 1000 CLITT0 6 8 10 500 1000 CSOMF0 1 2 3 500 1000 CSOMS0 0.01 0.02 500 1000 NLITT0 0.2 0.3 0.4 1000 2000 NSOMF0 0.1 0.2 500 1000 NSOMS0 1 2 x 10
  • 3
1000 2000 NMIN0 0.4 0.6 0.8 500 1000 FLITTSOMF 0.05 0.1 500 1000 FSOMFSOMS 2 4 x 10
  • 3
1000 2000 KDLITT 1 2 x 10
  • 4
500 1000 KDSOMF 1 2 x 10
  • 5
500 1000 KDSOMS 1 2 3 x 10 4 500 1000 VolTot (m3 ha-1) 1 2 3 x 10 4 500 1000 Vol (m3 ha-1) 1 2 3 x 10 4 10 20 30 CtreeTot (kg m-2) 1 2 3 x 10 4 10 20 30 Ctree (kg m-2) 1 2 3 x 10 4 5 10 Cstem (kg m-2) 1 2 3 x 10 4 1 2 Cbranch (kg m-2) 1 2 3 x 10 4 0.5 1 Cleaf (kg m-2) 1 2 3 x 10 4 2 4 Croot (kg m-2) 1 2 3 x 10 4 10 20 30 h (m) 1 2 3 x 10 4 2 4 LAI (m2 m-2)

Time

1 2 3 x 10 4 5 10 15 Csoil (kg m-2)

Time

1 2 3 x 10 4 0.2 0.4 0.6 Nsoil (kg m-2)

Time

5 x 10
  • 3
2000 4000 CB0T 0.005 0.01 2000 4000 CL0T 0.005 0.01 2000 4000 CR0T Prior param eter m arginal probability distributions (beta) 5 x 10
  • 3
2000 4000 CS0T 0.4 0.6 0.8 1000 2000 BETA 300 350 400 1000 2000 CO20 0.25 0.3 0.35 1000 2000 FB 0.25 0.3 0.35 1000 2000 FLMAX 0.25 0.3 0.35 1000 2000 FS 0.4 0.6 0.8 1000 2000 GAMMA 5 10 15 1000 2000 KCA 0.35 0.4 0.45 1000 2000 KCAEXP 2 4 x 10
  • 4
1000 2000 KDBT 0.5 1 x 10
  • 3
1000 2000 KDRT 10 20 2000 4000 KH 0.2 0.3 0.4 1000 2000 KHEXP 1 2 x 10
  • 3
1000 2000 KNMINT 1 2 x 10
  • 3
1000 2000 KNUPTT 0.02 0.03 0.04 1000 2000 KTA 10 20 30 1000 2000 KTB 0.5 1 1000 2000 KEXTT 4 6 8 2000 4000 LAIMAXT 1 2 3 x 10
  • 3
1000 2000 LUET 0.01 0.02 0.03 1000 2000 NCLMINT 0.02 0.04 0.06 1000 2000 NCLMAXT 0.02 0.03 0.04 1000 2000 NCRT 1 2 x 10
  • 3
1000 2000 NCW T 20 40 2000 4000 SLAT 4 6 8 1000 2000 TRANCOT 150 200 250 1000 2000 W OODDENS 0.5 1 1000 2000 CLITT0 6 8 10 1000 2000 CSOMF0 1 2 3 1000 2000 CSOMS0 0.01 0.02 1000 2000 NLITT0 0.2 0.3 0.4 1000 2000 NSOMF0 0.1 0.2 1000 2000 NSOMS0 1 2 x 10
  • 3
1000 2000 NMIN0 0.4 0.6 0.8 1000 2000 FLITTSOMF 0.05 0.1 2000 4000 FSOMFSOMS 2 4 x 10
  • 3
1000 2000 KDLITT 1 2 x 10
  • 4
1000 2000 KDSOMF 1 2 x 10
  • 5
1000 2000 KDSOMS

Prior pdf

Data Bayesian calibration

Posterior pdf

Dodd Wood Dodd Wood

slide-16
SLIDE 16

D a t a B a y e s i a n c a l i b r a t i o n D a t a B a y e s i a n c a l i b r a t i o n

5 x 1
  • 3
2 0 0 4 0 0 C B 0 T 0 . 0 0 5 0 . 0 1 2 0 0 0 4 0 0 0 C L 0 T 0 . 0 0 5 . 0 1 2 0 0 4 0 0 C R 0 T P r i o r p a r a m e t e r m a r g i n a l p r o b a b i l i t y d i s t r i b u t i o n s ( b e t a ) 5 x 1
  • 3
2 0 0 4 0 0 C S 0 T 0 . 4 0 . 6 . 8 1 0 0 0 2 0 0 0 B E T A 3 0 3 5 4 0 0 1 0 0 2 0 0 C O 2 0 . 2 5 0 . 3 . 3 5 1 0 0 2 0 0 F B . 2 5 0 . 3 0 . 3 5 1 0 0 0 2 0 0 0 F L M A X . 2 5 . 3 . 3 5 1 0 0 2 0 0 F S . 4 0 . 6 0 . 8 1 0 0 2 0 0 G A M M A 5 1 1 5 1 0 0 0 2 0 0 0 K C A . 3 5 . 4 0 . 4 5 1 0 0 2 0 0 K C A E X P 2 4 x 1
  • 4
1 0 0 2 0 0 K D B T 0 . 5 1 x 1 0
  • 3
1 0 0 0 2 0 0 0 K D R T 1 0 2 2 0 0 4 0 0 K H . 2 0 . 3 0 . 4 1 0 0 2 0 0 K H E X P 1 2 x 1 0
  • 3
1 0 0 0 2 0 0 0 K N M I N T 1 2 x 1 0
  • 3
1 0 0 2 0 0 K N U P T T 0 . 0 2 0 . 0 3 . 0 4 1 0 0 2 0 0 K T A 1 2 3 0 1 0 0 0 2 0 0 0 K T B . 5 1 1 0 0 2 0 0 K E X T T 4 6 8 2 0 0 4 0 0 L A I M A X T 1 2 3 x 1 0
  • 3
1 0 0 0 2 0 0 0 L U E T . 0 1 0 . 0 2 0 . 0 3 1 0 0 2 0 0 N C L M I N T 0 . 0 2 0 . 0 4 . 0 6 1 0 0 2 0 0 N C L M A X T . 0 2 . 0 3 0 . 0 4 1 0 0 0 2 0 0 0 N C R T 1 2 x 1
  • 3
1 0 0 2 0 0 N C W T 2 0 4 2 0 0 4 0 0 S L A T 4 6 8 1 0 0 0 2 0 0 0 T R A N C O T 1 5 2 0 2 5 0 1 0 0 2 0 0 W O O D D E N S 0 . 5 1 1 0 0 2 0 0 C L I T T 0 6 8 1 0 1 0 0 0 2 0 0 0 C S O M F 1 2 3 1 0 0 2 0 0 C S O M S 0 . 0 1 . 0 2 1 0 0 2 0 0 N L I T T 0 0 . 2 0 . 3 . 4 1 0 0 0 2 0 0 0 N S O M F . 1 0 . 2 1 0 0 2 0 0 N S O M S 1 2 x 1
  • 3
1 0 0 2 0 0 N M I N 0 . 4 0 . 6 . 8 1 0 0 0 2 0 0 0 F L I T T S O M F 0 . 0 5 0 . 1 2 0 0 4 0 0 F S O M F S O M S 2 4 x 1
  • 3
1 0 0 2 0 0 K D L I T T 1 2 x 1 0
  • 4
1 0 0 0 2 0 0 0 K D S O M F 1 2 x 1 0
  • 5
1 0 0 2 0 0 K D S O M S 1 2 3 x 1 0 4 5 0 0 1 0 0 0 VolTot (m3 ha-1) 1 2 3 x 1 0 4 5 0 0 1 0 0 0 Vol (m3 ha-1) 1 2 3 x 1 0 4 1 0 2 0 3 0 CtreeTot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 Ctree (kg m-2) 1 2 3 x 1 0 4 5 1 0 Cstem (kg m-2) 1 2 3 x 1 0 4 1 2 Cbranch (kg m-2) 1 2 3 x 1 0 4 0 . 5 1 Cleaf (kg m-2) 1 2 3 x 1 0 4 2 4 Croot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 h (m) 1 2 3 x 1 0 4 2 4 LAI (m2 m-2) T i m e 1 2 3 x 1 0 4 5 1 0 1 5 Csoil (kg m-2) T i m e 1 2 3 x 1 0 4 0 . 2 0 . 4 0 . 6 Nsoil (kg m-2) T i m e 0 . 5 1 x 1 0
  • 3
1 0 0 0 2 0 0 0 C B 0 T 1 2 x 1 0
  • 3
1 0 0 0 2 0 0 0 C L 0 T 2 4 x 1 0
  • 3
1 0 0 0 2 0 0 0 C R 0 T P a r a m e t e r m a r g i n a l p r o b a b i l i t y d i s t r i b u t i o n s 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 C S 0 T 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 B E T A 3 0 0 3 5 0 4 0 0 5 0 0 1 0 0 0 C O 2 0 0 . 2 5 0 . 3 0 . 3 5 5 0 0 1 0 0 0 F B 0 . 2 5 0 . 3 0 . 3 5 1 0 0 0 2 0 0 0 F L M A X 0 . 2 5 0 . 3 0 . 3 5 1 0 0 0 2 0 0 0 F S 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 G A M M A 5 1 0 1 5 5 0 0 1 0 0 0 K C A 0 . 3 5 0 . 4 0 . 4 5 5 0 0 1 0 0 0 K C A E X P 2 4 x 1 0
  • 4
5 0 0 1 0 0 0 K D B T 5 x 1 0
  • 4
1 0 0 0 2 0 0 0 K D R T 5 1 0 5 0 0 1 0 0 0 K H 0 . 2 0 . 3 0 . 4 5 0 0 1 0 0 0 K H E X P 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 K N M I N T 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 K N U P T T 0 . 0 2 0 . 0 3 0 . 0 4 5 0 0 1 0 0 0 K T A 1 0 2 0 3 0 5 0 0 1 0 0 0 K T B 0 . 5 1 1 0 0 0 2 0 0 0 K E X T T 4 6 8 1 0 0 0 2 0 0 0 L A I M A X T 1 2 3 x 1 0
  • 3
5 0 0 1 0 0 0 L U E T 0 . 0 1 0 . 0 2 0 . 0 3 1 0 0 0 2 0 0 0 N C L M I N T 0 . 0 2 0 . 0 4 0 . 0 6 5 0 0 1 0 0 0 N C L M A X T 0 . 0 2 0 . 0 3 0 . 0 4 1 0 0 0 2 0 0 0 N C R T 0 . 5 1 1 . 5 x 1 0
  • 3
5 0 0 1 0 0 0 N C W T 6 8 1 0 1 0 0 0 2 0 0 0 S L A T 4 6 8 5 0 0 1 0 0 0 T R A N C O T 1 5 0 2 0 0 2 5 0 1 0 0 0 2 0 0 0 W O O D D E N S 0 . 5 1 5 0 0 1 0 0 0 C L I T T 0 6 8 1 0 5 0 0 1 0 0 0 C S O M F 1 2 3 5 0 0 1 0 0 0 C S O M S 0 . 0 1 0 . 0 2 5 0 0 1 0 0 0 N L I T T 0 0 . 2 0 . 3 0 . 4 1 0 0 0 2 0 0 0 N S O M F 0 . 1 0 . 2 5 0 0 1 0 0 0 N S O M S 1 2 x 1 0
  • 3
1 0 0 0 2 0 0 0 N M I N 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 F L I T T S O M F 0 . 0 5 0 . 1 5 0 0 1 0 0 0 F S O M F S O M S 2 4 x 1 0
  • 3
1 0 0 0 2 0 0 0 K D L I T T 1 2 x 1 0
  • 4
5 0 0 1 0 0 0 K D S O M F 1 2 x 1 0
  • 5
5 0 0 1 0 0 0 K D S O M S 1 2 3 x 1 0 4 5 0 0 1 0 0 0 VolTot (m3 ha-1) 1 2 3 x 1 0 4 5 0 0 1 0 0 0 Vol (m3 ha-1) 1 2 3 x 1 0 4 1 0 2 0 3 0 CtreeTot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 Ctree (kg m-2) 1 2 3 x 1 0 4 5 1 0 Cstem (kg m-2) 1 2 3 x 1 0 4 1 2 Cbranch (kg m-2) 1 2 3 x 1 0 4 0 . 5 1 Cleaf (kg m-2) 1 2 3 x 1 0 4 2 4 Croot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 h (m) 1 2 3 x 1 0 4 2 4 LAI (m2 m-2) T i m e 1 2 3 x 1 0 4 5 1 0 1 5 Csoil (kg m-2) T i m e 1 2 3 x 1 0 4 0 . 2 0 . 4 0 . 6 Nsoil (kg m-2) T i m e

Prior pdf Posterior pdf

Bayesian calibration

Prior pdf

Dodd Wood Dodd Wood

New data

Rheola Rheola

slide-17
SLIDE 17

D a t a B a y e s i a n c a l i b r a t i o n D a t a B a y e s i a n c a l i b r a t i o n

5 x 1
  • 3
2 4 C B T . 0 5 . 0 1 2 4 C L T . 0 5 . 0 1 2 4 C R T P r i o r p a r a m e t e r m a r g i n a l p r o b a b i l i t y d i s t r i b u t i o n s ( b e t a ) 5 x 1
  • 3
2 4 C S T . 4 . 6 . 8 1 2 B E T A 3 3 5 4 1 2 C O 2 . 2 5 . 3 . 3 5 1 2 F B . 2 5 . 3 . 3 5 1 2 F L M A X . 2 5 . 3 . 3 5 1 2 F S . 4 . 6 . 8 1 2 G A M M A 5 1 1 5 1 2 K C A . 3 5 . 4 . 4 5 1 2 K C A E X P 2 4 x 1
  • 4
1 2 K D B T . 5 1 x 1
  • 3
1 2 K D R T 1 2 2 4 K H . 2 . 3 . 4 1 2 K H E X P 1 2 x 1
  • 3
1 2 K N M I N T 1 2 x 1
  • 3
1 2 K N U P T T . 0 2 . 0 3 . 0 4 1 2 K T A 1 2 3 1 2 K T B . 5 1 1 2 K E X T T 4 6 8 2 4 L A I M A X T 1 2 3 x 1
  • 3
1 2 L U E T . 0 1 . 0 2 . 0 3 1 2 N C L M I N T . 0 2 . 0 4 . 0 6 1 2 N C L M A X T . 0 2 . 0 3 . 0 4 1 2 N C R T 1 2 x 1
  • 3
1 2 N C W T 2 4 2 4 S L A T 4 6 8 1 2 T R A N C O T 1 5 2 2 5 1 2 W O O D D E N S . 5 1 1 2 C L I T T 0 6 8 1 1 2 C S O M F 1 2 3 1 2 C S O M S . 0 1 . 0 2 1 2 N L I T T 0 . 2 . 3 . 4 1 2 N S O M F . 1 . 2 1 2 N S O M S 1 2 x 1
  • 3
1 2 N M I N . 4 . 6 . 8 1 2 F L I T T S O M F . 0 5 . 1 2 4 F S O M F S O M S 2 4 x 1
  • 3
1 2 K D L I T T 1 2 x 1
  • 4
1 2 K D S O M F 1 2 x 1
  • 5
1 2 K D S O M S 1 2 3 x 1 0 4 5 0 0 1 0 0 0 VolTot (m3 ha-1) 1 2 3 x 1 4 5 0 0 1 0 0 0 Vol (m3 ha-1) 1 2 3 x 1 0 4 1 0 2 0 3 0 CtreeTot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 Ctree (kg m-2) 1 2 3 x 1 4 5 1 0 Cstem (kg m-2) 1 2 3 x 1 0 4 1 2 Cbranch (kg m-2) 1 2 3 x 1 0 4 0 . 5 1 Cleaf (kg m-2) 1 2 3 x 1 4 2 4 Croot (kg m-2) 1 2 3 x 1 0 4 1 0 2 0 3 0 h (m) 1 2 3 x 1 0 4 2 4 LAI (m2 m-2) T i m e 1 2 3 x 1 4 5 1 0 1 5 Csoil (kg m-2) T i m e 1 2 3 x 1 0 4 0 . 2 0 . 4 0 . 6 Nsoil (kg m-2) T i m e 0 . 5 1 x 1 0
  • 3
1 0 0 0 2 0 0 0 C B 0 T 1 2 x 1 0
  • 3
1 0 0 0 2 0 0 0 C L 0 T 2 4 x 1 0
  • 3
1 0 0 0 2 0 0 0 C R 0 T P a r a m e t e r m a r g i n a l p r o b a b i l i t y d i s t r i b u t i o n s 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 C S 0 T 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 B E T A 3 0 0 3 5 0 4 0 0 5 0 0 1 0 0 0 C O 2 0 0 . 2 5 0 . 3 0 . 3 5 5 0 0 1 0 0 0 F B 0 . 2 5 0 . 3 0 . 3 5 1 0 0 0 2 0 0 0 F L M A X 0 . 2 5 0 . 3 0 . 3 5 1 0 0 0 2 0 0 0 F S 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 G A M M A 5 1 0 1 5 5 0 0 1 0 0 0 K C A 0 . 3 5 0 . 4 0 . 4 5 5 0 0 1 0 0 0 K C A E X P 2 4 x 1 0
  • 4
5 0 0 1 0 0 0 K D B T 5 x 1 0
  • 4
1 0 0 0 2 0 0 0 K D R T 5 1 0 5 0 0 1 0 0 0 K H 0 . 2 0 . 3 0 . 4 5 0 0 1 0 0 0 K H E X P 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 K N M I N T 1 2 x 1 0
  • 3
5 0 0 1 0 0 0 K N U P T T 0 . 0 2 0 . 0 3 0 . 0 4 5 0 0 1 0 0 0 K T A 1 0 2 0 3 0 5 0 0 1 0 0 0 K T B 0 . 5 1 1 0 0 0 2 0 0 0 K E X T T 4 6 8 1 0 0 0 2 0 0 0 L A I M A X T 1 2 3 x 1 0
  • 3
5 0 0 1 0 0 0 L U E T 0 . 0 1 0 . 0 2 0 . 0 3 1 0 0 0 2 0 0 0 N C L M I N T 0 . 0 2 0 . 0 4 0 . 0 6 5 0 0 1 0 0 0 N C L M A X T 0 . 0 2 0 . 0 3 0 . 0 4 1 0 0 0 2 0 0 0 N C R T 0 . 5 1 1 . 5 x 1 0
  • 3
5 0 0 1 0 0 0 N C W T 6 8 1 0 1 0 0 0 2 0 0 0 S L A T 4 6 8 5 0 0 1 0 0 0 T R A N C O T 1 5 0 2 0 0 2 5 0 1 0 0 0 2 0 0 0 W O O D D E N S 0 . 5 1 5 0 0 1 0 0 0 C L I T T 0 6 8 1 0 5 0 0 1 0 0 0 C S O M F 1 2 3 5 0 0 1 0 0 0 C S O M S 0 . 0 1 0 . 0 2 5 0 0 1 0 0 0 N L I T T 0 0 . 2 0 . 3 0 . 4 1 0 0 0 2 0 0 0 N S O M F 0 . 1 0 . 2 5 0 0 1 0 0 0 N S O M S 1 2 x 1 0
  • 3
1 0 0 0 2 0 0 0 N M I N 0 . 4 0 . 6 0 . 8 5 0 0 1 0 0 0 F L I T T S O M F 0 . 0 5 0 . 1 5 0 0 1 0 0 0 F S O M F S O M S 2 4 x 1 0
  • 3
1 0 0 0 2 0 0 0 K D L I T T 1 2 x 1 0
  • 4
5 0 0 1 0 0 0 K D S O M F 1 2 x 1 0
  • 5
5 0 0 1 0 0 0 K D S O M S 1 2 3 x 1 4 5 0 0 1 0 0 VolTot (m3 ha-1) 1 2 3 x 1 4 5 0 1 0 0 Vol (m3 ha-1) 1 2 3 x 1 4 1 2 3 CtreeTot (kg m-2) 1 2 3 x 1 4 1 0 2 0 3 0 Ctree (kg m-2) 1 2 3 x 1 4 5 1 Cstem (kg m-2) 1 2 3 x 1 4 1 2 Cbranch (kg m-2) 1 2 3 x 1 4 0 . 5 1 Cleaf (kg m-2) 1 2 3 x 1 4 2 4 Croot (kg m-2) 1 2 3 x 1 4 1 2 3 h (m) 1 2 3 x 1 4 2 4 LAI (m2 m-2) T i m e 1 2 3 x 1 4 5 1 1 5 Csoil (kg m-2) T i m e 1 2 3 x 1 4 . 2 . 4 . 6 Nsoil (kg m-2) T i m e

New data Bayesian calibration

Prior pdf Posterior pdf Prior pdf

Dodd Wood Dodd Wood

0.5 1 x 10
  • 3
500 CB0T 1 2 x 10
  • 3
500 CL0T 5 x 10
  • 3
500 1000 CR0T 1 2 x 10
  • 3
500 CS0T 0.4 0.6 0.8 500 BETA 300 350 400 500 1000 CO20 0.25 0.3 0.35 500 FB 0.25 0.3 0.35 500 1000 FLMAX 0.25 0.3 0.35 500 FS 0.4 0.6 0.8 500 GAMMA 5 10 15 500 KCA 0.350.40.45 500 1000 KCAEXP 2 4 x 10
  • 4
500 KDBT 2 4 x 10
  • 4
500 KDRT 5 10 500 KH 0.2 0.3 0.4 200 400 KHEXP 1 2 x 10
  • 3
500 KNMINT 1 2 x 10
  • 3
500 1000 KNUPTT 0.02 0.03 0.04 500 KTA 10 20 30 500 KTB 0.5 1 500 KEXTT 4 6 8 500 LAIMAXT 1 2 3 x 10
  • 3
500 1000 LUET 0.01 0.02 0.03 500 1000 NCLMINT 0.02 0.04 0.06 500 NCLMAXT 0.02 0.025 0.03 200 400 NCRT 0.5 1 1.5 x 10
  • 3
500 NCWT 6 8 10 500 SLAT 4 6 8 500 TRANCOT 150 200 250 500 WOODDENS 0.5 1 500 CLITT0 6 8 10 200 400 CSOMF0 1 2 3 200 400 CSOMS0 0.01 0.02 500 NLITT0 0.2 0.3 0.4 500 NSOMF0 0.1 0.2 500 NSOMS0 1 2 x 10
  • 3
500 1000 NMIN0 0.4 0.6 0.8 500 FLITTSOMF 0.05 0.1 500 1000 FSOMFSOMS 2 4 x 10
  • 3
500 1000 KDLITT 1 2 x 10
  • 4
500 KDSOMF 1 2 x 10
  • 5
500 1000 KDSOMS

Rheola Rheola

0.5 1 1.5 2 2.5 x 10 4 200 400 600 800 VolTot 0.5 1 1.5 2 2.5 x 10 4 200 400 Vol

p p

0.5 1 1.5 2 2.5 x 10 4 10 20 30 CtreeTot 0.5 1 1.5 2 2.5 x 10 4 5 10 15 Ctree 0.5 1 1.5 2 2.5 x 10 4 2 4 6 8 Cstem 0.5 1 1.5 2 2.5 x 10 4 0.5 1 1.5 Cbranch 0.5 1 1.5 2 2.5 x 10 4 0.2 0.4 0.6 Cleaf 0.5 1 1.5 2 2.5 x 10 4 2 4 Croot 0.5 1 1.5 2 2.5 x 10 4 5 10 15 20 h 0.5 1 1.5 2 2.5 x 10 4 1 2 LAI

Time

0.5 1 1.5 2 2.5 x 10 4 8 10 12 14 Csoil

Time

0.5 1 1.5 2 2.5 x 10 4 0.3 0.4 0.5 Nsoil

Time

slide-18
SLIDE 18

Input data: atmosphere (UKCIP & NEGTAP)

5.91-6.66 6.66-7.4 7.4-8.15 8.15-8.9 8.9-9.64 9.64-10.4 10.4-11.1 11.1-11.9 11.9-12.6 12.6-13.4 Temperature (C)

Mean temperature for 1920-

  • 2000. Data source: UKCIP

0.621-1.4 1.4-2.18 2.18-2.96 2.96-3.74 3.74-4.51 4.51-5.29 5.29-6.07 6.07-6.85 6.85-7.63 7.63-8.41 Rain (mm d-1)

Mean precipitation for 1920-

  • 2000. Data source: UKCIP

2.87-5.83 5.83-8.79 8.79-11.7 11.7-14.7 14.7-17.7 17.7-20.6 20.6-23.6 23.6-26.5 26.5-29.5 29.5-32.5 N-deposition (kg/ha/y)

Atmospheric N-deposition in

  • 2004. Data source: NEGTAP /

R.I. Smith

slide-19
SLIDE 19

Input data: soil (IGBP-DIS)

7-14 14-21 21-28 28-35 35-42 42-49 49-56 56-63 63-70 70-77 C(soil) (kg/m2)

Total carbon in top 100 cm

  • soil. Data source: IGBP-DIS

63-82.7 82.7-102 102-122 122-142 142-161 161-181 181-201 201-220 220-240 240-260 PAWC (mm)

Maximum plant available water in top 100 cm soil. Data source: IGBP-DIS

0.797-1.39 1.39-1.97 1.97-2.56 2.56-3.15 3.15-3.74 3.74-4.33 4.33-4.92 4.92-5.51 5.51-6.09 6.09-6.68 N(soil) (kg/m2)

Total nitrogen in top 100 cm

  • soil. Data source: IGBP-DIS
slide-20
SLIDE 20

Model results, using env. input data & posterior parameter pdf’s (Bayesian calibr.)

  • 0.255--0.166
  • 0.166--0.0767
  • 0.0767-0.0127

0.0127-0.102 0.102-0.191 0.191-0.281 0.281-0.37 0.37-0.46 0.46-0.549 0.549-0.638 C-seq. (kg/m2/y) (1920-2000)

Simulated average annual C- sequestration (in soil, living trees and wood products) for 1920-

  • 2000. Results from model

BASFOR

4.68e-006-0.00416 0.00416-0.00831 0.00831-0.0125 0.0125-0.0166 0.0166-0.0208 0.0208-0.0249 0.0249-0.0291 0.0291-0.0332 0.0332-0.0374 0.0374-0.0415 SD(C-seq.) (kg/m2/y)

Uncertainty (standard deviation) in simulated average annual C- sequestration (in soil, living trees and wood products) for 1920-2000. Results from model BASFOR

slide-21
SLIDE 21

Effect of env. change on C-sequestration

  • 0.0357--0.014
  • 0.014-0.00775

0.00775-0.0295 0.0295-0.0513 0.0513-0.073 0.073-0.0948 0.0948-0.117 0.117-0.138 0.138-0.16 0.16-0.182 Change C-seq. (kg/m2/y)

Simulated change in average annual C-sequestration (in soil, living trees and wood products) from 1920-2000 to 2000-2080. Results from model BASFOR Increase in [CO2] Change in N- deposition Climate change

slide-22
SLIDE 22

Environmental factor analysis

(Dodd Wood)

Impact of environmental change Ecosystem variable Dodd Wood value Effect of temperature (per °C) Effect of [CO2] (per 100 ppm) Effect of N- deposition (per 10 kg N ha-1 y-1) Yield class (m3 ha-1 y-1) 7.91 ± 1.11 0.18 ± 0.05 1.32 ± 0.38 0.74 ± 0.26 C-sequestration (t C ha-1 y-1) 3.99 ± 0.64 0.10 ± 0.03 0.76 ± 0.21 0.41 ± 0.14 C-sequestration, soil (t C ha-1 y-1) 1.58 ± 0.31 0.05 ± 0.01 0.36 ± 0.10 0.18 ± 0.07 C-sequestration, trees and products (t C ha-1 y-1) 2.41 ± 0.34 0.05 ± 0.02 0.40 ± 0.12 0.23 ± 0.07

Conclusions from factor analysis:

  • Uncertainties (standard deviations) of factor effects: 20-40%

`But …:

  • Actual uncertainty is larger, as only parameter uncertainty was

quantified (not uncertainty about inputs or model structure)

  • Sensitivities are nonlinear, so site-specific, and thus need to be

calculated UK-wide

slide-23
SLIDE 23

Possible use of yield class modifiers with uncertainties (Bayesian): input to CFLOW

Impact of environmental change Ecosystem variable Dodd Wood value Effect of temperature (per °C) Effect of [CO2] (per 100 ppm) Effect of N- deposition (per 10 kg N ha-1 y-1) Yield class (m3 ha-1 y-1) 7.91 ± 1.11 0.18 ± 0.05 1.32 ± 0.38 0.74 ± 0.26 C-sequestration (t C ha-1 y-1) 3.99 ± 0.64 0.10 ± 0.03 0.76 ± 0.21 0.41 ± 0.14 C-sequestration, soil (t C ha-1 y-1) 1.58 ± 0.31 0.05 ± 0.01 0.36 ± 0.10 0.18 ± 0.07 C-sequestration, trees and products (t C ha-1 y-1) 2.41 ± 0.34 0.05 ± 0.02 0.40 ± 0.12 0.23 ± 0.07

Woody biomass Non-woody biomass Woody litter Non-woody litter Soil organic matter Wood products thinning and harvesting Transfer of residues to soil Natural mortality Thinnings Harvest debris

Yield tables & expansion factors

Woody biomass Non-woody biomass Woody litter Non-woody litter Soil organic matter Wood products thinning and harvesting Transfer of residues to soil Natural mortality Thinnings Harvest debris

Yield tables & expansion factors Environmental response modifiers

slide-24
SLIDE 24

Conclusions

1. Method of sequential data assimilation & uncertainty quantification by Bayesian calibration works well 2. Tree data for calibration and environmental data

  • f model drivers still limited. Key issue: soil

nitrogen 3. Environmental factor analysis for Dodd Wood showed importance of elevated CO2, but …

  • needs to be repeated after better tree & soil information

has become available for further Bayesian calibration

  • needs to be repeated UK-wide because the same change

in an environmental factor has different effects on different sites

4. Because of the above, methodology not yet ripe for the official inventory

slide-25
SLIDE 25

Outlook

Bayesian methods used in other European collaborations:

NitroEurope: European Union IP aimed at quantifying nitrogenous GHG budget for Europe

  • Protocol for Uncertainty Quantification & Uncertainty Analysis
  • At least 2 models for each vegetation type (4 for forests)
  • Bayesian methods also used to assess structural uncertainty (“how

plausible are the different models?”

CarboEurope: (just starting, for crop models) Forest Focus (forests & env. change) WINSUR (grasslands & env. change)