Contributor:

Rothamsted Research3rd Annual Meeting Month 40 of 42

November 2008

TSEC-BIOSYSTheme 2 - Topic 2.2

Modelling biomass supply

Modelling bioenergy crops -key objectives

Purpose I is to assess– Production potential of bioenergy (BE) at the sub-regional

scale, – Trade-offs of BE vs. Food within land use change (LUC), – Cost-based supply as an option within the UK energy mix, and– Environmental implications, like GHG-balance and hydrology

Purpose II is to– Describe, quantify and predict system behaviour – Underpin processes in aide of crop selection/breeding (G x E)– Identify the most important genotypic traits and – Locate crucial control points of yield formation

Task within TSEC-BIOSYS

Theme 2: Evolution of UK biomass supply• Topic 2.2: Bioenergy Models resources

Biofuel from arable crops – models @ RRES Winter wheat, sugar beet,

Oilseed rape, maize

Biomass from grasses, mainly Miscanthus Empirical model for Miscanthus (& switchgrass)

Maps of yield under current climate

Process model for Miscanthus is available; parameterized, calibrated and evaluated;

Ready to be used for predictive purposes

0

5

10

15

20

Arthur

Rick

wood

Boxwor

th

Bridge

ts

Brooms B

arn

TG

Buckf

ast A

bbey

Gleadth

orpe

High M

owth

orpe

Rosem

aund

Rosewar

ne

Rotham

sted

408

Rotham

sted

480

Rotham

sted

TGSCRI

Wobu

rn m

ain T

G

Wobu

rn m

icro

TG

Yie

ld (

dry

mat

ter

- t

ha-1

)

0

2

4

6

8

10

12

14

16

18

20

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Yie

ld (

dry

ma

tte

r -

t h

a-1 )

RES 408

RES 480

Empirical yield model for MiscanthusRichter, G. M. et al. (2008) Soil Use and Management 24 (3), 235

Application of empirical yield maps

Aide to

Producers &

LUC planners

BE Allocation

Trade-offs

Economic of BE Supply & Demand

Assess environmental impact/benefit

Richter et al., Soil Use Manage 24, 235 (2008)

GHG H20

Land use trade-offs - Methods

• Incorporated a range of constraints on energy crops– environmental, physical– agricultural, agronomic – socio-economic

• Accounted for currently grown food crops

• Used Miscanthus yield map for England

Lovett, A. A. et al., BioEnergy Research (u. rev.)

Land use trade-offs – Results

• Regional contrasts occur in the importance of different constraints

• Between 80 and 20% of are below an economic threshold of 9 t/ha

• Areas with highest yields co-locate with important food producing areas

Lovett, A. A. et al., BioEnergy Research (u.rev.)

Supply & Demand Modelling

• Majority of land would yield between 10 - 14 t odm/ha/yr

• Cost map gives annual cost of 20 to 60 £/t odm

• Switch from yield to cost optimal crop affects only a small fraction of land

• Preference map shows 4.4 Mha of Miscanthus and 6 Mha of SRC

Conclusions for integration (Theme 4) - based on working paper between IC, UoSo, RRes, FR

• Yield maps are available for Miscanthus, willow and poplar

• Overlay of yield maps implied some exclusion criteria (slope > 15%, organic soils)

• Yield and cost advantage maps have been created

• Potential availability of 10 Mha preferably used for willow and Miscanthus (ratio 6:4)

• Suitability and constraint maps reduced area to about 3 Mha (preference of food production given to high grade land) – cooperation with UEA (Lovett)

• Simulations of biomass crop allocation based on opportunity costs confirmed expansion of lower grade land being used under higher BE-demand

• Paper is based on empirical models describing current (past) yields only – future scenarios (2050) are excluded up to now

• Future scenarios must be based on process-based models

Modelling Purpose II

Describe, quantify and predict system behaviour at process-level

Underpin the processes in aide of crop selection and breeding (G x E interaction)

Identify the most important genotypic traits that can be easily quantified and

Locate crucial control points of yield formation

Experimental basis for Process Model

• Long-term, highly resolved data at Rothamsted– Light interception (LAI)– Dry matter – Leaf senescence, loss

(litter)

• Morphological data – Stem number, height &

diameter– Leaf length, width

• Growth dynamics of belowground biomass (rhizomes)

0

5

10

15

20

25

01/05/96 26/06/96 21/08/96 16/10/96 11/12/96 05/02/97 02/04/97

Dry

ma

tte

r [

t h

a-1 ]

Total

Stems

Leaves

Dead Leaves

0

2

4

6

8

10

12

14

16

18

20

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Yie

ld (

dry

ma

tte

r -

t h

a-1

)

RES 408

RES 480

Christian, D. G. et al., Biomass & Bioenergy 30, 125 (2006)

Christian, D. G., Riche, A. B., Yates, N. E., Industrial Crops and Products 28, 109 (2008)

kfrost

Carbohydrates

Reserves10-20%

StemsDensity (n),

Ht, Wt

RhizomesRGR(T), SRWT,

[RhDR(t)]

fT(A)

rad, P, T,..

θfc, θpw, depth, ...

crf

fsht

cL/P

Source Formation Sink Formation

MorphologyWD(L), SLA,

nV, nGMaxHt, SSW(d)

PER

Photo-synthesis

Flowers

fw Phenology

Phyllochron, nLTb, TΣ(e, x, a),

cv2g

Energy Balance

Water Balance

Ta

Leaves

LAI

Inter-ception

kext

Roots

PhysiologyAsat, φ

rs, ksen,,fW, fT

rdr, halflife

ksen

A sink-source interaction model

Tillering

Sensitivity of model parameters

Δyield/Δparameter

Parameter sensitivity for Miscanthus

• Grouped according to

– Initial establishment

– Phenology

– Physiology

– Morphology

0

100

200

300

400

500

0 500 1000 1500 2000 2500 3000

μ_Change

σ_

Ch

an

ge

physio- pheno-

morpho- initial

Model evaluation – Sensitivity Analysis

cv2g

Toptv2g

TΣ(x)

φ

kext

Tb(A)

Asat

Tn(A)

Tx(A)

WDL

cL/P

fsht

SLAx

cSSW

Tb(sht)DMrhz

Sink – Source Balance

0

10

20

30

40

50

60

70

80

1 91 181 271 361 451 541 631

Day after start of simulation (1/1/94)

Car

bo

hyd

rate

S&

D [

g m

-2 d

-1 ]

ShootGrowthPotn AGGrowthSourceLimited

Leaf DM & GLAI dynamics

0

1

2

3

4

5

6

01/01/94 01/01/95 01/01/96 31/12/96 01/01/98 01/01/99

Le

af

dry

ma

tte

r [

t h

a-1 ]

0

1

2

3

4

5

6

7

Jan 94 May 94 Sep 94 Jan 95 May 95 Sep 95

GL

AI [

m2 m

-2 ]

Model evaluation – shoots

0

50

100

150

200

01/01/94 01/01/95 01/01/96 31/12/96 01/01/98 01/01/99

Sh

oo

t n

um

ber

0

50

100

150

200

250

300

01/01/94 01/01/95 01/01/96 31/12/96 01/01/98 01/01/99

Hei

gh

t [

cm ]

• Shoot ≡ Generative Tiller– Initially fixed No. of VegTiller

– cv2g is an important factor

– Tiller dynamics linked to height

• Height dynamics– Increases with GY

– PER function of T & CHORes

– Partitioning PER using cL/P

• Stem weight evaluation– Discrepancy is consequence of

height estimate, tiller dynamics

– Loss of stem weight at harvest is due to stubble

0

5

10

15

20

25

01/01/94 01/01/95 01/01/96 31/12/96 01/01/98 01/01/99

Ste

m d

ry m

atte

r [

t h

a-1 ] Harvested

Leaf area dynamics and water stress

0

1

2

3

4

5

6

7

8

9

10

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

LA

I

-1.5

-1

-0.5

0

0.5

1

Wat

er s

tres

s fa

cto

r, k

w

LAI [-] k_w

Yield prediction over 14 years

9495

96

9798

9900

01

02

03

0405

06

07y = 1.03x

6

10

14

18

22

6 10 14 18 22

Observed yield [ t ha-1 ]

Sim

ula

ted

yie

ld [

t h

a-1 ]

0

5

10

15

20

25

Jan94

Jan95

Jan96

Jan97

Jan98

Jan99

Jan00

Jan01

Jan02

Jan03

Jan04

Jan05

Jan06

Jan07

Jan08

Ste

m d

ry m

att

er

[ t

ha-1

]

Harvested

Conclusions for Process-based Model

• A generic grass model was successfully adopted to simulate dry matter production of Miscanthus x giganteus– Identified important morphological traits– Calibrated & evaluated for one site, one variety– Ranked parameter using OAT sensitivity analysis– Exploring sink-source balance, tillering dynamics

• Future applications of this model are needed– For different species & varieties to identify optimal

grass ideotypes – In different environments (G x E interaction)

Thank you for questions !

0

0.2

0.4

0.6

0.8

1

1.2

0 10 20 30 40

T air [ oC ]

f so(

Tai

r)

Naidu rel(Asat)

Farage rel(Asat)

Naidu rel(φ)

Farage rel(φ)

Asat, φ = f(Ta)

Naidu, S. L. et al., Plant Physiology 132 (3), 1688 (2003).Farage, P. K., Blowers, D., Long, S. P., and Baker, N. R., Plant Cell and Environment 29 (4), 720 (2006).

T-scale function, photosynthesis

Water stress function

0.0

0.2

0.4

0.6

0.8

1.0

0 0.2 0.4 0.6 0.8 1

Relative soil water content

Ra

te r

ed

uc

tio

n

ws-factor = 12

ws-factor = 6

kws = 2 / ( 1 + exp (-Ws-factor * relSWC))

early response

late response

Sinclair, T. R., Field Crops Res. 15, 125 (1986).

Richter, G. M., Jaggard, K. W., Mitchell, R. A. C., Agric For Meteorol 109, 13 (2001).

Morphological Parameters – Leaf

• Leaf extension rates (L/PER)– A priori parameters from Clifton-

Brown & Jones (1997)– Simplified either as linear model

or Arrhenius function (Q10)– Compared to in situ

measurements

• Specific area (SLA) – Unchanged principle from LinGra

giving a min-max range– Range adjusted to observed SLA

• Dynamic components– Number of leaves growing

simultaneously (nL 2.7 → > 3)– Senescence rates (age, shading,

drought) determine tiller density

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 15 20 25

Temperature

PE

R [

mm

hr-1

]

Measured (C-B&J)

3-polyn (C-B&J)

Linear

Arrhenius

Morphological parameters – Shoot/Stem

• Stem extension rate– Related to leaf extension rate

e.g. le ~ 0.83 ±0.07; (Clifton-Brown & Jones 1997)

• Shoot density [ m-2 ]– Initially 100 to 140 m-2

(Danalatos et al. 2007; Bullard et al. 1995)

– 50 to 80 m-2 at equilibrium (Clifton-Brown & Jones 1997; Danalatos et al. 2007)

• Specific stem weight– 10 to 11 g m-2

(acc. to Danalatos et al., 2007)

– Changes with height and plant age (unpublished)

Maximum specific stem weight

0

5

10

15

20

25

0 0.5 1 1.5 2 2.5 3

Stem height [ m ]

Sp

ec

ific

ste

m w

eig

ht

[ g

/m ]

Sensitivity Analysis

• Morris-method varies parameters as one-at-a-time at discrete levels (4 to 8)

• Parameters given as mean ± % variation, randomly generated within 5-95%

• “change” is defined as Δyield/Δparameter

• μ / μ* are means of distribution of the “global” parameter effect

• “σ” is an estimate of second- and higher order effects of parameter (interactions with other factors, non-linearity)

• Simultaneous display of μ* and σ allows to check for non-monotonic models (negative elements in distribution)

ReferencesMorris (1991) as described in Saltelli et al. (2004)*

Morris M.D. Technometrics 33(2) 161-174; Saltelli A., et al.. Sensitivity analysis in practice. WILEY