Sirius wheat simulation model: development and applications Mikhail A. Semenov Rothamsted Research, UK IT in Agriculture & Rural Development, Debrecen,

Sirius wheat simulation model: development and applications

Mikhail A. SemenovRothamsted Research, UK

IT in Agriculture & Rural Development, Debrecen, 2006

Sirius: wheat simulation model

Initially developed by Peter Jamieson from Crop & Food Research, NZ; since 1992 development in collaboration with Mikhail Semenov at Rothamsted Research, UK

Intensively tested in different environments and used in many countries (www.rothamsted.bbsrc.ac.uk/mas-models/sirius.php)

Sirius is a part of the GCTE International Wheat Network

Sirius: wheat simulation model

Sirius

Inputs Outputs

Daily Weather

Soil

Management

Cultivar

Grain yield

Grain quality

N leaching

Water and N uptake

Sirius: a process-based model

Radiation use efficiency and biomass accumulation

Phenological development Canopy model Nitrogen uptake and redistribution Evapotranspiration and water limitation Soil model

Modelling growth: Radiation Use Efficiency

Radiation Use Efficiency

0

1

2

3

4

5

6

7

8

9

10

0 1 2 3 4

Light intersepted, Mj / m2

Bio

ma

ss

g /

m2

Beer's Law

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10

Leaf Area Index

Lig

ht

Inte

rse

pte

d, %

Biomass = RUE*R

R intercepted radiation

P = 1-exp(-k LAI)

P proportion of light intercepted

Modelling canopy

Phenology is used to predict emergence times of individual leaves

Deal with leaf “layers” avoid consideration of tillers avoid adding extra parameters for calibration

Define genetic potential growth

Modelling Canopy: Leaf Area Index

Leaf Area Index

0

1

2

3

4

5

6

7

8

9

10

0 5 10 15 20 25 30 35 40 45 50

time

LA

I

anth

esi

smax LAI Sirius grows a canopy Sirius grows a canopy

(LAI) according to simple (LAI) according to simple rules involving rules involving temperature, water and temperature, water and N supplyN supply

Modelling phenology

Pre-emergence and after anthesis calculations are based on thermal time

Calculation of anthesis is based on the final leaf number and the value of phyllochron

Calculation of the final leaf number includes vernalization and daylength responses

Em

erg

enc

e

Anthesis

Matu

rit

y

N Limitation

Leaf Area Index

0

1

2

3

4

5

6

7

8

9

10

0 5 10 15 20 25 30 35 40 45 50

time

LA

I

anth

esi

s

max LAI

Green area contains 1.5 g N/m2 ; “non-green” biomass can store 1% labile N

Daily N-demand is set by the increment of new GA and biomass

Unsatisfied demand limits the GA increment and/or causes N release through premature GA senescence

Calibration and validation

Calibration – measuring (direct) or fitting (indirect) model parameters to observed data

Validation – using independent (not used during calibration) observed data for testing model skills

Validation of Sirius: N experiments

0

1

2

3

4

5

6

7

20-Jan 19-Feb 21-Mar 20-Apr 20-May 19-Jun

GAI

Low N High N Obs LoN Obs HiN

FACE, Maricopa, 1996/97

Sirius: soil, evapotranspiration & water limitation

Soil model is based on modified SLIM (UK) and DAISY (DENMARK) models

ET is calculated as the sum of transpiration and soil evaporation after Ritchie (1972). The upper limit is given by the Penman potential ET rate or the Priestley&Taylor equation

Water stress factor reduces leaf expansion and accelerate leaf senescence.

Validation: water-limited grain yield

0

2

4

6

8

10

12

0 2 4 6 8 10 12

Measured yield (t/ha)

Sim

ula

ted

yie

ld (

t/h

a)

Y = X

Canterbury, NZ

Rothamsted, UK

Maricopa, H2O

Free-Air CO2 Enrichment Project (FACE)

RUE = f(CO2)

USDA-ARS U.S. Water Conservation LaboratoryUSDA-ARS U.S. Water Conservation Laboratory, Maricopa, USA, Maricopa, USA

Model complexity

Model complexity is related to a number of model parameters and model equations

Hierarchy of complexity: Meta-model (Brooks et al, 2001); Sirius (Jamieson et al., 1998), AFRCWHEAT

(Porter 1993), CERES-Wheat (Ritchie and Otter, 1985);

Ecosys (Grant, 1998).

Simplifying model

Mimic model output by non-linear regression

0

1

2

3

4 0

1

2

3

4

-1

0

1

0

1

2

3

4

Model response surface

0

1

2

3

40

1

2

3

4

-1-0.5

0

0.5

1

0

1

2

3

4

Fitted approximation

Simplifying model

Simplify a model by analysing model structure, model processes and its interactions

0

1

2

3

40

1

2

3

4

-1-0.5

0

0.5

1

0

1

2

3

4

0

1

2

3

4 0

1

2

3

4

-1

0

1

0

1

2

3

4

Comparison between Meta-model and SiriusRothamsted, UK, 1960-1990 (50% precipitation)

4

5

6

7

8

9

10

4 5 6 7 8 9 10

Sirius

Meta

Simulation results, Andalucian region, Spain, 1988-1999

3

4

5

6

7

8

9

3 4 5 6 7 8 9

observed

sim

ulat

ed

Meta Sirius

Obs Meta Sirius

Obs 1.00

Meta 0.92 1.00

Sirius 0.63 0.74 1.00

ApplicationPrediction of grain yield in real time

Observedweather

Management

Soil

Sirius

Generatedweather

0

0.05

0.1

0.15

0.2

0.25

0.3

0 4 8 12 16 20

ApplicationPrediction of grain yield in real time

Observedweather

Management

Soil

Sirius

Generatedweather

0

0.05

0.1

0.15

0.2

0.25

0.3

0 4 8 12 16 20

Weather uncertainty in real-time predictions

0

100

200

300

400

500

600

700

800

900

1000

Sep Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep

Accumulated Rainfall (mm)

0

100

200

300

400

500

600

700

800

900

1000

Sep Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep

Accumulated Rainfall (mm)

0

100

200

300

400

500

600

700

800

900

1000

Sep Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep

Accumulated Rainfall (mm)

Accumulated rainfall, mm

Yield prediction using mixture of observed and generated weather at Rothamsted, 1997

4000

4500

5000

5500

6000

6500

7000

7500

8000

8500

9000

0 50 100 150 200 250 300 350

No. observed days

Grain Yield (kg/ha)

Lead-time for predicting wheat growthat Rothamsted

0

0. 1

0. 2

0. 3

0. 4

0. 5

0. 6

0. 7

0. 8

0. 9

1

0306090120150180210240270300330360

No. days bef ore matur i ty

Probability of prediction

Fi nLN

Ant hD

Mat D

Bi omass

Yi el d

Lead-time for predicting grain yield in diverse climates

Yi el d

0

0. 1

0. 2

0. 3

0. 4

0. 5

0. 6

0. 7

0. 8

0. 9

1

0306090120150180210240270300330360

No. days bef or e mat ur i t y

Probability of prediction

Tylstrup Debrecen Toulouse Lincoln Munich RothamstedTylstrup Debrecen Toulouse Lincoln Munich Rothamsted

Grain yield can be predictedwith 0.9 probability:in Toulouse 40 days and in Tylstrup 65 days before maturity

Publications

Jamieson PD, Semenov MA, Brooking IR & Francis GS (1998) Sirius: a mechanistic model of wheat response to environmental variation. Europ. J. Agronomy, 8:161-179

Jamieson PD & Semenov MA (2000) Modelling nitrogen uptake and redistribution in wheat. Field Crops Research, 68: 21-29.

Brooks RJ, Semenov MA & Jamieson PD (2001) Simplifying Sirius: sensitivity analysis and development of a meta-model for yield prediction Europ. J. Agronomy 14:43-60

Lawless C, Semenov MA & Jamieson PD (2005) A wheat canopy model linking leaf area and phenology Europ. J. Agronomy, 22:19-32

Lawless C & Semenov MA (2006) Assessing lead-time for predicting wheat growth using a crop simulation model Agric Forest Meteorology 135:302-313

WWW: www.rothamsted.bbsrc.ac.uk/mas-model/sirius.php