Current developments in plant modelling at Wageningen-UR

Wageningen’s domain: �Food and Living Environment

• Sustainable production �and food processing
• Animal feed and �biobased products
• International food �chains and networks
• Food security and food�health aspects

• Nature and landscape
• Land use
• Water, sea and natural �resource management
• Biodiversity

• Food and Living environment
• Lifestyle
• Perceptions
• Governance
• Market and chains
• Social innovations

Wageningen University & Wageningen Research

Partners

Wageningen University

• 13,153 BSc/MSc-students from > 100 countries
• 2,303 PhD candidates
• 3,767 faculty and staff (3,277 fte)
• Revenue in 2021: € 431 million

​

Wageningen Research

Partners

​

​

​

​

• 3,471 staff (3,143 fte)
• Revenue in 2021: 373 million

Wageningen University

Wageningen Research

Organigram

About myself

• Working at Wageningen-UR since 1997
• Senior scientist on agrometeorology and crop modelling
• Maintainer and developer of the WOFOST cropping system model
• Main lines of work
• Operational crop monitoring and yield forecasting (Europe, China, Morocco, Russia)
• Agromet systems for supporting smallholder farmers (advisory, finance, weather)
• Education, teaching and capacity building

​

7

Crop models developed in Wageningen

• Quantitative analysis of development, growth and production of annual field crops as well as perennials (cocoa, palm oil, banana)
• Simulate for different agro-ecological production levels
• Characteristics:
• Mechanistic
• dynamic
• Different approaches ranging from summary models to complex biochemistry/physics
• Public models, open source implementations available

​

8

Pedigree of models of “School of De Wit”

1965

1970

1975

1980

1985

1990

1995

2000

Photosynthesis of leaf canopies

(de Wit 1965)

ELCROS

(de Wit et al. 1970)

BACROS

(de Wit et al. 1978)

PAPRAN

(Seligman & van Keulen 1981)

ARID CROP

(van Keulen 1975)

​

ARID CROP

(SAHEL)

(van Keulen et al. 1986)

PHOTON

(de Wit et al. 1978)

MICROWEATHER

(Goudriaan 1977)

ORYZA

(Kropff et al. 1995)

​

​

ORYZA2000

(Bouman et al. 2001)

​

ORYZA v3

(Bouman et al. 2013

WOFOST 7

(van Diepen et al. 1988)

(van Keulen & Wolf 1986)

​

​

​

​

​

​

​

WOFOST 7.2

(De Wit et al 2019)

​

WOFOST 8

(Berghuis et al. In prep.)

SUCROS

(van Keulen et al. 1982)

​

​

SUCROS87

(van Laar et al. 1992)

​

​

SUCROS1

(Goudriaan & van Laar 1994)

SUCROS2

(van Laar et al. 1997)

MACROS

(Penning de Vries et al. 1989)

INTERCOM

(Kropff & van Laar 1993)

GECROS

(Yin and Struik 2017)

SWHEAT

(van Keulen & Seligman 1987)

2022

​

​

​

​

​

​

LINTUL

(Adiele et al 2022)

Functional-Structural plant models (FSPM)

• Explicitly represent plant 3D structure
• Deal with the complex 3-D structure of plants and to simulate growth and development occurring at spatio-temporal scales:
• From cell arrangement to forest structure
• seconds to decades and many plant generations
• Currently being implemented in the WUR “Virtual Tomato” Digital Twin

​

​

10

Functional-Structural plant models (FSPM)

11

Functional-Structural plant models (FSPM)

12

Use plant phenotyping for FSPM

13

Institute for Advanced Studies for Photosynthetic Efficiency (IASPE)

14

Categories of mechanistic crop models

15

WOFOST/GECROS model

LINTUL model

WOFOST – Agro-ecological production levels

16

WOFOST 7.2

WOFOST 8

Phenological development in wheat

17

Source Hyles et al. 2020. https://doi.org/10.1038/s41437-020-0320-1

BBCH scale for phenology

18

Understanding phenology is important

• Temperature, radiation and CO₂ drive crop growth
• Phenological development steers crop growth

Thus:

• Phenological development is a parallel controlling process

​

Correctly simulating phenological development is the first priority of crop models and model calibration always start with phenology!

19

WOFOST modelled processes

20

Source: Hammer et al. 2019. DOI: 10.1093/insilicoplants/diz010

• WOFOST modelling domain is limited to above-ground processes + roots
• Exchanges between soil-plant are related to water, carbon and nutrient fluxes

​

• Modelling soil processes can be done with models of different complexity:
• SWAP, tipping bucket, VIC, Noah-LSM

​

WOFOST process flow diagram

21

Limitations of WOFOST 7.2

Inherent to simulation technique:

• Multi-parameter model, difficult to calibrate and validate
• Sensitive for initial state of soil and crop

​

Chosen generalizations:

• Growth is source driven, not sink limited
• No crop architecture (plant height, branching, individual leaves, number of grains)
• Homogeneous canopy, no effect of rows, N-S orientation)
• No translocation of assimilates between organs. (Will be included in v8)

Limited knowledge of crop response relations:

• Empirical relations, e.g. partitioning, mechanism not well understood
• Best for near optimum growth conditions
• Severe stress difficult to quantify
• Recovery mechanisms unknown

​

22

WOFOST implementations

23

WOFOST - Documentation

24

How accurate is WOFOST?

• Difficult to quantify, as it depends on quality of calibration and input data
• WOFOST often over-estimates yield compared to farmer’s fields (e.g. agro-ecological production levels)
• Getting interannual variability is often more important then the absolute yield value
• Post-processing techniques can be used to correct for bias and trends in yield data

​

25

Potential production for Potato: accuracy

26

Actual production for sugar beet

27

WOFOST model output (biomass) was used to predict actual sugar yield at the level of a factory sourcing area

​

Statistical techniques were used to predict sugar yield from WOFOST results

WOFOST applications: MARS

• Backbone of MARS Crop Yield Forecasting System (MCYFS)
• Provides timely, independent and scientifically relevant:
• Information on crop development and growth
• Information on short-term effects of meteo �events on crops
• Seasonal yield forecasts of key crops (cereal, oil seed crops, protein crops, sugar beet, potatoes, pastures, rice) at EU and member state level
• Developed by WUR in 90s, acknowledged by EU parliament, operationally run by WUR since 2000

​

28

https://ec.europa.eu/jrc/en/research-topic/crop-yield-forecasting

WOFOST applications: MARS

29

WOFOST applications: MARS

30

WOFOST applications: GYGA

• GYGA explores possibilities for extra production and intensification under full irrigation & rainfed conditions for crops across the globe
• GYGA is an international initiative with:
• National agronomists with local knowledge
• Standard protocol for assessing yield potential (Yp),�water-limited yield potential (Yw) and yield gaps (Yg)
• Based on best available local data
• Robust crop simulation models (WOFOST, �CERES-MAIZE embedded in DSSAT 4.0 etc.)
• Bottom-up approach to upscale results from location�to region and country

31

www.yieldgap.org

WOFOST applications: GYGA

32

yield gap of rainfed maize

WOFOST in WaterWijzer Landbouw

33

Further developments

• Maintain, support and document current model implementations
• New developments:
• WOFOST 8.1 will have improved implementation of N limited growth through the leaf N concentration
• Data assimilation strategies for taking into account spatio-temporal variability (EnKF, 2DVAR)
• Better integration with soil processes including water, C/N, temperature (also in SWAP/WOFOST)

​

​

34

Conclusions

• Range of plant modelling approaches developed/used in Wageningen
• Traditional crop models, FSPM models, strip cropping
• Well supported by experimental evidence (field trials, NPEC)
• Efforts sometimes fragmented across groups
• Additional focus required on:
• Soil-Plant interactions
• Photosynthesis
• Nutrients, extreme events

35

Additional resources

• AgERA5 dataset: https://doi.org/10.24381/cds.6c68c9bb
• Daily meteo variables 1979-present, global, 0.1 degree
• Available from Copernicus Climate Data Store

​

• Global crop productivity indicators: https://doi.org/10.24381/cds.b2f6f9f6
• Estimates of crop productivity for wheat, maize, soybean, rice
• Global, 0.1 degree, 2000-present

​

36