Current developments in plant modelling at Wageningen-UR

Summary:

• Focus on mechanistic dynamic models of crops

  • Detail: from summary models to complex biochemistry/physics

  • History:

    • Photosynthesis & canopies: square meter of plants

      • LINTUL: Light intensity and utilization

      • GECROS: Genes, biochemistry and environment/management

      • WOFOST

      • ORYZA

    • Functional Structural plant models (FSPM): individual plants

      • Explicit 3D project

      • Cell arrangement to forest structure

      • WUR Virtual tomato

      • Can be used to design greenhouses

      • Netherlands plant eco-phenotyping centre (use lidar and reflectance measurements)

      • Institute for Advanced Studies for Photosynthetic Efficiency

• Mechanistic crop models

  • Radiation use efficiency models: simple (LINTUL)

  • Water use efficiency model: simple

  • Canopy photosynthesis models: depends on more factors, more sophisticated (WOFOST, GECROS)

•  Drivers

  • Potential production: defined by CO2, Temp, Radiation, crop features

    • (limit of biology)

  • Attainable production: limited by water & nutrient availability

    • (limit of farming practice)

  • Actual production: reduced by weeds, pests, diseases, pollutants

    • (limit of environment)

  • WOFOST 7.2: potential and partial attainable production

  • WOFOST 8: adds nitrogen availability constraints

• Phenological development of wheat (different detail levels)

  • Vegetative -> Reproductive

  • Tillering -> Stem extension -> Heading -> Ripening

  • Seedling growth -> Tillering -> Stem elongation -> Booting/Ear Emergence -> Flowering

  • BBCH Scale of phenology: common schema for crop lifecycle

• WOFOST: above-ground portions of a plant

  • Soil processes modeled by other models: SWAP, tipping bucket, VIC, Noah-LSM

  • Modules: Soil, Weather, Light, Crop

  • Limitations:

    • Many parameters to configure

    • Sensitive to model’s initial state

    • Growth is source-driven, not sink-limited (e.g. if grains become sterile they cannot be filled anymore; this limitation is not modeled by WOFOST)

    • No explicit crop architecture

    • Canopy is homogeneous, not organized into rows, etc.

    • No translocation of assimilates between organs (will be in v8)

    • Limited knowledge of crop response relations (mainly empirically-observed relationships)

  • Best for near-optimal growth regime

  • Many implementations of the model in different languages

    • Extensive test set to validate new implementations

    • Extensive documentation: “gentle introduction”, user manual, reference manual (most detailed)

    • Jupyter notebooks of Python implementation

  • Accuracy:

    • Depends strongly on calibration data

    • Decent accuracy in reproducing major patterns of their tests sets

  • Applications:

    • MARS Crop yield forecasting system

      • Applied to Europe, Russia, Uk

      • Regional crop yield forecasts and analyses

      • Level 1: meteorology -> regional interpolation maps

      • Level 2: crops -> WOFOST -> crop simulations

      • Level 3: Stats on yield, area, indicators -> Estimates of future yield

    • GYGA: Global yield gap analysis

      • Explores possibilities for improved production across the world

      • International initiative with local agronomists

    • WaterWijzer in Landbouw

      • Impact of water supplies on crop yields

  • Further developments of WOFOST

    • Improved N limited growth through leaf N concentration

    • Data assimilation to account for spatio-temporal variability (Ensemble Kalman filters, 2DVar)

    • Better integration with soil models

• Digital Future Farm

• Resources:

  • AgERA5: daily meteo variables

    • .1 degree

    • 1979 - Present

  • Global crop productivity indicators: 

    • estimates of crop productivity of wheat,e maize, soybean, rice

    • .1 degree

    • 2000-present