Primary Navigation

Case study

Creating a digital orchard

11 Jan 2024

Taking the roots and branches of an apple orchard and transferring them into computer code for manipulation and modelling was once the stuff of scientists’ dreams. But Plant & Food Research’s Digital Horticultural Systems (DHS) project promises to unleash the digital potential of sensing technologies, artificial intelligence (AI) and modelling to enable that dream to be realised.

Creating a digital orchard

Plant & Food Research science group leader Dr John Mawson says digitally replicating an orchard takes multiple modelling tools describing individual facets of orchard production, and puts them into a single unified modelling system.

“There are a lot of different models for the various components of an apple orchard ecosystem, including light, water, nutrients, tree and fruit growth, and crop treatments. The key challenge for us is to build a platform to put them all upon, with software ‘conductors’ to manage and integrate all those models. We need to enable them to communicate to one another across the different computer languages often used.”

John says building “digital twins” is not a new concept in the realm of manufacturing and processing, with the airline industry as a prime example.

“But what is rarer is applying that digital twin technology to complex biological systems.”

The Plant & Food Research team are aiming to combine two modelling approaches, mechanistic modelling, and AI modelling into the digital twin orchard’s architecture, incorporating the best aspects of both.

Where mechanistic modelling draws on understanding a system and how it responds to stimuli and forces, AI pulls on data driven relationships.

“We want to bring the computational power of AI models that are easily deployed into play with the scientific knowledge of our mechanistic, biological models. These models can be used to train the AI which, for its part, has the capacity to provide new insights from large data driven relationships.”

The foundational stages of the project are drawing on data from apple trees growing in Motueka and Hawke’s Bay.

Data are being collected on a per tree basis from static imaging technology and by building 3D reconstructions of trees and their fruit. From these, the science teams hope to measure and model the growth rates of the tree, its leaves, and the fruit. Meantime a moving imaging rig is analysing between-tree differences within the orchards.

The ability to layer up the complexity of the digital twin orchard over time will progressively increase its predictive ability, and the ability of researchers to run “what if” scenarios over an entire orchard environment.

Initially the scientists will have a prototype digital twin by mid-2024 that can predict fruit yield, productivity and quality as measured by fruit firmness and size. It will be capable of analysing the outcome over an entire season from bud-burst to harvest.

“We are starting with something we know we have a reasonable chance of doing accurately and will continue to build our capacity up from there. Each successive year of data will increase our ability to test the accuracy of our models.”

Other dynamic inputs that will be added in soon afterwards will include irrigation water and nitrogen fertiliser input. After that will come pests and disease, understorey growth, fruit storage behaviour, and ultimately consumer acceptance modelling.

The researchers are basing the digital twin on the newly developed Planar Cordon configuration growing Royal Gala apples.

The Planar Cordon configuration is a 2D tree layout that has proven to deliver significant increases in yield and is well suited to future automated operations like harvesting. Researchers also have one eye on the commercial potential of the digital twin, with the ability to incorporate proprietary apple brand data for operators.

“We can see this project providing new knowledge and understanding to enhance the apple industry’s resilience, but it will also be able to be applied to other crops in the future as it is developed.

“It is quite a moon-shot project for us, and we also expect to gain a lot of new understandings from it.”

Get in Touch

Interested?

Get in touch