Kiwi Fruit Yield Prediction

The Challenge:

In horticulture, uncertainty around yields and crop sizing has a significant impact on market price. Yields and crop sizing are affected by weather, but in complex ways that are not well known.

In kiwifruit crops, these complex interactions between the environment and plants cause uncertainty and variability in yield, fruit quality, sizing, and maturity. Uncertainty in kiwifruit yield and fruit sizing is a significant financial cost to Zespri, New Zealand’s largest marketer of kiwifruit.

There are many complexities in supply and production planning in fresh produce systems with many variables to consider for accurate forecasting of yield, including weather patterns. The weather experienced by crops at key growth stages can cause significant variations in yield outcomes and sizing. It is important to know what the weather will be as accurately as possible during these growth stages.

Weather prediction in New Zealand is difficult, as is the geographical scale of the industry growing regions. One of Australia’s leading agtech companies, The Yield, will therefore assess yield prediction models using gridded weather inputs to improve kiwi fruit yield predictions.

The Solution:

The technology, developed by leading Australian agtech company The Yield, uses historical records, seasonal conditions and farm management practices to create accurate data-driven forecasts of the size of fruit and the volume of fruit produced. Improving yield forecasting will reduce waste from over-production and reduce costs from under-production, better aligning supply and demand.

The project team will deliver predictive models of fruit yield count and fruit sizing for kiwi fruit crops by combining three key capabilities:

o gridded weather capabilities, crop models, and cloud computing and app solutions from The Yield

o crop data from the New Zealand kiwi fruit industry

o data science, climate and crop modelling expertise of UTS.

Data science, historical and current season grower data, combined with gridded weather data, will be used to develop a modelling framework that predicts kiwi fruit yield counts and fruit sizing.

Predictions from these models will be delivered to Zespri through The Yield’s digital solutions which will enable Zespri to improve its price realisation capabilities, providing greater returns to kiwi fruit growers.

Outcomes:

• Improving yield forecasting will reduce waste from over-production and reduce costs from under-production, better aligning supply and demand.
• Readily available future predictions of fruit count and size made available to Zespri at an improved level of accuracy from current standards.
• The prediction models developed will be delivered to Zespri through The Yield’s digital solutions enabling Zespri to improve its price realisation capabilities and in turn providing greater returns to New Zealand kiwi fruit growers.
• The products developed in this project will be available to Australian customers growing similar vine-based crops, for example avocados and citrus fruits.
• The Yield and a University of Technology Sydney (UTS) Data Science team are building on harvest prediction frameworks, boosting the scalability of the technology that can be transferrable across industries such as citrus and stone fruit.

‍