Skip to Content Skip to Navigation
Back to Highlights

World-first yield forecasting technology offers avocado growers 93% accuracy

Words by Casey Dunn

Accurate yield forecasts are essential for determining avocado harvest requirements. But estimating how many fruit you’ll pull off a tree is not easy. On a medium-sized orchard of 10,000 trees, growers estimate yield by counting anywhere from 70 to 1,000 trees, twice during the growing season. At five minutes per tree, it’s a laborious and costly task.

And is often inaccurate. A 2017 trial on a 30-hectare orchard conducted by the University of New England (UNE) indicated avocado yield forecasting can be out by up to 120%. Taking into account labour, land management, packing and marketing costs, that overestimation, constituted an error worth $1.5 million. With more than 13,000 hectares of avocado orchards across Australia, the need for better yield forecasting is enormous.

RELATED: Multi-armed mango auto-harvester set to speed up picking by 90%

The UNE’s Applied Agricultural Remote Sensing Centre (AARSC) has the solution: CropCount, the yield monitoring app, which puts more accurate yield forecasting in the hands of every avocado grower. 

Related organisations

Logo for Hort Innovation
Multiple industries

Hort Innovation

Hort Innovation is the grower-owned, not-for-profit research and development corporation for Australia's horticulture industry As one of the nation's 15 Rural …
  • Location

    Australia wide

  • Organisation type

    Research funding body

A clever refinement that delivers over 90% accuracy, in a fraction of the time

Funded by Hort Innovation, and co-developed with the expertise of Sydney-based creative agency, Circul8, CropCount combines satellite imagery with targeted on-ground sampling to produce a yield forecast with better than 90% accuracy. Using a sophisticated algorithm, CropCount identifies which trees should be counted to give the most accurate picture of health and yield across an orchard. 

As AARSC director Professor Andrew Robson explained, “It reduces the number of tree counts required from 70 to 1,000, to just 18. With CropCount, growers take 9 hours to do their yield forecast, not the 500 hours they once did.”

“With CropCount, growers record the fruit count for those 18 trees on the app, which sends the data to us. We do the processing and derive a yield map, fruit size map, tree disease map and a forecast for the block [or paddock], which goes back to the grower to inform their decision-making. The accuracy at the block level is better than 90%. At the farm level, it’s generally better than that,” said Andrew.

The methodology has been years in development, with industry engaged from day one. “We started seven years ago and we’ve gone through lots of different methodologies that didn’t work before we got back to this one – which is the most reliable and the most simply adopted. Because it’s not really changing what growers do. It’s just refining where they do it – with a lot less labour and better results,” explained Andrew.

World-first forecasting tool looks at not just the yield – but why it varies

The dramatic improvement in accuracy is possible because CropCount considers the yield variability that is so common in avocado production. “Avocados can suffer from irregular or alternate bearing. In some years, the same trees produce 1,000 fruit; the next year or two years later, they’ll produce two fruit. So it is very difficult to do a generic prediction for yield forecasting." 

"Secondly, if you prune the trees or if they are suffering from any diseases like phytophthora, or from a weather event that has reduced yield because you are manually calibrating the variation for each season, you get the most accurate answer."

RELATED: Innovation in horticulture to benefit people, plants and the planet

CropCount is a world first. “There are lots of providers who provide imagery – but a lot of the imagery is just about health. It’s what’s called NDVI [Normalised Difference Vegetation Index]. It says, ‘This is a healthy part of an orchard; this is a sad part’. But for us, that’s just the starting point. All the work we’ve done for the last seven years is to work out what the imagery means for disease, health, yield and quality – and how robust is it, across seasons, regions, and varieties,” said Andrew. 

It is powerful information for growers. “Knowing yield variation guides harvesting logistics - labour, bin placement etc, as well as informing targeted agronomy to determine what may be causing low production – for example, irrigation, nutrition, pests or disease. The app supports producers to do that. So that’s our difference.”

CropCount’s initial business, commercialisation plan and development of the ‘pitch video’ featured below was completed through the GATE incubator program, run by NSW Department of Primary Industries.

CropCount

Presented by Janine Googan

Putting CropCount in the hands of every Australian grower is top priority 

Hort Innovation funded the development of the app to the prototype stage and Andrew is excited about the next phase. “The next 6 to 12 months is about completing the full version of the app from the Minimum Viable Product – with extension and validation beyond Queensland to growing regions in Western Australia and South Australia.” 

While the commercialisation pathway is still to be finalised, delivery to the Australian market – to levy payers who funded the development – is our number one priority. 

But the opportunity extends beyond Australia. “We’re creating something that globally doesn’t exist,” said Andrew. So it is little wonder CropCount has drawn the attention of major commercial avocado producers internationally, including from South America and South Africa – who together account for 26% of global avocado production (compared to Australia’s 2%). The potential in these two markets alone is enormous.

Looking for partners to help with scaling CropCount to other industries

But the CropCount opportunity goes beyond just avocados. “The process we’ve developed for CropCount is directly relevant for many tree crop industries. We have another project going on at the same time in mangoes, citrus, olives and macadamias. For mangoes and citrus in the first instance, because they’re big fruit and they’re easily observable and countable, CropCount is easily translated – potentially in the next 12 months once we’ve completed the development for avocado.”

“For apples, pears, stone fruit – crops we haven’t even looked at yet – this would work. The benefits to Australian industries across the board are really achievable,” said Andrew. 

There is opportunity for investors who recognise the benefits of CropCount to come on board to test, refine and scale the technology to suit their industry. “We’re really confident and we’re really happy with the approach and the methodology, and think it offers a lot of benefit to other industries.”

For years, tree crop growers have had to make do with best guesses on how much fruit they’ll pull from a tree that season – and worn the cost of those mistakes. CropCount is set to change that.

 

 

Enjoyed this story?  Sign up for our newsletter here and stay up to date with our latest stories, news and opportunities.