• All of the 2020 Viti-Culture Seminars are now available to watch at your leisure!

    Please note that replies to comments and questions may not be responded to straight away. Please login back in to check.

11.30am - PLUMPTON COLLEGE - "Yield Forecasting, Getting it Right" - with Dr Greg Dunn, Head of Plumpton Wine Division.

admin

Administrator
Staff member
In this first seminar from Plumpton College, head of its wine division Dr Greg Dunn will look at yield forecasting and how to get it right.
Having studied a doctorate in botany, Dr Dunn has spent the last two decades developing extensive international research experience, education and
industry networks through many collaborative and viticulture-based projects. He has worked for DPI Victoria, the University of Melbourne,
and the National Wine and Grape Industry Centre at Wagga Wagga, where he was deputy director. He also took on the role of associate editor and then deputy editor of the Australian Journal of Grape and Wine Research

.1594218997439.png

TO ASK A QUESTION, PLEASE CLICK ON THE "POST REPLY" BUTTON BELOW
 
Last edited:

greg

New member
Really good Greg - looks like the only good way to predict bunch weight is berry counting?
Hi Matt, Bunch weight is not easy to predict. Weighing them at veraison and multiplying that weight by a factor hasn't really worked. Neither has using allometric relationships. What I am suggesting, and what seems to work, is weighing a subset of berries (100 or more) from bunches you have randomly collected and weighed and then predicting berry weight at harvest and using the ratio of predictedberrywt (harvest)/berryweightattimeofcollection. You multiply this ratio by bunch weight at the time of collection to predict bunch weight at harvest. For it to work berry weight needs to be reasonably constant from year to year and all your sampling needs to be random and representative. I can chat to you about it sometime.
 

Matt Strugnell

New member
Hi Matt, Bunch weight is not easy to predict. Weighing them at veraison and multiplying that weight by a factor hasn't really worked. Neither has using allometric relationships. What I am suggesting, and what seems to work, is weighing a subset of berries (100 or more) from bunches you have randomly collected and weighed and then predicting berry weight at harvest and using the ratio of predictedberrywt (harvest)/berryweightattimeofcollection. You multiply this ratio by bunch weight at the time of collection to predict bunch weight at harvest. For it to work berry weight needs to be reasonably constant from year to year and all your sampling needs to be random and representative. I can chat to you about it sometime.
Hi Greg
Thanks that would be good - I do have some berry weights from a couple of years back taken over 3 years (I think)

Cheers
Matt
 

Giz Gaskin

New member
Hi Greg, going back to your map of yield, it would be interesting to take a Yara N sensor over the field to see if it can detect variation in crop N vs Yield delivered. Is this something you would be interested in look at with us ?
 

greg

New member
Hi Greg, going back to your map of yield, it would be interesting to take a Yara N sensor over the field to see if it can detect variation in crop N vs Yield delivered. Is this something you would be interested in look at with us ?
I believe that CSIRO and others have looked at this recently using multispectral sensors and indices. Very happy to discuss with you. In any given season there is a positive relationship between yield and vine/canopy size and the NDVI studies bear this out. This issue in viticulture is that, unlike something like wheat, much of the seasonal variation in yield comes from within buds that develop in previous season. This means that canopy characteristics might be similar from one season to the next but carry quite different crop loads. Nitrogen is a particular issue in grapevines too - too much N favours vegetative growth and can lead to depressing yield through effects of shade (and other things) on bud development and fruit set. Happy to discuss.
 
Thanks @greg brilliant and clear overview! Very interesting to hear your focus on berry number: Not sure if you have seen our app Sectormentor (you can see more at vines.vidacycle.com)? Our Yield Prediction Tool follows your research in terms of making it as easy as possible to do effective random sampling and get a good bunch count, and we've had a number of people tell us it really improved the accuracy of their yield prediction. Where our methods differ slightly, until now, is how to further optimise the yield prediction: We had been working with some vintners in Oregon looking at the lag weight method to optimise it further and they do seem to be having good results with this on their Pinot Noir, but we would definitely be keen to incorporate some of the work you shared about optimising the prediction using a berry count. Maybe something we could work on together @Matt Strugnell @greg ?
 

greg

New member
Thanks @greg brilliant and clear overview! Very interesting to hear your focus on berry number: Not sure if you have seen our app Sectormentor (you can see more at vines.vidacycle.com)? Our Yield Prediction Tool follows your research in terms of making it as easy as possible to do effective random sampling and get a good bunch count, and we've had a number of people tell us it really improved the accuracy of their yield prediction. Where our methods differ slightly, until now, is how to further optimise the yield prediction: We had been working with some vintners in Oregon looking at the lag weight method to optimise it further and they do seem to be having good results with this on their Pinot Noir, but we would definitely be keen to incorporate some of the work you shared about optimising the prediction using a berry count. Maybe something we could work on together @Matt Strugnell @greg ?
Yes, we should think about this
 

greg

New member
Hi Greg, 3DBunch app for ios/Apple phone and iPad now available/just released for bunch visualising and berry counting. Have you reviewed this
Thinking about this Peter - I think that this software can get berry number, average berry size and bunch porosity in a few seconds - well worth looking at in the context of speeding up and improving forecasting
 
Top