Kimberley Low / Wine Australia
Kimberley Low / Wine Australia
08 Jun 2018
Previous  | Next News

This project is supported by funding from the Australian Government Department of Agriculture and Water Resources as part of its Rural R&D for Profit program.

The idea of using drones to provide accurate, immediate and cost-effective snapshots of the micro-climate anywhere in or above vineyards is a step closer to reality. As Professor Anthony Finn puts it: ‘We’ve got the physics sorted, now we have to work on some of the engineering’.

The Research Professor of Autonomous Systems at UniSA’s Defence and Systems Institute in Adelaide is conducting a project funded by Wine Australia and the Australian Government Department of Agriculture and Water Resources as part of its Rural R&D for Profit program, which has just completed a first series of trials in the Barossa, working with Pernod Ricard Winemakers, Treasury Wine Estates and consulting firm Barton Vale Technologies.

To everyone’s delight, the concept worked. In simple terms, that concept goes like this. If you equip an unmanned air vehicle (UAV) with a normal camera and a long-range infrared camera set up to work as a temperature sensor you can measure the temperature of vines, grapes, buds and the ground itself.

As UAVs also make a very distinctive noise, by measuring this and converting it to sound speed between the ground sensors and the UAV you can create 3D snapshots of temperature and wind velocity up to a height of 500 metres. When you put all the bits together it allows you to monitor and assess temperature and wind conditions at the meso- and micro-scale without needing masses of very expensive equipment.

Aerial image showing the vine/ground temperatures at approximately 2pm. The vines (which transpire) are significantly cooler (~40oC) than the ground temperatures (> 50oC); and the darker (cooler) areas along the vine rows show areas where irrigation pipes have burst.

‘We ran some experiments during heat stress and obtained a bunch of air temperature profiles in 3D and a series of surface temperatures for the vineyard and you can clearly see the temperatures in the vineyard get up to an average ground temperature of around 50°C, even above 60°C in some areas, but it’s much cooler in others’, Prof Finn said.

‘So we’ve been able to generate evapotranspiration maps and crop stress indices at the micro-scale level, which is the goal. 

‘We still have to do all this for frost, but we know the concept works. Now the challenge is to fit it all on a smaller and more manageable UAV and go for it.’

There have been challenges getting to this stage, of course, not the least being when the drone manufacturer quietly changed its interface just weeks before the trials began. That meant rewriting not only all the software for the on-board equipment, but also the software off-board because all the parameters had changed.

‘It was a massive task and we missed a few trials, but fortunately we caught the last heatwave in February about a fortnight before picking began.’

3D images of air temperature wind velocity taken at approximately the same time as the aerial image (2pm). These show good correspondence to measures taken by independent instruments deployed in the field (LiDAR wind measurements and drone-carried instruments).

‘We will be looking for very different things during frost events’, Prof Finn said. ‘For instance, when we fly the drones carrying the infrared camera during heat events we’re obviously looking at full foliage vines and vineyard temperature differences from 60-odd down to 35°C, possibly even 20°C. It’s a massive variation, but it’s around that.

‘When we are looking for frost we’ll be looking for budburst and the challenge there is getting enough pixels onto what are essentially sticks and being able to pull the key aspects within the imagery out with sufficient clarity to be able to do the calculations that we need to do. The temperature variations are also likely to be much smaller and more subtle.’

The other priorities are to add some additional sounding to get bi-directionality and, most importantly, to improve the networking and the efficiency of communications back to a ground station.

‘We need things to be much more automated and in real time so we are telling people what the weather conditions are now, not two weeks ago.

‘We still have a lot of work to do, but we know the mathematics works in so far as we know how to compare sound fields and derive time delays and covert that to air speed, wind speed and temperature.’

Things began in summer, Finn says, in large part because it is easier to predict when heat is coming than frost. The technology still needs to be tested in cold and frosty conditions, however, so the team is aiming to move down to Coonawarra in September.