Biomass variations help locate weed targets

Active sensors that measure plant biomass are being tested on mapping weeds in-crop and could potentially determine on-the-go variable spray rates

In a paddock, variation in plant biomass relates to the crop and weeds. At early crop-growth stages crop biomass tends to be relatively uniform, while weed biomass varies according to weed density. The patchy distribution common with annual ryegrass makes it suitable for site-specific management.

Research by Sam Trengove, of Allan Mayfield Consulting, funded as part of a GRDC Science and Innovation Award and as part of a GRDC project (through the South Australian No-Till Farmers Association), on direct injection of herbicides in a boom spray is looking at possibilities for the use of precision agriculture (PA) for weed control.

The three optical biomass sensors Crop Circle™, GreenSeeker® and Yara N-Sensor® have each been found to be effective at readily detecting patches of young ryegrass at early growth stages in lentils and canola.

These sensors measure light reflectance in the red and near infrared (NIR) wavebands. Red light reflectance is sensitive to changes in plant chlorophyll content and NIR light reflectance is sensitive to changes in plant biomass.

The data from these sensors can be used to map weed densities for targeting specific in-crop treatments or to manipulate weed-management practices in the following year.

Weed biomass maps or, ultimately, on-the-go weed identification could be combined with direct-injection spray systems to deliver more appropriate types or rates of herbicide to specific weeds or weed densities.

In trials, biomass maps from the previous year (in lentils) were used to locate pre-emergent herbicide trial sites in low and high-density ryegrass patches. The pre-emergent (cereals) application of trifluralin was spiked with tri-allate. This achieved improved weed control at the high-density site, but provided no significant benefit at the low-density sites (Figure 1).

When the same techniques were used in-crop in lentils to apply higher rates of clethodim (240 grams per litre ai at 150, 250, 350 and 500 millilitres per hectare), improved ryegrass control was achieved in high-density ryegrass patches but only with rates above 250 millilitres per hectare. At the low-density sites there was no significant difference between the three high rates.

GRDC Research Code SAN00013

More information: Sam Trengove, 0428 262 057, email

High resolution can tell a much different story

In many situations management practices are the main cause of in-paddock yield variability

Paddock images captured remotely via high-resolution satellites are showing that, in many situations, management practices can be the main contributor to in-paddock yield variation.

Queensland agronomist Tim Neale, from CTF Solutions, suggests this type of imagery enables ‘forensic agronomy’, which helps a grower understand the causes of yield variation. When the cause is understood, then variation can be managed.

Much satellite data used in precision agriculture (PA) has been based on coarse data. That is, each pixel that makes up the image represents a large area. For example, a 25-metre pixel (625m2) for a satellite image gathered from the LANDSAT or a 10m pixel (100m2) from SPOT. At this resolution much of the variability is averaged out in each pixel and consequently masked (Figures 1a and 1b).

High-resolution satellite imagery provides pixel sizes of less than 2.5m (6.25m2) and much more variability is seen. Yield, EM and other proximally sensed data are generally produced at lower resolutions (100m2 to 400m2) which, according to Tim, immediately triggers the user to think that the only effects are soils or other extensive effects.

Tim has been using high-resolution satellite data with his clients for several years and has been amazed at how often the variation in a paddock is related to management or to factors not identified by other forms of imagery.

From his experience, the timing of data capture during the growing season is not crucial, as poor areas generally appear poor throughout the season.

Using satellites that provide 1m and 0.6m resolution images CTF Solutions has now captured high-resolution satellite images of more than 850,000 hectares across Australia.

Four wavelength bands are captured allowing true colour analysis, so images can be delivered in a range of formats, including aerial-like photos or normalised difference vegetation index (NDVI), within about a week of capture. At about 50 cents per hectare, high-resolution imagery has become cost-effective (minimum areas are required).

The causes of variation being identified by high-resolution imagery include poor calibration of fertiliser spreaders resulting in up to 50 per cent of the paddock being under-fertilised, soil compaction caused by random traffic and poor plant vigour caused by uneven seeding depth/poor seeder set-up.

Spray damage, variety differences, disease and insect damage have also been recorded. Paddock history and previous management, such as header rows, also clearly show up in this high-resolution data, and all have considerable effects on crop productivity.

Tim acknowledges that proximal sensing allows a grower to gather data and use data from a paddock at a specific time. However, until systems cover a greater proportion of the paddock he believes there is great value to be obtained from the high-resolution satellite imagery. As with all remotely or proximally sensed data, on-ground truthing is essential.

More information: Tim Neale, CTF Solutions, 0428 157 208, email

Airborne crop watch on the horizon

Relatively cheap remote-control helicopters are being modified by students at the University of New South Wales (UNSW) to provide ‘autonomous crop scouts’.

Undergraduate Harry Xiao, pictured with PhD student Edward Mak, aims to develop the control systems and programs to enable the larger helicopter to fly to a defined location and execute specific tasks without human intervention.

Powered by a two-stoke engine, the helicopter has a two to three-kilogram payload and can operate in winds of up to about 10 knots.

The addition of cameras and sensors would allow the helicopter to gather data at predefined locations across a paddock or crop canopy. With further development such autonomous helicopters could even be programmed to gather crop samples.

It is planned that the autonomous scouting helicopter would communicate with other autonomous vehicles, such as the robotic tractor and the robotic GreenWeeder currently being developed by researchers at the Australian Research Council’s Centre of Excellence for Autonomous Systems at UNSW.

Harry is excited about the possibilities and believes the ‘sky is the limit’ ... literally. By the end of his degree he hopes to produce an autonomous airborne vehicle that can be used as a practical grain-management tool.

GRDC Research Code UNS00002

More information: Dr Jayantha Katupitiya, UNSW, 02 9385 4096, email

Crop scanning, sampling the next PA step-change

With support from the GRDC two Southern Precision Agriculture Association (SPAA) committee members visited the US to learn and share experience of precision agriculture (PA) in Australian production systems.

Presenting a paper on the ‘Economics of adoption of PA on Australian farms’, based on results from SPAA research, was the catalyst for the trip. However, Malcolm Sargent and Ashley Wakefield also wanted to gain a better perspective and understanding of the adoption of PA in North America and to see what developments were in the pipeline.

Attending the 9th International Conference on Precision Agriculture in Denver, Colorado, gave them a chance to learn about new and emerging PA technologies and their application.

From the conference Ashley Wakefield reports the following observations:

there was an emphasis on crop scanning research using real-time scanning, airborne imagery and satellite imagery to target variable rate application of nitrogen fertilisers;
there were several presentations on real-time protein measurements, including using this technology to predict test weight in cereals;
research is being carried out using crop scanning (NDVI) to determine whether crops are suffering from water stress or lack of nitrogen, before applying in-crop nitrogen;
various on-the-go soil-sampling techniques for a range of soil properties are being developed – for example, a group in Korea is developing a mobile, motorised, digital, cone penetrometer for measuring soil strength and compaction; and
weed recognition to guide precision weeding robots is being developed.
Prior to the conference Malcolm and Ashley visited Case IH and John Deere, major manufacturers of farm equipment for PA systems, and met with organisers and members of organisations similar to SPAA in Nebraska and Kansas.

They emphasised to the manufacturers the need for equipment compatibility and for calibrations that are suitable for use in Australia.

“Achieving this will be an ongoing challenge as, from their questions and observations, many of the people we met in the US had little or no concept of how we farm in Australia,” Malcolm says.

Attending the conference and these visits highlighted that the adoption of PA in the US is at a similar stage to Australia.

However, at the University of Nebraska, every agricultural course offered includes a PA component; education is considered an important part of achieving technology adoption.

Groups such as the Nebraska Agricultural Technology Association (NeATA) and the Kansas Agricultural Research Association (KARA) have been formed by innovative people (growers, researchers and industry) to make agiculture more productive. It is hoped that SPAA can maintain and build on the contacts made with these groups.

The next International Conference on Precision Agriculture will be held in 2010.

More information: www.icpaonline.org, www.spaa.com.au

Back to the list of articles