September 27, 2013
Opening the door to precision black-grass management
Targeted grass weed management based on automated detection and mapping is set to join the growing arsenal of precision farming technologies helping UK arable farmers produce more from less, thanks to a pioneering new project.
Offering the opportunity to improve black-grass control, in particular, while significantly cutting herbicide costs, the four year eyeWeed project is co-funded by the Technology Strategy Board. Its Agrii-led consortium of commercial and academic partners includes the University of Reading, Knight Farm Machinery, Syngenta, Patchwork Technology and NIAB-TAG. Now into its second full season, it aims to turn Reading University’s successful 2009/10 proof of concept into a practical automated weed mapping system robust enough for reliable use on the farm.
Following preliminary trials in 2011/12 using a cab-mounted camera, the system being tested this summer on a number of farms involves four forward-facing cameras along the sprayer boom to capture images during T3 spraying. Black-grass heads are identified, densities estimated and weed patches mapped. These maps can used for accurate pre- and post- emergence patch spraying in the following crop as well as to highlight possible areas of herbicide resistance.
Agrii agronomist Carl Flint, who leads the project, agrees that the ideal would be a ‘real-time’ system to detect weed patches and spray them during the same pass. Unfortunately, however, such a system would preclude pre-emergence patch spraying. So, with experimental evidence showing a good association of black-grass head counts in June/July with black-grass seedling counts the following autumn, the ‘off-line’ approach of detection and mapping at T3 for subsequent autumn control offers the most practical solution.
“The eyeWeed system is part of our company-wide focus on precision agronomy and decision support,” he explains. “Potentially there are big savings to be made in one of our biggest crop protection costs by spraying only where the weeds actually are. This is also very positive environmentally. Indeed, it might even be the key to retaining chemistry that would otherwise be lost in a future of growing legislative pressures.
“Knowing precisely where our black-grass problems are without the need for time-consuming manual mapping opens-up better targeting of cultural controls too. With variable seed rates and more competitive varieties, for instance, we have the opportunity to increase local plant populations to improve the crop competition which our industry-leading black-grass research at Stow Longa research shows to be particularly valuable. Doing this without the need for blanket seed rate increases is the best way of minimising associated problems like greater disease and lodging risk as well as cost.”
While the current project uses an on/off approach to patch spraying, the eyeWeed monitoring and mapping technology is set to offer even greater benefits with the progressive development of twin line and direct chemical injection spraying systems that enable herbicide mixes to be varied ‘on the move’.
The success of the system, however, depends on its ability to detect low levels of black-grass with sufficient accuracy. A key advantage over other weed mapping approaches is the use of high resolution digital cameras close to the crop canopy. This eliminates the need for ‘ground truthing’ the field after image capture to confirm it really is black-grass that has been detected. To ensure eyeWeed is sufficiently robust to do this reliably in practice, the team is testing it extensively at different timings and under a range of crop conditions.
“The complexity and amount of data processing required has certainly been a challenge,” explains Reading University scientist, Dr Alistair Murdoch. “It has taken a lot of software development to get to this stage.
“However, we’re confident we can separate black-grass at low levels from wheat once the heads are above the canopy. We’ve also proved we can develop accurate maps of black-grass infestations using current geo-referencing technology. And we know patches of seed heads in one season are closely correlated to areas of infestation in the next.
“We’re fine-tuning our camera set-up this season to work equally well over 24m and 36m boom widths,” he reports. “At the same time, we’re extending our field testing to more farms under more conditions. And we’ll be validating the degree of the black-grass control we achieve across these sites through patch pre-emergence spraying from our maps with a 6m buffer zone around the patch to allow for GPS drift and some seed dispersion.
“There’s a way to go yet before we have a fully functioning farm system. But we’ve got all the elements in place now. So it’s only really a matter of time.”
Amongst other things, the current validation work will include re-mapping of black-grass patches in second wheat crops in herbicide-treated fields as well as visual assessments to maintain a record of the changing black-grass population; something Carl Flint sees as a very valuable management tool for future commercial use.
“My growers and I would really value black-grass maps produced automatically as a matter of course from T3 monitoring in every wheat crop in the rotation,” he stresses.
“They’d enable us to accurately target the key elements of our grass weed management every autumn, saving time and money. They’d also give us the best possible indication of the success we’re having in getting on top of this troublesome weed. And by highlighting poor control they should help us identify any areas of developing resistance we can tackle more aggressively before they become even more problematic.
“Our project has concentrated almost exclusively on black-grass to date, but there’s the clear potential to develop eyeWeed to detect and map a range of other problem weeds which form stable patches, like barren brome, ryegrass, wild oats and cleavers,” Carl Flint concludes.
“Then, there’s the possibility of using other colour and texture changes across field crops to accurately pinpoint and manage a number of other agronomic challenges. Add on the extent to which the system can be integrated with other key components of precision agronomy like Soil Quest mapping and it’s future becomes even more exciting.”