Jury still out for selective harvesting
As precision agricultural techniques become more widely adopted, researchers have investigated if grain quality can be accurately measured on the go
Reading Dr Rob Bramley’s final report on his GRDC research project titled ‘More good, less bad and ugly – extracting additional value from grain production through selective harvesting’, brings to mind the television program MythBusters.
While he produced some conclusive outcomes, the overall picture was that the concept of selectively harvesting barley to produce parcels of grain with more uniform protein was unproven. If it is a myth it was yet to be busted.
“A lack of robustness in the on-harvester protein sensors meant that we failed to prove the selective harvesting concept, although equally it was not unproven,” Dr Bramley, principal research scientist with CSIRO, reported.
Previous research by Dr Bramley showed the advantage of selectively harvesting wine grapes. By segregating grapes from within or across vineyards by end-product quality requirements, improvements in returns can be large. In one commercial example he found a nine per cent or more financial benefit for grape growers and as much as 58 per cent for winemakers.
When it comes to malt production, a grain parcel with a narrow protein range, for example nine to 12 per cent (to provide an average of 10.5 per cent), is superior to a parcel with a protein range of 7.5 to 13.5 per cent.
“Economic benefits have been estimated for the on-farm segregation of barley by protein percentage, but the degree of benefit depends on the price differential on offer, the degree of protein variation in the paddock and whether or not on-farm storage is required.”
For example, if a 100-hectare barley paddock is conventionally harvested as a single unit and yields 240 tonnes with an average protein of 9 per cent, it would be worth $37,360 or $374/ha.
If 60 hectares yielded 2t/ha at 10 per cent protein and 40 hectares 3t/ha at 8 per cent protein, by differentially harvesting and not blending, the return would increase to $43,360 or $434/ha – an increase of 16 per cent.
If storage is required, based on about nine weeks’ storage, the cost is calculated to be $46.22/t. This would be the break-even premium required for selectively harvesting and storing the barley, assuming no additional harvesting costs. Based on 2011 prices for barley, achieving this premium would have been difficult.
Working with three South Australian growers experienced in the use of precision agriculture, the project team aimed to establish the feasibility and potential financial value of selectively harvesting malting barley crops in commercial paddocks.
Immediately prior to the 2009 harvest, about 70 grain samples were collected from the trial paddocks across the full range of soil and yield variation.
The mean protein content was 10.7 per cent and the range of variation in protein percentage was 8.8 to 14.4 per cent in the Mid North paddock. Similar figures were recorded for a paddock on Yorke Peninsula (range 10.6 per cent, variation 8.1 to 13.5 per cent).
Protein maps showed this variation to be spatially structured rather than random. These initial results therefore confirmed the potential opportunity for selective harvesting.
Spatial datasets were used to assess if differences in grain protein could be associated with differences in yield, elevation and electromagnetic and gamma-radiometric soil maps, as well as in-crop biomass. If such relationships exist then selective harvesting zones could be delineated.
In the year of the trial, the highest protein was consistently harvested in areas of low yield in the paddock on Lower Eyre Peninsula.
However, when historic yield and protein data was analysed only a poor relationship was established.
In the Mid North paddock a relationship was found between areas of high protein and a combination of historic yield data and electromagnetic soil data and this was used to delineate two harvesting zones.
An AccuHarvest on-combine grain analyser (Zeltex Inc., Hagerstown, Maryland, US) was fitted to two harvesters for an on-the-go assessment of protein during harvest. A Cropscan on-harvester analyser (NIR Technology Systems, Condell Park, NSW) was already in use on the third machine. Both systems use near infrared (NIR) transmission as the basis of protein sensing.
The trials to date have proved that monitoring grain protein presents many challenges, especially with the equipment that is currently available.
Irrespective of the parameter being measured, the more frequently data is collected the more reliable the result when the data is interpolated into a map.
When working properly, the protein sensors should log approximately every seven seconds. However, in this study, some data was logged at much larger intervals, apparently due to instrument instability. In contrast, a yield monitor typically logs data every one to two seconds.
“Given the tasks that the protein monitor has to execute, it could be argued that its performance is impressive. However, the data density delivered conspired against the delineation of protein zones that could form a robust basis for a selective harvesting decision,” Dr Bramley said.
The map confidence interval was typically greater than one per cent protein. This generates a problem because the acceptable range for malt grade is only three per cent (nine to 12 per cent protein).
Without marked improvements in on-the-go protein sensors, or radical changes to the protein standards, the application of selective harvesting based on protein percentage alone is going to be limited. Based on this research, the concept remains unproven.
Dr Rob Bramley, CSIRO Ecosystem Sciences,
08 8303 8594,
GRDC Project Code CSA00024