Yield variability - the 'flip-flop' scenario

Figure 1. Temporal average map of yield/biomass for the paddock

By P Fisher, M Abuzar, F Best, T Fay, P Rampant, D Rodrigues, R Armstrong and C Bell

Most growers know that within any paddock there are high-yielding areas, low-yielding areas, and areas that yield somewhere in-between. An objective of Precision Agriculture (PA) is to find the optimal economic and environmental management practice for each of these areas.

The increased availability of yield monitors has enabled growers to map yield variability within paddocks. However, as the quantity of yield maps has increased it is becoming apparent that some areas within paddocks can “flip-flop” between high, medium or low yielding, in different years.

This inconsistency between seasons is the focus of the component of the Precision Agriculture Initiative SIP09 that is co-funded by the Victorian Department of Primary Industries.

A common approach to PA is to ignore the individual seasonal variability in yield maps and to use either average results over a number of years, or a single season"s results to decide the optimal management practices for different paddock zones. This can provide significant increases in growers" returns.

However, it is not uncommon to find that 40 to 60 percent of a paddock can be defined as seasonally variable. In these areas, a management practice based on an average season will often be economically under-performing. If we could predict how crops would respond in these variable zones under different seasonal conditions, growers could gain important financial benefits.

DPI Victoria is undertaking, in collaboration with the Birchip Cropping Group and farmers Ian and Warrick McClelland, a detailed study of spatial and temporal variability of yields in two paddocks in the southern Mallee region o f Victoria.

“On these paddocks, as with many of my paddocks, I have observed different reactions to seasonal conditions on different soils within the same paddock,” Ian McClelland says.

The paddocks have been classified by the project into areas of high, medium, or low-average yield (fig 1), and have also been classified into areas where the yields are either consistent (temporally stable) or variable (temporally unstable) over a range of years (fig 2).

Figure 1. Temporal average map of yield/biomass for the paddock

Figure 2. Temporal viability map of yield/biomass for the paddock

Figure 2. Temporal viability map of yield/biomass for the paddock

Yield mapping is still a new technology and few growers have yield maps for the same paddock and crop over several seasons. For this reason co-researcher Dr Mohammad Abuzar has used a technique to construct average-yield and yield-variability maps by combining actual yield maps with data from satellite images.

Satellite images in the form of a spectral index called “NDVI”, taken close to flowering, provide an estimate of plant canopy density, which can be related to final yield. The method used by Dr Abuzar is extremely powerful as judgments about a particular paddock"s variability can be made without needing to wait years for a grower to accumulate actual yield data.

However, the method also has drawbacks due to biomass at anthesis not always being a good predictor of final yield, such as in seasons with severe frost or terminal drought stress. Better understanding the relationship between crop density around flowering and final yield will help the rapid analysis of seasonal variability.

The DPI team will continue to investigate this relationship over the project"s next three years.

Birchip Cropping Group economist Fiona Best is undertaking an economic impact analysis of various zonal management strategies at the experimental site. Preliminary analysis suggests that over three years of wheat yield data, when the paddock is managed uniformly, the high yielding zones have a gross margin 150 percent higher than the low yielding zones (fig 3). This provides an excellent opportunity to improve the management system so that gross margin returns are maximised in all areas.

Longterm gross marginThe analysis suggests that precision management of just a single input, such as nitrogen, could have significant economic benefits. For example, reducing the nitrogen application to match the lower yields in the medium and low-yielding zones, could result in a saving of about $3300 across these two paddocks.

Nitrogen management is one of the most promising options for farmers to manage the seasonal effects of paddock yield variability. This can include varying the total nitrogen applied to match the zone yield potential, but also splitting the applied nitrogen between pre-drilling and top-dressing applications.

The objective of this practice is to manipulate the quantity of leaf area (green-area index) to match the availability of soil-water during the season. Too much nitrogen early in the season will produce big, leafy plants that will deplete soil-water more rapidly, and with a dry finish to the season the chances of the crop haying off are high.

Too much nitrogen is also an unnecessary expense and a potential environmental hazard. In contrast, too little nitrogen and the crop will not reach its potential yield or quality.

Ian McClelland believes that PA offers farmers more scope for tactical management of crop canopies. He says growers do not have to operate only from a sowing perspective but can also make adjustments as the crop is growing.

During 2004, the project team, with the Birchip Cropping Group, will test these concepts to find better ways to manage spatial variability and reduce growers" exposure to the uncertainty introduced by climate variability.

The impact of the different nitrogen rates on the economics, water use and final yields will be monitored.

The intention is to demonstrate the opportunities for using PA techniques to improve the management of crop variability under commercial conditions.

However, the project is also investigating the factors causing spatial and temporal variability so that data collected on the experimental paddocks can be more precisely applied to other sites.

The first step in understanding the causes of variability on the experimental paddock has been to characterise how the soil"s physical and chemical properties vary across the site. DPI researcher Paul Rampant has assessed the paddock using the latest sensors. These include electromagnetic induction (EM), ground-penetrating radar (which can detect changes in soil layers and compaction zones), and radiometrics (which is a measure of the very low levels of naturally occurring radiation emissions from the soil).

Some of the remotely sensed data shows a remarkable similarity to the measured yield maps in some seasons.

Single sensors (particularly EM) are being used to identify changes in soil conditions, and while these are useful tools for understanding paddock variability, they often don"t tell the full story. The project is investigating how the integration of multiple sensor data will result in much better tools for high precision soil mapping.

Information from the field trial and the site characterisation will eventually be used to develop future and more detailed experiments. It is hoped that these will unravel the complicated interactions between seasonal climatic conditions and soil constraints.

The project team is interested in hearing from growers who have yield maps of the same crop and paddock for five or more years, and would be willing to make them available for analysis.

For further information:
Dr Peter Fisher, 03 5833 522, peter.fisher@dpi.vic.gov.au

GRDC Research code: DAV 00030

Region South