Utilising spatial data for within-paddock soil and crop management

Take home messages

• A flexible approach is required when developing zonal management plans. Multiple tools are available to help define within-paddock variability, yet no one tool can do it all.
• Explore opportunities to unlock yield potential by addressing any known soil constraint/s. Soil zones are generally stable and easy to define. Common applications include variable rate lime and gypsum, and management of waterlogging with paddock design and strategic drains.
• Focus on site-specific crop management once a healthy soil base has been established. Management zones for variable rate fertiliser, fungicides, weedicides, pesticides and growth regulants will typically vary between seasons. They may or may-not align with your soil zones. Appropriate selection of data, farmer knowledge and careful ground-truthing is the key to a successful variable rate program.
• Continue to challenge your farming system and management decisions with seasonal crop monitoring (normalised difference vegetation index (NDVI) imagery, yield maps, plant testing, etc.) and ideally couple with a coordinated on-farm trial program.
• Precision farming should not be difficult or confusing it should integrate seamlessly with your established farm management plan. A team approach with the farmer, agronomist and precision agriculture adviser delivers the best results.

Introduction

All farmers make observations of their soil and topography and assess the implication on yield to help inform how they manage a paddock. In addition, single sample or transect soil testing is commonly used. More advanced approaches may have included sap or petiole testing in-crop for nutrient management. In recent years we have had increasing access to low-cost spatial data which can enable variable rate soil and crop management (i.e. yield, imagery, soil conductivity, topography and soil tests). While not yet perfect, we have seen advancements in the useability, interoperability and connectivity of precision farming hardware. Yet despite these developments the adoption of variable rate management remains incredibly low, especially in South Eastern Australia. Lack of technical support to assist farmers with this management approach (hardware, software and agronomics) and perception of low return on investment are just a few known barriers to adoption.

Sub-optimal allocation of inputs (potential lost opportunity for both input savings and yield increase) and missed opportunity to resolve variable soil constraints remains an issue for many paddocks with a paddock scale management unit approach.

There is a range of tools that can inform or define paddock variability. The applicability of the tool(s) used to define and describe within-paddock variability will be determined by the soil constraint or productivity issue that is being tested. Often a single layer of data may not define the issues and the treatment.

Common tools used to define and describe variability include:

• Yield data – can identify zones of variable productivity which warrant further investigation but will not identify factors that are constraining yield without additional knowledge and/or data. Yield data can also be used to determine nutrient removal maps (phosphorus is the most common nutrient to be managed with the influence of this data).
• Electromagnetic induction (EM-38 dual dipole) measures a combination of soil salinity, clay and soil moisture at depth (approximately 1.5m). For many paddocks it can provide an accurate soil map. Soil cores and associated analyses (ideally segmented cores to explore soil profile) are essential to extract paddock management value from such soil maps.

Figure 1. Maps from three data tools highlighting the value of integrating datasets for developing zones.

• Gamma radiometrics are another soil mapping tool used predominantly in combination with EM-38 on sand plains and in gravel soils where soil conductivity is less effective in isolation. It effectively measures the natural radioactive decay of potassium, thorium and uranium in the top 30cm to 45cm of the soil.
• NDVI provides an assessment of canopy density, biomass and plant vigour. NDVI will not define the underlying factor promoting or limiting plant growth, it simply indicates where to investigate (visual, plant testing, soil testing, etc.) and helps define boundaries around such factors. Multiple layers of data and or integration of existing paddock knowledge will be required to validate NDVI imagery. For example, nutrient imbalance would require additional soil testing to understand what is deficient, while suspected diseased crop would need to be tested to identify the pathogen and to prescribe a treatment.

Satellite derived NDVI maps (30m to 30cm resolution) offer a low-cost whole-farm monitoring opportunity which can help benchmark paddock performance and assess management (i.e. detect uneven seeding or spreading, spray misses or uneven irrigation).

Drones or unmanned aerial vehicles (UAVs) can provide similar services as a satellite but also have the flexibility to target timing and the area that is monitored. They offer a superior resolution (1m to 1cm) for more detailed insights into crop and even individual plant performance.

From an active crop management point of view, NDVI imagery can inform the need for a range of activities including variable rate nitrogen (N), variable rate fungicide and crop growth regulants, strategic weed control (spray and/or cut for hay) and definition of seasonal yield constraints such as waterlogging, pests and sub-soil constraints.

• Global Navigation Satellite System (GNSS), in addition to machinery guidance underpins elevation mapping. This data can be converted into digital elevation models, informing water management such as drainage design, water logging zones, erosion control and water harvesting.
• Soil testing can define a broad range of soil physical, chemical, biological and hydraulic properties. Disease pathogen and nematode monitoring can be included in this process.

Our understanding of economic response curves to a range of crop inputs links back to many years of research in soils with known chemical element concentrations. Farmers and agronomists utilise keys measures for management decisions such as Colwell phosphorus coupled with phosphorus buffering index (PBI) for phosphorus (P) applications and soil pH(CaCl₂₎ coupled with crop sequence plan to determine a lime application rate. We can extend this thinking to how to develop an accurate variable rate management application in areas of the country where temporal yield variability is less consistent/predictable (i.e. anywhere other than low rainfall dune-swale systems).

We ideally want to use the same soil measurement from the input response research for developing variable rate management zones. This takes us to the most common method of paddock zoning in the world, grid soil mapping (soil transect sampling per grid unit, typically 1-4ha). The simplicity is appealing as we no longer rely on building assumptions around relationships between spatial datasets. A grid soil map can provide a baseline for several seasons (especially for farmers who can link nutrient removal maps via their yield data). Grid Colwell or Olsen P, pH(CaCl₂) and exchangeable cations (potassium (K), sodium (Na), magnesium (Mg), aluminium (Al)) are the more common soil elements tested when grid soil mapping (0cm to 10cm).

Soil probes/sensors which assess soil elements in-situ are becoming increasing popular as they offer a lower cost per sample enabling greater sampling density. In our experience, these sensors struggle to generate accurate results across all soil types and moisture levels and thus it is essential to calibrate every paddock with traditional wet chemistry laboratory analysis.

When addressing within-paddock variability the highest return on investment (ROI) is often achieved by addressing the key soil health factor/s limiting crop performance. Soil acidity (VR lime) and sodicity (VR gypsum) are common strategies to ameliorate soil constraints and in doing so unlock yield potential. For example, the aim of the VR lime is to establish a minimum target pH level (i.e. pH(CaCl₂) of 5.2) across the entire paddock, while avoiding issues associated with over-liming.

As the key soil constraints are ameliorated and broader measures of soil health are improved, site-specific crop management can be implemented. Seasonal management strategies (predominantly VR nutrition, disease and pest control) are developed with the combination of underlying soil levels (grid or zonal soil testing) and crop monitoring information (NDVI, yield maps and boots on the ground). Macro nutrient supply (predominantly N, P and K) via seeding and spreading are the most common VR strategies employed.

A note of caution must be applied when yield and or NDVI maps are solely used to define zones for VR management strategies. Plant growth and final yield is influenced by a multitude of factors, with the dominant influences being the key plant growth constraints for that particular season. Paddock knowledge is essential when interpreting yield and NDVI maps and such maps can be valuable in defining dominant constraints but rarely offer ability to map sub-clinical factors. For example, crop performance zones for a certain paddock may be great at defining severe sub-soil constraints (rooting depth/plant available water capacity (PAWC)) but do not accurately inform the variability of soil P and/or surface soil pH.

Figure 2. Comparison of results from 2ha grid soil mapping of 80ha paddock near Rokewood. Colwell P (left), variable rate lime prescription map (middle) and Colwell K (right).

Lumpy capital investments (i.e. five yearly lime application costing +$200/ha) attract an interest in a VR approach as the immediate input savings (i.e. VR lime average saving of 30%) more than offset the cost of mapping, plus such jobs can be implemented by contractors with VR technology.

Other paddock mapping investments may be applied over several seasons such as a grid soil P map coupled with P removal (yield) maps. A quick evaluation of what a 15% input efficiency gain can help determine the likely ROI. For example P @ 80kg/ha x $680/t x 5yrs = $272/ha. 15% decrease in input cost = $40.80/ha. It can be difficult to model the extra income derived by improved P management (especially if areas of the paddock are P deficient), yet such value should be considered when evaluating the investment.

Conclusion

We are seeing an explosion of interest in ag-tech and we can expect to see a myriad of technologies hit the market in coming years. When evaluating the value of a sensor technology for your farm, start with a clear understanding of exactly what it is measuring and how this data can be converted into an action. A well-designed ground-truthing process must be employed when relying on surrogate datasets to determine VR application plans. Where possible link paddock mapping analysis with well-established science-based principles. In essence, precision agriculture is simply the intensification of an agronomic management decision. Secure partnerships with businesses who can support you with all steps of the VR management process: agronomy, data collection/processing and application.

Useful resources

Precision agriculture - common sense with a dash of technology

Variable-rate application of grain cropping inputs

References

Precision agriculture - variable rate

Contact details

Andrew Whitlock
0458 312 589
andrew@precisionagriculture.com.au
@andrew_whitlock