All posts by LandWISE Admin

Agronomy, Automation, LandWISE 2015, LandWISE People, Precision Agriculture, Sensing

Agri-Intelligent Systems: robots, data, and decisions

Leave a comment

LandWISE 2015 Presenter, Tristan Perez

Tristan Perez Professor of Robotics and Autonomous Systems, Queensland University of Technology, QLD, Australia
Tristan Perez
Professor of Robotics and Autonomous Systems,
Queensland University of Technology, QLD, Australia

Since the 1960s, agriculture has seen significant advances in agrochemicals, crop and animal genetics, agricultural mechanisation and improved management practices. These technologies have been at the core of increased productivity and will continue to provide future incremental improvements. Data analytics, robotics, and autonomous systems are transforming industries such as mining, manufacturing, and health. We are starting to see automation of single agricultural processes such as animal and crop remote monitoring, robotic weed management, irrigation, nutrient decision support, etc. However, we envisage that the integration of these technologies together with a systems view of the farming enterprise and its place within the agri-food value chain will trigger the next wave of productive innovation in agriculture.

The challenge of the next agricultural revolution is to assist farming enterprises to make the management and business decisions that will optimise inputs such as labour, energy, water and agrochemicals and explicitly account for variability and uncertainty across the production system and along the agri-food value chain.

The opportunity for increased profitability, sustainability and competitiveness from finer-scale sensing and whole-farm decision-making and intervention requires farmers to have greater access to digital data and technologies to extract information from data. The agricultural landscape will rapidly change due to low-cost and portable ICT infrastructure.

Agri-intelligence is the integrated collection of tools and techniques – from robots, unmanned airborne vehicles (UAVs) and sensor networks to sophisticated mathematical models and algorithms – that can help farmers make sense of large amounts of data (agronomic, environmental and economic) to make risk-informed decisions and run their farms more profitably and sustainably.

Perez AgriIntelligence

 

The figure below shows the ubiquitous emerging vision of a farm in the second machine age, where computer systems are used to augment human perception and capacity for decision making in complex situations.

PerezAgriIntelligentFarmSystem

The farming enterprise is considered a system that interacts with the environment (through climate, markets, value chain, etc.) The key objective is to make sound decisions about management in order to optimise inputs, yield, quality, and at the same time make the system robust against yield and quality volatility due to climate, commodity market fluctuations, and incomplete information about the state of crop, soil, weeds and pests.

Automation, LandWISE 2015, Precision Agriculture, Sensing

AgBot Design Challenge

Leave a comment

Delegates at LandWISE 2015 formed small groups to specify a “robot” that would address some issue of relevance.

Development of any such machine should follow a systems engineering approach. Four critical features of systems engineering described by Marcel Bergerman are:

  1. Take a top down approach
    • Focus on the “what” not the “how”
  2. Formally define requirements
    • A thorough test plan can be developed before the product/system is
  3. Have a life cycle orientation
    • Consider all aspects from the cradle to the grave
  4. Inter-disciplinary in nature
    • Systems are too complex to do it all on your own

For this exercise, delegates were asked to concentrate on the first two; define what their dream machine had to achieve and describe the constraints under which it must operate.

It is important to take this approach to avoid limiting the possible solutions too early. Deciding your scout robot would be electrically powered three wheel drive eliminates aerial, two wheel or four wheel and wind or fuel powered machines. Sure, the answer may seem obvious, but the final form of many creative solutions that are optimal are not envisaged at first.

Some examples from the exercise include:

A root vegetable harvester that must have minimise soil compaction, remove soil from the produce, minimise mechanical crop damage, and work in a range of weather conditions

A bird scarer for seed crops that must keep birds off the crop, not damage the crop, and must stay within the boundaries of the crop or farm

A pea crop sampler that picks plants randomly through the crop, analyses samples on the go in the paddock and sends the resulting information back to the office. It must cope with the height of the crop and not damage it, and work in a wide range of weather conditions and cope with pugging and mud

An automated machine to remove weeds from vegetable crops grown on beds. It may also monitor plant population and variability, health and vigour, and pests and forecast yields. Working in a range of weather and soil conditions it must have high productivity (ha/hr)

Determine insect populations across field prior to planting of spring crops, including density, location and spread of identified species – include a lure for underground species

Harvest ripe mandarins after determining brix and colour levels and sort according to quality and size with rejects dumped and in-grade fruit transferred to bins for transport

An automatic pine tree pruner that removes every branch up to a height of 6m. Cuts must be clean with no damage to branch or trunk, and it must operate in difficult terrain

An apple harvester that harvests only export grade fruit such that no post-harvest grading is required. Must work 24/7 during harvest within current tree architectures

A fresh broccoli with automated cutting, picking and packing into bins. Will select/sort according to size and colour with no flowering heads.

Agronomy, LandWISE 2015, Nutrients, Precision Agriculture, Soil

Fertiliser Ballistics: must know facts

Leave a comment

LandWISE 2015 Presenter, Miles Grafton

MilesGrafton
Miles Grafton NZ Centre for Precision Agriculture, Massey University Institute of Agriculture and Environment

crops and pastures since upgrading to spreaders capable of spreading at increased, bout or swath widths. This issue is more prevalent where fertilizer blends of products with dissimilar ballistic properties are sown simultaneously.

The problem is more obvious when applicators have purchased modern top of the range twin disc spreaders with the ability to spread at an acceptable spread pattern at tram lines at or greater than 30 m. These spreaders have increased tram or bout widths of spread from 20 – 24 m, to greater than 30 m thus reducing the number of tram lines, increasing output and reducing trafficking of the crop or pasture.

Spreading at a tram line of 30 m requires a total spread pattern to be around 45 m, allowing for a pattern overlap of around 50%, to achieve the desired accuracy (Chok et al., 2014). Given that the spreading discs are around 0.5 – 1.5 m above ground level, then fertilizer particles must be discharged at some considerable speed.

GraftonTable1
Table 1: Distances typical fertiliser particles will travel when ejected at various speeds, in a horizontal plane from 1.5 metre height. (SSP is single superphosphate at 3 different sizes, KCl is potash, MAP is mono-ammonium phosphate, DAP di-ammonium phosphate). The distances are lateral from each disk; total spreading distance is twice that in Table 1.

In order to achieve an optimal even distribution, spreaders deliver 50% of the required amount of fertilizer on each side, which is overlapped with another 50% when the vehicle makes the next run in the opposite direction, or an adjacent run in a round and round spreading pattern. Therefore, the area closest to the border of the distribution area only receives half of the recommended rate as there is no pattern overlap and the desired fertilizer response, would then be reduced as if 50% of the desired rate on average has been applied.

Border spreading refers to the capacity to reduce the application distance on the side towards the border in order to minimize the amount of fertilizer applied outside the zone. Yield spreading compensates for the need to overlap by doubling the amount of fertilizer applied in the boundary side.

Product separation can be avoided by soil testing early, then addressing fertility issues by direct drilling or broadcast application prior to the crop establishing. Then use tram line application on crops for the one product intended to be side dressed during the crop life cycle.

Agronomy, Controlled Traffic Systems, LandWISE 2015, LandWISE People, Nutrients, Precision Agriculture, Projects

Testing On-Farm Fertiliser Spreading

Leave a comment

LandWISE 2015 Presenter, Dan Bloomer

DanBloomer200
Dan is the Manager of LandWISE Inc, an independent consultant, and a member of the Precision Agriculture Association of New Zealand Executive.

The SFF “On-Farm Fertiliser Applicator Calibration” project arose from repeated requests by farmers for a quick and simple way to check performance of fertiliser spreading by themselves or contractors.  They wanted to know that spreading was acceptable.

A calibration check includes assessment and correcting of both application rate (kg/ha) and uniformity (CV). Farmers indicate determining the rate is reasonably easy and commonly done. Very few report completing any form of uniformity assessment.

There are many protocols internationally relating to the spreading of fertiliser products. Lawrence (2007) compared six test methods.

1400 tray matrix used to collect 18 simultaneous transverse tests on a Transpread “W” twin chain spreader From Lawrence, 2006
1400 tray matrix used to collect 18 simultaneous transverse tests on a Transpread “W” twin chain spreader, From Lawrence, 2006

Most used 0.5 m trays organised in a single transverse row to capture the spread pattern of the spreader. No account is taken of the longitudinal variation between individual rows when multiple tests are carried out.

Larwnce Comparisons

The results of the test are given as the bout width where the coefficient of variation (CV) does not exceed a specified level. In all cases the maximum allowable CV is 15% for nitrogenous fertilisers and 25% for low analysis fertilisers.

An On-Farm Protocol

There is no set method dor assessing uniformity. On-farm testing could use a set number of collectors per swath (spacing changes with swath width) or a set spacing between containers (container number varies with swath width). Farmers can decide.

There are however some important principles:

  • Uniformity requires collection of samples from a spreading event and calculation of a uniformity value.
  • Set equipment up correctly according to manufacturer’s instructions
  • Ensure the spreader is horizontal, and at the correct height off the ground
  • Use standard test trays, given the need for baffling to stop fertiliser bouncing out
  • Ensure the spreader is driven well past the trays to capture all fertiliser
  • If a larger sample is wanted, two or more runs at the chosen application rate should be made rather than applying a higher rate.
FertSetUp
Set equipment up correctly according to manufacturer’s instructions
FertTrays
A line of trays laid out across the full width of spread to catch fertiliser. Use standard test trays, given the need for baffling to stop fertiliser bouncing out

Weighing samples is complicated by the very small quantities involved – often a single prill in the outer containers. Scales weighing to 0.01g are required, but satisfactory options are readily available at reasonable price.

An alternative is to assess the volume of fertiliser captured in each tray. Disposable syringe bodies make good measuring cylinders.

FertTestRack
Disposable syringe bodies make good measuring cylinders

Determining a field uniformity will involve either physical or theoretical over-lapping of adjacent swaths.

FertOverlap
Determining a field uniformity will involve either physical or theoretical over-lapping of adjacent swaths

On-line software is being developed to process data and generate statistical reports. Key outputs will be measured application rate, the CV at the specified bout width and the bout width range at which CV is within accepted limits.

Test spread-pattern checks performed to date show there is a need for wider testing by farmers. Unacceptable CVs and incorrect application rates are not unusual.

 

There remains some question about the percentage of fertiliser caught in some types of tray

The SFF project is co-funded by the Foundation for Arable Research and the Fertiliser Association

FertResearch

FAR

 

Agronomy, Automation, LandWISE 2015, LandWISE People, Planting, Precision Agriculture

Micromanaging Your Planter

Leave a comment

LandWISE 2015 Presenter, John Ahearn

John Ahearn
John Ahearn, GPS Control Systems

Modern planters are amazing machines.  Electric and hydraulic drives with precise speed control have replaced ground engaging wheels.

Variable rate is achieved using prescription maps.  Individual row control is achieved using GPS location. So seed populations, fertiliser rates and spray rates are accurate. Up hill, down hill, on curves, headlands, and point rows – the precise rates are applied with zero over-lap.

How do you check your new planter is performing as advertised?  Before looking for improvements we need to get a benchmark of where the performance is now. During planting there are some problems you can identify and fix on the go and some you can’t.  The trick is to record planter data, match it with yield data, and use that analysis to fix or modify the planter for next year.

Most planters have a simple population monitor but the addition of a few extra sensors, and a data management tool, gives an idea of how these variables affect your planter, and ultimately – your yield.  Variables to monitor include singulation, ground contact pressure, ground wheel slippage speed.

When you know what happened during planting, and can compare that to the end of season yield map, you can then figure out the impact these variables have on yield.  That yield impact can be converted to a dollar cost which leads to a discussion on planter improvements for an old planter, and a set-up review for a new planter.

An aftermarket planter control system can be retrofitted to most old planters.  The system will monitor and record the planter variables during the planting operation for each field.

Ideally, the same system will also be used to collect and record yield data during harvest. This completes the loop for data collection and gets all the data in the same place on the same software.  Later analysis is then simple and it is easy to identify opportunities, problems and create prescription maps for next season.

What goes on the planter?

  • Monitor /computer screen in the cab.
  • population sensors – use existing or fit newer better units
  • down force sensors and hydraulic down force control units
  • electric or hydraulic drive system for seed, fertiliser, and other inputs
  • section control

Process improvement is a continual process that leads to higher yields, reduced inputs, and higher profitability.  There are a lot of variables we can’t control, but closing the loop on planter variability finally adds some true value to your yield maps.

Agronomy, LandWISE 2015, MicroFarm, Nutrients, Precision Agriculture, Soil

Monitoring peas from the greenhouse to paddock

Leave a comment

LandWISE 2015 Presenter, Christina Finlayson

Christina Finlayson, Research Associate, Plant and Food Research
Christina Finlayson, Research Associate, Plant and Food Research

Peas grown for processing can be a notoriously variable crop both in terms of overall productivity and maturity at harvest. In turn, this can have a significant impact on paid yields and grower profitability.

Many factors are thought to contribute to variability in peas – some of which are beyond grower control and some of which can potentially be influenced through management.

In recent years we’ve worked to consider a few of these ‘manageable’ factors, and specifically, the effect of seed characteristics on early crop vigour and the response of crops to plant growth regulators (PGRs).

The aim of this work has been to identify practices that reduce plant-to-plant variability in the field and maximise paid yield returns for growers and processors.

In our talk we present results from a greenhouse trial looking at the effect of seed characteristics on early crop growth, and how that relates back to the previous pea seed crop. Preliminary results suggested that different-sized pea seed can result in big differences in early plant size and growth.

Finlayson_PeaSeedlingWeightVsSeedWeight
Different-sized pea seed can result in big differences in early plant size and growth

 

These early differences in productivity are seldom overcome, so getting a consistent ‘start’ is important in minimising future variability issues. However additional variability develops following establishment .

In the field, we have conducted several trials assessing the potential efficacy of a range of PGRs on flowering dynamics and pea yields. Across these trials we’ve seen few clear effects of the PGR types, rates and timings that have been tested.

Finlayson_PGRs

 

Agronomy, LandWISE 2015, LandWISE People, Nutrients, Precision Agriculture, Sensing, Soil, Water

“V” is for variability – origins of variation within a crop and contributions to yield

Leave a comment

LandWISE 2015 Presenter Bruce Searle

BruceSearle200
Bruce Searle, Crop Physiologist and Modeller Plant and Food Research

Why is a crop so variable and what can be done about it? And of that variability, how much can management practices actually affect?

Those who work with crops are well aware of this variability, but ways of addressing it are less clear.

Variability affects profitability so managing it is important. The more variable a crop is, the more the yield will be reduced compared to an even crop that produces the maximum attainable yield, and so profitability is reduced.

Increasingly, the value of a vegetable crop is dependent on providing product that meets some quality criteria – initially, usually size. This means that growing to maximise yield does not always increase profitability if the crop does not meet the quality standards. If variability is high then less of the crop will fall in the grades desired by the market and so profitability decreases.

Causes of variability are complex. It starts with the variability in seed size. Overlaid on this is variability in emergence time, variability in seedling size, variability in plant spacing and effects on competition, individual variability in relative growth rate, and differences in spatial supply of nutrients and water in the field and patchiness in pest and disease.

All these factors interact in different ways. Models are an ideal way to quantify some of this complexity and provide good insight, but the large number of measurements needed and the large number of interactions to compute can limit the value of such approaches.

Here we start to develop a conceptual framework that allows the causes of variability in a field to be identified and enables the contribution of each cause to be considered. We have called this framework the ‘V of variability’.

VforVariability

 

We group causes of variability into spread of emergence, establishment, population and growth, and examine how these factors change for crops of onions and potatoes and affect the variability of size in onions and potatoes and dry matter % of potatoes. We use this to identify how much variability is manageable and the appropriate key management practices to consider.

This framework, when linked with digital information capture at a field scale, could provide a powerful tool for management of variability.

Presentation Authors: Bruce Searle, Jeff Reid and Paul Johnstone – Plant and Food Research

 

Agronomy, Automation, LandWISE 2015, Nutrients, Precision Agriculture, Sensing, Soil

Converting urine-nitrogen into grass

Leave a comment

LandWISE 2015 Presenter Geoff Bates

geoff-low-res-e1423206134526
Geoff Bates, Pastoral Robotics

Environmental pollution results, both from nitrate leaching and greenhouse gas emissions that arise from the high urea content of urine patches. This problem is possibly the biggest single threat to dairying in New Zealand. Pastoral Robotics’s solution addresses this threat whilst increasing the bottom line through increased grass growth without requiring large capital investment. This is in contrast to other methods of nitrate leaching reduction, all of which either reduce profitability and/or involve significant capital investment.

Pastoral Robotics’s solution allows the customer to increase profitability in the following ways:

  • Directly through increased grass growth
  • Indirectly through allowing the continuation of their current farming practice without additional costs
  • Indirectly through opening options to increase milk solids production within regional council-imposed nitrate-leaching or fertiliser N use limits.

The system

After the last cow leaves the paddock Spikey® is towed over the freshly grazed land detecting and treating the urine patches with ORUN®. ORUN® is sprayed only onto recent urine patches, meaning typically only 5% of the pasture has to be sprayed, a huge saving in chemicals.

Spikey® detects changes in the electrical properties of the soil caused by the presence of urine.

ORUN® is a combination of a commonly-used urease inhibitor NBPT (not DCD), and the widely-used growth-promotant gibberellic acid variant (GA3).  ORUN® greatly increases the already large amount of extra dry matter grown in the vicinity of the urine patch, and with it recovery of urine-nitrogen. The NBPT increases the lateral spread of urea out from the urine patch; the GA increases the vigour of the growth. This increase in growth covers all costs of using Spikey®.

Increased pasture growth automatically means reduced environmental losses of N – independent testing of ORUN® by Massey University indicated a reduction in nitrate leaching from urine patches of up to 50% and N20 emissions by 27-37%.

Mini-ME and Spikey

The first commercial version of the Spikey® urine detector and simultaneous treatment with ORUN® spray have been designed to be towed behind an ATV or other farm vehicle.

Mini-ME®, an electric robotic tow vehicle, is under development to eventually take over the towing role, leaving the farmer free to concentrate on his stock.

Presentation authors: Geoff Bates and Bert Quin, Pastoral Robotics Limited

Agronomy, LandWISE 2015, MicroFarm, Nutrients, Precision Agriculture, Sensing, Soil

Monitoring and Mapping Crop Development

Leave a comment

LandWISE 2015 Presenter, Dan Bloomer

DanBloomer200
Dan is the Manager of LandWISE Inc, an independent consultant, and a member of the Precision Agriculture Association of New Zealand Executive.

In an attempt to understand variability in crops, smartphone photos were processed to assess canopy size. By geo-referencing such images, they can be used for spatial analysis. Preliminary results showed considerable promise and a tool has been developed.

One use is for detailed nutrient planning and variable rate application, which requires spatial knowledge of final yield. A case study of an onion crop at the LandWISE MicroFarm is used as an example. Onions have a potential yield of around 100t/ha but the mean national yield in an average year is only 35t/ha.

The onion crop was planted on 7 June 2014. Fertiliser was applied at three intervals, 2 August, 27 September and 24 October 2014. Yields before curing within this 1ha paddock ranged from 0 – 85 t/ha.

Overhead photos of the crop were taken across 18 crop beds on 1 October, 28 October and 14 November 2014. The images were processed by ASL Software Ltd to determine an estimate of ground cover. The crop was lifted on 8 January 2015 and fresh weights taken from each bed.  These final yield results were compared to the images taken during crop development.

Data collected on 14 November show strong correlation with final yield; R2 = 0.86.

OnionCover November

However, these data were collected three weeks after our final fertiliser application.OnionCover November Correlation

Photographs taken on 1 October had an average ground cover of 4.6% (range of 1.1 – 8.4%).

OnionCover November

The measurements at this stage  showed good correlation with final yield; R2 = 0.71 once one image with areas of surface algae was discounted.

OnionCover October Correlation

The October images were taken four days after the second fertiliser application, which could easily have been delayed. They were collected three weeks before the third fertiliser application.

This research suggests simple image analysis can provide early indications of final yield. It also suggests such images can provide timely information for adjusting rates and variable rate fertiliser application in onion crops.

To investigate the potential to create canopy maps, we automatically captured GPS referenced images. The images were processed to determine ground cover and displayed on Google Earth.

CoverMap Onion Map

All image capture and processing was built into a smartphone application by ASL Software. There was a strong spatial pattern that could allow variable rate application.

ASL_CoverMapThe accuracy of the smartphone GPS may be adequate for large scale assessment of crops in big paddocks. However, it was not able to correctly locate the images and subsequent ground cover factors within the correct onion bed. Connection to an accurate GPS signal is being included to better locate each image point.

We thank ASL Software for app development, the LandWISE MicroFarm sponsors Ballance AgriNutrients, BASF Crop Protection, Centre for Land and Water and MicroFarm supporters for access to the onion crop.

 

 

Agronomy, Automation, LandWISE 2015, LandWISE People, Nutrients, Precision Agriculture, Sensing, Soil

Hyperspectral remote sensing to assess pasture quality

Leave a comment

LandWISE 2015 Presenter Ian Yule

Ian Yule, Centre for Precision Agriculture, Massey University
Ian Yule, Centre for Precision Agriculture, Massey University

A presentation by Ian Yule, Reddy Pullanagari, Gabor Kereszturi, Matt Irwin, Ina Draganova, Pip McVeagh, Tommy Cushnahan, Eduardo Sandoval.

Remote sensing methods are becoming much more accessible for end users in terms of access to results and the method in which they are presented. They can be developed into systems for herbage analysis which will measure every square meter of a farm or river catchment and publish the results in the form of a map, as below, rather than complex hyperspectral analytical measurements.  

The team at Massey University have been using a hyperspectral imaging tool called Fenix. It is flown in an aircraft usually at around 500-800 m above ground level and has been used to measure hill country properties within New Zealand in the first instance. The map below shows the level of ME in pasture for an example scene, but the major nutrients can also be mapped in this way.

MasseyHyperspectral

 

The catalyst for purchasing the sensor was the Ravensdown/ MPI, funded PGP project; Pioneering to Precision: Fertiliser Application to Hill Country.

The first scientific objective of the project is to map the nutrient concentration of pasture over hill country properties. The business objective is to provide much better information around the productivity of hill country in order to calculate the fertiliser requirements more accurately and improve the overall utilisation of nutrients.

Previous work indicated that significant financial benefit could be achieved from this approach. This is the first time that an imaging tool has been used and this is important because it overcomes many of the difficulties of on-the- ground sampling and then using these samples to represent the whole farm. It is basically impossible to capture the true variability of these properties from ground sampling.

The sensor detects light in the Visible (VIS), Near Infrared (NIR) and short wave infrared (SWIR) parts of the electromagnetic spectrum. This gives it the ability to determine the bio-chemical properties of vegetation that it observes. It has been shown to be a very robust technology for laboratory analysis and this new development takes it out of the laboratory and in the field.

The images captured in strips are mosaiced together in order to develop a single image for the whole area. Each pixel has 448 different layers of information, corresponding to 448 different wavebands which make up the spectral signature for each pixel. It is by comparing the spectral signature using a number of statistical techniques that the nutrient concentration within the vegetation can be identified.

The big advantages with this approach are that the results can be presented in the form of a map, information can be produced quickly with limited need for laboratory based chemical analysis. All of the complex statistical and analysis processes have happened in the background and the results can be presented with in a Geographical Information System (GIS). In a GIS environment data can be linked to decision making software which will help farmers decide on the optimal fertiliser policy for the farm.

Acknowledgement. The image was produced from a survey from the PGP Project, Pioneering to Precision: Fertiliser application in hill country, which is funded by Ravensdown Fertiliser Cooperative and the Ministry for Primary Industries MPI.

New Zealand Centre for Precision Agriculture, Massey University, Institute of Agriculture and Environment. Palmerston North, New Zealand.