Cheryl McCarthy is a researcher at the National Centre for Engineering in Agriculture, University of Southern Queensland based in Toowoomba.
As inputs costs continue to rise, on-farm productivity gains will come from greater sophistication in managing inputs like labour, water, chemicals and energy. Robotics is enabling the development of farming equipment and systems that can precisely sense and control to manage inputs and save labour.
NCEA is conducting a range of research projects that integrate autonomous sensing and control with on-farm operations, to robotically manage inputs within a crop. Major projects are being conducted in on-farm automation for weed spot spraying, adaptive control for irrigation optimisation, and remote crop surveillance using cameras and remotely piloted aircraft.
Cheryl is developing machine vision and sensing systems for agriculture. Machine vision-based weed detection systems have been developed for the sugar, cotton and pyrethrum industries. A Depth and Colour Segmentation process enables weed detection and a new processing technique enables the vision systems to operate at commercial ground speeds of 10-15 km/h.
Optimal irrigation strategies for overhead and surface irrigation systems are being investigated in projects for the cotton and horticultural industries. Trials in the cotton industry at sites on the Darling Downs and Central Queensland have demonstrated 10-30% water savings with 10% increase in yield, as well as labour savings, when using adaptive and automated irrigation systems which combine soil and crop monitoring sensors and variable rate applicators, together with software to calculate optimal irrigation amount.
Insufficient sampling for diseases or pests in crops and pastures can lead to misdiagnosis of the presence or level of infestation in a field, or uniform application of pesticide in a field where infestation is not distributed uniformly. Similarly, field conditions, including crop growth, water stress and weed coverage, vary spatially and require frequent monitoring to optimise management.
NCEA is developing technology that will couple rapid, field-scale data collection from RPAS with automated data and image analysis to automatically diagnose unhealthy areas of crop (see below).
3D model of cotton crop generated by RPAS and photogrammetry software
Centre for Land and Water residents, AltusUAS are creating detailed farm and crop maps of the LandWISE MicroFarm.
Altus Unmanned Aerial Solutions specialises in the manufacture of professional UAS systems for wide-ranging applications. They build and operate systems of high specification with features including built in redundancy, custom control interfaces and integrated emergency parachute. They offer platforms with class leading flight performance as well as all-weather capabilities.
An AltusUAS Quadcopter sets off on a mission to map the MicroFarm
Even a simple aerial image is highly informative – the view from above changes perception immensely!
MicroFarm fields and Green Shed meeting venue – image captured by AltusUAS
At the MicroFarm, they are using their technologies to survey the site, photogrammetry to process imagery and further analysis to create a 3D model of the MicroFarm and crops.
A composite image created from many overlapped photographs taken from a GPS guided UAV – the basis for a 3D MicroFarm modelA zoom-in on part of the image showing the Green Shed and corner of a crop of mustard
View a flythough of the model (it looks like a video of the site) on YouTube here>
As well as terrain models/topographic maps, they can produce detailed NDVI information.
We are using this information to understand our site in much more detail. With individual pixels as small as about 4cm, we can zoom in practically to individual leaf scale. Do we need that? Not for many applications, but it does raise new possibilities around pest and disease identification and definitely enable us to view individual plants.
On Day 1, presenters discussed sensing, control and robotics. Developments in this area are proceeding remarkably fast, with prototype machines finding their own way around farms, identifying weeds by species and applying custom treatments including sprays only to leaves. Additional presentations on pasture and plant quality detection, grapevine pruning and fruit quality analysis made it a full informative day.
At the end of the day, delegates formed small teams to design their dream agricultural robots – an excellent way to consolidate information. The key however, was identifying what their robots should do (not how) and describing the constraints under which it would need to operate.
As soon as you state a “how” you limit the options that can be considered in determining the final design. Maybe it shouldn’t be a 4-wheeled rover, but an aerial vehicle, or even a ground crawler. Get the specifications right, and the design will identify itself.
AgBot – image from Queensland University of Technology
Day 2 began with discussions around variability. Identifying what variability exists, where it is and whether it justifies custom management is a critical starting place. Speakers also focused on managing two important farm inputs to ensure the right job is done – seed placement and fertiliser application.
Day 2 was completed at the LandWISE MicroFarm at the Centre for Land and Water. There were demonstrations of in-field nitrogen testing, a soil pit to examine, a robot pulling a urine patch detector, a one pass strip-till and planting machine, testing fertiliser spreaders and UAVs.
An AltusUAS UAV takes off at the Centre for Land and Water, a mission to collect data at the LandWISE MicroFarm
Something for every forward thinking agriculturist!
Long-time LandWISE member, Andy Lysaght is featuring on Rural Delivery with his mechanical weeder for squash and other crops.
Agricultural contractor Andy Lysaght has invented a machine to weed squash and similar crops mechanically. Client Gareth Holder, who was paying significant amounts of money for hand weeding of his squash crops, encouraged him to design something mechanical. About a week later Andy came back to Gareth, and said: “Come and play with this.”
Andy has won two major national awards with his Andweeder; the Ravensdown Innovation Award at the National Horticultural Field Day in Hastings, and the Launch NZ Innovation Award at National Fieldays.
Judges said the Andweeder is a step change for the industry, converting an intensive manual process to an automated precision one.
A comparative 10ha weeding trial between the Andweeder and human weeders saw the three-row machine take 6 hours 45 minutes, compared with 125 hours by hand.
Andy says: “We are doing our best to get away from chemicals. We can’t keep doing what we are doing and the rest of the world doesn’t want it either.”
To see Andy on TV, watch Rural Delivery Series 11, Episode 4 on 28 March 2015. First screening, Rural Delivery on TV One, Saturday at 7.00am. Repeats on TV One Sunday at 6.00am and Heartland channel Sunday at 6:30pm, Monday at 12.30am, 12.30pm and 6.30pm
The aerial view above of the MicroFarm was provided by new Centre for Land and Water residents, AltusUAS. This is a wide angle colour image, but we are also looking forward to capturing a range of other image types over coming weeks.
So far this season we have completed fourteen irrigation rotations across our various crops. We have guidance from HydroServices soil moisture monitoring and advice. Their weekly reports are posted on the MicroFarm Irrigation Monitoring page.
The linear-move irrigator that the Ritchie family provided has been critical to success this year – one of the driest we’ve seen for a while. In August we posted a story about rebuilding the linear from parts.
Now, part way through Cyclone Pam, Hawke’s Bay is getting much needed rain. In our case, this has been a manageable amount and we have been fortunate to avoid the strong winds that have caused trouble north of us.
Since last week we’ve received regular showers, with daily rain totals of 20mm, 10mm, 0mm, 5mm and 25mm and moderate rain continuing. So 75mm so far, with about 12mm ET in that time. You can access our weather data from the MicroFarm site weather page, provided for us by HortPlus MetWatch.
Cover crops planted last month are coming away well. We irrigated them to get them started, but with the current rain, that is probably it for the season. The mustard and the oat/vetch mix were both direct drilled by Mike Kettle Contracting.
Our two crops awaiting harvest are sweetcorn for McCain Foods and navy beans for Heinz-Watties. Both crops are looking good.
Looking at the AltusUAS image above, the gaps in the sweetcorn on the left are where crop samples were taken to compare growth of seedlings with and without cover.
At the far end of the MicroFarm, the navy beans glow yellow as they reach maturity. Despite the colour change, HydroServices soil moisture monitoring shows water use has remained high, suggesting plants remain active.
The four white squares in the paddocks on the right are cover cloth on our cover crops: mustard and on an oat/vetch blend. We will remove two covers soon and monitor growth. Two other covers will remain in place for a few more weeks.
We are keen to involve even more technology partners at the MicroFarm.
Wintec has established a trial network of soil moisture sensors which are being compared with the HydroServices neutron probe results.
Installing the WINTEC wireless soil moisture sensor array
ASL software, (LandWISE Yield Estimation software) and apps for pipfruit and irrigation industries) have developed a ground cover measurement tool. We used it to map MicroFarm onion crop canopies and found good correlation with final yields.
If you have something to test or demonstrate, or know someone that might add value, let us know!
Remember, you can follow @LandWISENZ on Twitter, or keep an eye on the MicroFarm website for the most up to date information on happenings.
Peas are one crop that has huge variation. It’s hard to know if the crop will yield four tonnes per hectare or twelve. Even within small sampling plots we measured yields less than 4 t/ha and greater than 13 t/ha. More>
Our first MicroFarm onion crop is extremely variable. We want to measure variability so we can better assess it. If we can measure objectively we can make better decisions. We are interested in spatial variability and temporal variability. More>
Variability in crops implies lost production. While the very best parts in a paddock may still give less than potential yields, anything less is a sure sign of loss.
LandWISE developed a simple “crop loss calculator” called YieldEst. The purpose of YieldEst is to assess how much crop variability is costing a grower.
We have been watching establishment of the onion and pea crops at the MicroFarm. Both have shown high levels of variability, apparent very early in crop establishment. From there it can only increase.
We have visited a number of other sites this spring, looking in particular at crop establishment and seeking reasons for variations seen. We have been a little surprised at the level of acceptance among growers; not happy but somewhat resigned.
Where does variability start? What are the most important causes? What can we do about it?
Variability starts at the start; with variations in genetic potential within seeds in a line. It is surprising how little we really know about this. We do know hybrid lines can be more uniform that open pollinated ones, mainly because they have a narrower genetic spread. But gene differences are hugely complex too. One gene can express differently depending on environmental factors. For example, we don’t know much about the effect that growing conditions of the mother plant may have on the seed we sow.
We have an idea about soil conditions for best establishment, although we wonder if we don’t over-cultivate to get a satisfactory seedbed. We hope we don’t have carry-over herbicides or weeds so growers have started using precision ag section control to get the right dose in the right place. We try not to plant if soil is too wet or too dry. Sometimes it is both in the same paddock.
We do know more about plant density and plant spacing. We know crops have a target population for best returns, and that many are sensitive to variable spacing between neighbours. Precision ag again helps with planter controls. But maybe we are trying to plant too fast when under pressure to get seed into the ground. A lot of research has shown planting speed has a direct impact on yield, largely through spacing uniformity effects.
We know that planting equipment set up and operation has a big effect. Check planting rate is as intended. Get seed to even depth into moisture, cover it properly. Ensure every seed gets sufficient nutrients, but don’t put fertiliser too close.
There is some evidence that weather conditions straight after planting can have a big effect on emergence and establishment. Heavy rain caps weak soils, and if that dries we suffer.
Heavy rain after planting capped this soil and created poor conditions for germination and emergence.
If plants do emerge, they have to survive strong winds, hail, frost, birds and slugs. And our herbicides.
And that is just getting to establishment. There is still most of the crop life to go, but it seems a fail to thrive infant never does as well as a bonny one.
Two thirds of the way through spring our micropaddocks are approaching full canopy. Onions that progressed so slowly over winter are growing rapidly and peas planted in September are well established. A key question now is when to irrigate. More>
Crop covers laid on onions and peas
Squares of Cosio cover cloth placed on peas and onions
Following the lead of Dr Charles Merfield at the BHU Future Farming Centre at Lincoln, we laid some trial crop covers at the MicroFarm. More>
Linear Move Irrigator
The MicroFarm has a linear move irrigator. The Ritchies replaced the original linear at Drumpeel, and after Hugh made some necessary adjustments to the old one, arrived with a truck and trailer carrying a single span linear and a Manitou to put it up with. More>
Queensland University of Technology is investigating robotic technologies as a new generation of tools for site-specific crop and weed management. Tristan Perez described “AgBot”, a platform currently being manufactured to a design by QUT. AgBot is 2m long, 3m (adjustable) wide and 1.4m high.
AgBot – image from Queensland University of Technology
Tristan suggests the use of a swarm of small light robots, that operate at lower speeds and have a suit of sensor-acting devices, could lead to a better application of agrochemicals. He also sees them having a key role addressing herbicide resistance as they could enable use of mechanical or microwave weed destruction techniques.
Now we have your attention:
LandWISE’s Dan Bloomer attended the 17th Precision Agriculture Symposia of Australasia in Adelaide. This event is a collaboration between SPAA and the University of Sydney, bringing together researchers and practitioners with the aim of promoting the development of PA to profit agricultural production.
The Symposium has a broad coverage including new technologies, big data, precision cropping and viticulture and spatially enabled livestock management.
As well as an excellent range of speakers, the symposium is a very good networking opportunity for people active in this space. There is a close alignment with LandWISE interests and the supporting Trade Displays were very relevant and informative
A few notes about some of the other presentations:
Lucas Haag, Professor at Kansas Sate University and Partner and agronomist in his family’s large farming operation, described the evolution of PA in the dryland environment of the US High Plains. While a very different context to New Zealand agriculture, the lessons appear readily transferable.
He noted the critical role of autosteer, rather than yield mapping, in accelerating the adoption of precision agriculture tools among a wide spread of cropping farmers, and the subsequent search by those farmers to gain additional return on that significant investment. This is a pattern very familiar to us.
A take home message from Lucas was the use of PA technologies to make better whole field or whole farm analysis and decisions. While they use PA tools to help evaluate new varieties, seed treatments and other new product options, machinery management decisions have added considerable value to their business.
Examples included evaluating the economics of grain stripper versus conventional header harvesting, grain cart logistics and the value of a dedicated tender truck to support spraying operations – all applications that were not anticipated. Better telematics and machine monitoring technologies and costs of machinery suggest this will continue to be an area of focus.
As manufacturers have increased claims of spreading width, and farmers and contractors have increased bout widths accordingly, arable farmers have noted increased striping and lodging in crops where blended fertilisers are applied. The Massey studies identify the different ballistic properties of blend components, and the increased bout widths, explain these symptoms.
A number of presentations included reference to UAV technologies. Some are very sophisticated and used to carry very high-spec sensors. Some are just used to get up above the crop for a new perspective. Regardless, the potential benefits are clear and the price dropping and capability rapidly rising.
Luke Schelosky of RoboFlight Australia their approach using either piloted or unpiloted aerial vehicles to capture ultra-high resolution imagery create 2 cm resolution maps. As we have seen before, the key to the technology is the processing software rather than the choice of vehicle.
Miles Grafton also reported Massey work using remote sensing for pasture management. They use a number of Remote Piloted Aerial Systems (RPAS) including multi-rotor (e.g. QuadCopter) and fixed wing (e.g. Trimble UX5) as well as a range of multi- and hyper-spectral sensors and imaging systems. While at an early stage, Massey research is showing promise in remotely sensing pasture quantity and quality, including assessing pasture nutrient levels.
Lucas Haag also discussed the role of UAVs, suggesting three unique features create special potential:
Temporal Resolution -The data are fresh, not from the last cloud free satellite pass
Spatial Resolution – the user can control flight height and pattern to gain the redolution needed for the intended use of the data
A separate step – because it requires a separate trip to the field, there is opportunity to add external knowledge before inputs are applied. This allows, for example, adding historical yield monitor knowledge (and perhaps knowledge of herbicide mistakes) to UAV NDVI imagery when creating a Nitrogen application map.
Lucas further addressed spatial and temporal variability measures determined from Yield Maps, contrasting Normalised Yield and Yield Stability. Multi-year Normalised Yield provides the measure of Spatial Variability or spatial yield potential, and the standard deviation (defined perhaps as stable high, stable low and unstable yield) provides a measure of temporal (over time) variability.
Cheryl McCarthy presented crop sensing and weed detection work being undertaken at the National Centre for Engineering in Agriculture at the University of Southern Queensland. Colour and depth image analysis is enabling them to identify weeds in real time, and spot spray them at 10 – 15 km/h. Interesting to note this requires analysis of 30 images per second, as at least three image frames per plant are needed for sufficient confidence.
Two Trade Displays that caught attention were SST Software and PA Source; both aimed at helping farmers and their advisors access the benefits of Precision Agriculture.
Mark Pawsey demonstrated SST Software’s Sirrus programme, a cloud based data access, storage and analysis package. Of additional interest was their decision to open their agX Platformto other developers creating supplementary apps. This reflects the opening also of the John Deere platform and similar moves by other organisations.
Ben Jones of PA Source spoke at LandWISE in 2012 and has facilitated our use of that platform for supply and analysis of spatial data including yield maps, EM maps and satellite data. New offerings include www.watch.farm and www.pastack.com.
Watch.farm delivers Landsat satellite sourced vigour maps to your email every 16 days (cloud cover permitting). Part of the watch.farm package includes change maps, so you can track growth change in individual paddocks. We want to try this service in the current summer season and see it having a significant role to play if the resolution coverage is satisfactory in the “Land of the Long White Cloud”.
PAStack also uses Landsat imagery, “stacking” images from as far back as 1999 to 2013 to see which areas provide more or less biomass and how consistent they are. We will also investigate this product as there appear to be a number of uses of potential benefit to LandWISE farmers and their supporters.
At the MicroFarm, we just harvested our second lot of peas. We tracked their water use since planting to build on learning from our first crop (see the December 2013 issue of “Grower” reproduced here>).
Once again, HydroServices’ Melanie Smith established three neutron probe access tubes in each of two crops. These were read weekly and analysed to tell us paddock soil moisture content down to 80 cm.
Both pea crops were planted on the same day with the same drill. Paddock 1 is dryland and Paddock 2 has drip irrigation installed 200mm deep.
Figures Paddock 1 and Paddock 2 show soil water content for each crop.
Paddock 1_PeasPaddock 2_Peas
We see the crops tracked about the same at the start. In mid-December, Paddock 2 received two 9mm irrigations from our buried dripline.
Melanie estimated that the irrigation was 80% efficient, so only added about 7.5mm to the budget each time. Paddock 1 is not irrigated, and continued to drop towards stress point.
Paddock 1 reached Refill Point on Boxing Day two days before rain fortunately lifted it back out of stress. Paddock 1 again hit stress point on about 6th January. With no more significant rain, it stayed stressed. With irrigation applied as required, Paddock 2 remained stress-free throughout.
Overall, the two crops used similar amounts of water through until early January. After that the 0 – 30 cm soil reached stress point, and water use from the unirrigated Paddock 1 began to taper off. The steeper lines in the bottom part of the graphs show it began to get more water from deeper in the profile.
By harvest, Paddock 1 was using only about half as much water as the drip irrigated Paddock 2 and drawing it from much deeper in the profile.
The difference in what a crop did use and what it could have used if the water was available is described by Potential Soil Moisture Deficit. We estimate that by harvest, Paddock 1 suffered about 100mm of PSMD. I am not sure what the pea response is to stress. I am told it is a “very elastic” crop. For many crops this would indicate a growth reduction of about 20%.
So did irrigation pay?
We sampled each crop pre-harvest and found Paddock 2 had about 30% more fresh weight canopy than Paddock 1. The difference was easily seen, being significantly taller and generally more “lush”. The peas in Paddock 1 reached harvest maturity at least three or four days before the irrigated peas in Paddock 2.
We get paid for peas not canopy. We also sampled yields and quality as measured by TR (pea tenderness) and found differences.
There was a lower tonnage in Paddock 2, but the quality (and pay-out value) was much higher.
At harvest the Paddock 1 tonnages were reasonable at 6.85 t/ha paid yield. But TR was 137; a bit high and the lowest pay-out grade.
We delayed harvesting Paddock 2 for two days. The paid yield was similar at 6.55 t/ha but the TR was 102, a 30% higher pay-out grade.
Paddock 1 returned $2,059/ha and Paddock 2 returned $2,625/ha gross, so a benefit of $566/ha from irrigation.
We applied 81 mm so our return from irrigation was $6.99/ha/mm applied. Many people quote an irrigation cost of about $2/ha/mm so let’s claim a benefit of $5/ha/mm applied.
Looking at it another way. If we had a 20ha paddock, irrigation would have made us about $8,000 better off. If we also sold pea hay, the benefit would be even greater.
Answer: Irrigation pays!
Thanks to: Centre for Land and Water, ThinkWater, Netafim, HydroServices, McCain Foods, Ballance AgriNutrients, BASF Crop Protection, FruitFed Supplies, Agronica NZ, Nicolle Contracting, Drumpeel Farms, Greville Ground Spraying, True Earth Organics, Tasman Harvesting, Plant & Food Research, Peracto NZ
Cheryl McCarthy is a researcher at the National Centre for Engineering in Agriculture, University of Southern Queensland based in Toowoomba.
ASL software, (LandWISE Yield Estimation software) and apps for pipfruit and irrigation industries) have developed a ground cover measurement tool. We used it to map MicroFarm onion crop canopies and found good correlation with final yields.
