ADAS hosts: Use of crop sensing in field vegetables and potato crops
Monday 9th November: 9:30-11:00
The event brought together drone imagery and remote-sensing experts, together with researchers and growers from the UK, the Netherlands, Sweden and Australia.
Theory and practice: How to use Ag-tech in the real world?
Bert Rijk of Dutch-based Aurea Imaging
Aurea started off as building their own unmanned aerial vehicles but have now moved to data processing and analysis. The team today consists mainly of data analysts who now work closely with a network of drone pilots, described by Bert as ‘Uber for drone pilots’
Drones used consist of fixed wing drones, which are better suited for larger areas, and ‘(heli)copter’ drones, which can fly closer to the ground. Different types of sensors are used also; from simply visual to multispectral and thermal imaging. More recently, it’s the introduction of RTK GPS technology that has been a real game changer for remote sensing, explained Bert. This type of technology enables precise aligning of images taken from several different flights over time – this is particularly useful in spot spraying and phenotyping scenarios where precision is key.
‘One of the biggest challenges is how you integrate drone imagery into your business profitably’ highlighted Bert. This is why Aurea works with growers and breeders to support them with their decision-making in a way that makes financial sense for the business.
Bert then went on to highlight examples where drone imagery has made a positive difference to their clients. Potato growers have been able to use emergence maps of potato fields to use for insurance purposes when dealing with seed producers that will have supplied the potato seeds they have planted. In other cases, breeders have come to Aurea to assess the potential use of remote-sensing compared to manual observations. For example, flower breeders concluded that the duration and intensity of flowering was better assessed with Aurea’s imagery analysis than manual observations.
Bert’s key message was to ‘start with what your client wants, don’t over focus on the technology and start simple’.
Analysis of field scale crop reflectance data using ADAS Agronomics data analysis methods
Susie Roques of ADAS
For root crops and other vegetables, assessing and predicting marketable yield is still a hot question for growers. As a result, scientists at ADAS and other partner research institute in Belgium, Netherlands and France are exploring whether imagery data be used as a proxy for marketable yield for potatoes and vegetables, as part of the EU-funded Interreg 2 Seas ‘INNO-VEG’ project.
The project has made use of Agronomics, a statistical tool developed by ADAS in conjunction with the British Geological Survey. This has enabled the modelling of underlying field variation in on-farm strip-line trials. And as a result, has made results of over 200 on-farm trials to date more robust.
Susie, crop physiologist at ADAS, shared the results collected from onion and vining pea trials, where different types of imagery based on different vegetation indices were assessed (NDVI and NDRE for example). For vining peas, results have shown that NDRE provides a better yield prediction compared to NDVI, which is consistent with the known suitability of NDRE for assessing dense canopy crops. For onions, a good correlation was observed between drone measurements and marketable yields. This time, it was NDVI that showed the best correlation. These couple of examples show how it’s important to consider the type of crop you are working with, and which vegetation index might be better suited.
With now up to 200 trials being assessed with Agronomics, some important lessons have emerged:
- An even field is key for more precise results.
- Variation across tractor tramlines is OK, as all treatments will be exposed to that variation.
- Less is more – keeping the number of treatments relatively low, for example by simply testing against the farm standard, will keep things simple.
Crop sensing making sense?
Jacob van den Borne of Van den Borne Farm
Jacob farms over 500 hectares of potatoes in the Netherlands. Innovation and the integration of different technologies are integral to the success of his business.
On the farm, sensor data is collected from tractor cabs, spray booms, drones and satellites. This crop data is collected with soils and weather data and used to inform and continually improve crop management decisions such as irrigation management, fungicide and fertiliser applications. This successful integration of data has translated in significant reductions of input use across the farm.
- Emerging weeds have been reduced by 90 % thanks to spot spraying
- Fungicide application to treat late blight has been significantly reduced by implementing variable rate application and only targeting the plants that need treating
- Water usage has been reduced thanks to the use of thermal imaging to assess soil moisture content and integrating it with weather data to optimise irrigation scheduling.
Jacob does not solely rely on sensor technologies. For him, ground trothing the data is as important so regular sampling of tubers and leaf canopy also form part of field evaluations.
Validating precision ag tech for vegetables
Julie O’Halloran of Queensland Department of Agriculture & Fisheries
Julie has been working with farmers in an Australian government-funded project looking to encourage the adoption of precision agriculture technologies on more vegetable farms. Imagery from drones has received relatively little take-up on vegetable farms compared to arable farms.
The project aimed at putting case studies together to showcase how commercially-available remote sensing technology and data analytics can make a difference on farm and to get a clear breakdown of the cost-benefit ratio for farmers.
Some of these case studies included assessing flowering area in maize and head counting in lettuces. The data obtained from maize flowering areas enabled the scheduling of spray timing applications around this critical growth stage. For lettuces, the automated plant count at different stages of the growing season has highlighted a 40 % discrepancy between the number of planted and packed lettuces.
Julie also pointed to the barriers to technology adoption in vegetables that were encountered during the project. Some of these barriers included a low awareness of commercially-available technology, a lack of technical support in adopting new tools and their perceived vs actual cost when accounting for time and money invested.
Jeff from Hummingbird Technologies highlighted some of the data analytics expertise available at the company and its increasing focus on using remote sensing data and associated AI-powered analytics to develop sustainability assessment tools, such as carbon sequestration.
Russel from AgriVue gave an example of how AgriVue’s drone flying technology was used in a field experiment to compare the suitability of the resulting imagery with that of manual assessments. A 0.9 correlation was achieved, illustrating that crop sensing can go hand in hand with manual scores.
Igor from Solvi showcased how Solvi’s farmer-friendly data collection and analytics platform can be used for estimating yield in cabbages, automating plant counts and monitoring weeds.