Dan Bloomer attended the 20th Symposium on Precision Agriculture in Sydney.
The PA Symposium brings together farmers, growers, researchers, advisors and industry to discuss and absorb developments. Speakers covered cutting edge research, on-farm application by researchers, advisors and farmers, and industry background information such as the state of telecommunications and data ownership.
As Brett Whelan told delegates, “The purpose of precision agriculture has always been to increase the number of correct decisions made in the businesses of crop and animal management. It is a logical step in the evolution of agricultural management systems toward increased efficiency of inputs relative to production, minimized waste and improved product quality, traceability and marketability.”
Crop and soil sensing continues to develop, and there is increasing use of new approaches. Canopy assessment has relied heavily on NDVI, the 1970s vegetation index chosen for distinguishing forest from desert and ocean. In recent years a wider range of sensors capturing more light bands (blue, green, red and infrared) have become affordable and available. Some look at red-edge and thermal infra-red, two bands often related to crop stress of some form. Off the shelf cameras that fit simple UAVs are within farm budgets now.
Ian Yule described research with hyperspectral sensors that capture very detailed images with hundreds of light bands. Hundreds of ground control samples provide “real” information and enormous amounts of data get analysed to identify relationships. The capacity of this to determine species, plant nutrient status and other useful information is remarkable. The current research equipment and processing is very expensive but assume price drops as commercialisation progresses.
Machine vision including object shape, texture and colour is being used to recognise individual objects such as plants, parts of plants or specific weeds. Discussing robotics research to guide decision making on vegetable farms Zhe Xu noted, “If a human can recognise something, a machine can be taught to as well.” Get used to artificial intelligence, neural programming and autonomous phenotyping!
We presented our own onions research which is using smartphone cameras to capture very useful crop development information quickly and cost effectively. Combined with crop models and web based calculation we can predict final yields with fair accuracy early enough to support crop management decisions.
An Australian vegetable research project is using similar approaches to support decision making in carrot crops and investigating others with promise. That team includes researchers and farmers, and is increasingly using yield monitors for crops such as potatoes and carrots. Converting yield data to value allows farmers to estimate costs of variability and how much to invest to fix problem areas.
Data capture, communications and analysis was a key theme. Kim Bryceson described the establishment of a sensor network and analytics using IoT (internet of things) tools at Queensland University Gatton. Rob Bramley explained a process that predicted sugar yields at regional scale to promote better fertiliser management in that industry. Patrick Filippi presented a “big data” approach to predicting grain yield.
The data revolution is changing our world in ways we can’t yet imagine. The increasing amount of things measured, the spatial scale and time span of collection and development of data science to analyse huge streams of information revolutionise our understanding. These are exciting times. Some jobs are going to go, but others will be created as we require completely new skills for jobs not heard of a decade ago.
“We are all in the position of making decisions from a limited understanding or a particular perspective, working with biological systems that are incredibly complex and impossible to fully understand, “ said Ian Yule. “Recent experience with new sensing technologies and data processing has produced new information that challenges our preconceived ideas and understandings,” he said.
The PA Symposium is presented by SPAA, the Society for Precision Agriculture Australia, and the Precision Agriculture Laboratory at the University of Sydney. There has always been a New Zealand presence because while some details are unique, the tools and processes are for the most part generic.
James has over 15 years’ experience in software development, advanced analytics and data visualisation.
Tony has a distinguished track record in predictive analytics and data mining. His specialties include machine learning and computer vision. 