1University of Melbourne
Project supervisors: Tony Weatherley1, Alexis Pang1 and Liz Morse-McNabb (DEDJTR Epsom)
Key Words: precision agriculture, variable rate, NDVI, LANDSAT, crop zones, management zones
- Zones showing low, medium and high NDVI (crop ‘greenness’ or vigour) matched similar zones generated from yield monitor data with 59—74% accuracy
- Mid-season NDVI zones predicted medium yield most accurately, followed by low- then high-yield zones
- Single satellite images from early August predicted the location of yield zones with up to 73% accuracy
- Satellite NDVI has potential to guide paddock zoning and in-season nutrient management in Mallee soils
Why was the project done?
Many Mallee cereal growers use variable rate (VR) equipment and zone their paddocks from local knowledge to reduce input wastage, but uniform paddock management is also common. Complementing growers’ knowledge, yield maps and satellite images can potentially be used to guide VR management in the Mallee’s highly variable soils, making precision agriculture cheaper and more accessible to growers. The project builds on previous work on yield monitor data conducted at Loxton and Carwarp, work incorporating yield logs and remote sensing in the Southern Mallee and studies of site-specific N strategies at Karoonda. MSF had also identified a need for better understanding of yield log data.
How was the project done?
Harvester yield data for winter cereals from between 2007—2015 at Loxton (five crops) and Wargan (two crops) were supplied by MSF members. Free Landsat multispectral satellite images, with 30 m x 30 m pixel resolution, were downloaded for as many dates as possible for the yield-mapped paddocks/seasons. Using QGIS software, these were used to develop Normalized Difference Vegetation Index (NDVI) maps, showing spatially variable crop vigour for several dates. Yield and NDVI data were processed into paddock maps showing zones of low, medium and high yield / vigour, relative to each year’s average. These maps were compared to assess how well yield maps agreed year-on-year, and how effectively vigour could be used to predict yield. Yield maps were also compared with each other to assess their stability across different seasonal conditions.
This was a University of Melbourne Master of Agricultural Science research project, co-supervised and supported by DEDJTR and supported by MSF and participating farmers. Thanks to these organisations and individuals.