Abstract | The ability to collect robust yield data has been demonstrated in a previous SRA-funded project CSE022 (=2008/022) across a range of different yield sensing options. This same project documented a set of procedures and protocols with which to handle the data (detailed in Bramley and Jensen 2013), in order to generate robust yield maps. It was noted during CSE022 however that the ability to generate reliable maps is compromised by consignment errors (attributing the cane cut to the wrong fields), and more importantly, by how the data is converted to yield (t/ha). Sensor 'noise' derived from the vagaries of harvest, especially at row ends/ haulout changeover/ stoppages is also an issue. How do we take the research outputs relating to yield mapping and monitoring to a commercially implementable product without compromising robustness? This key question will be broken down into two components; 1. When we know the consignment is reliable, how reliable will the yield monitor data be? 2. What are the key performance indicators and methods for assessing whether the data underpinning a yield map is accurate and reliable? |