Abstract
The monitoring of grape quality is a key viticultural process which facilitates improved business performance and increases business resilience. Airborne remote sensing is a methodology which has shown potential for identifying variation in grape quality, however, the scientific development of these approaches and associated methodologies within viticulture is so far limited in scope and extent. This thesis focuses on the use of small Unmanned Aircraft Systems (sUAS) and satellite-based multispectral and hyperspectral imagery for monitoring grape quality parameters. Studies were undertaken at a world-renowned vineyard in the South-East of England over three growing seasons (2019-2022). Airborne data collection, soil sampling, and grape harvesting was undertaken at three locations within the site. Three individual analysis chapters assessed capabilities of machine learning, cover crop segmentation and hyperspectral imagery for monitoring grape yield and quality variation.Results indicate that more advanced remote sensing approaches, involving multiple-VI machine learning and high spectral resolution hyperspectral imagery, can lead to substantial improvements to grape quality monitoring. Machine learning was shown to outperform linear models which are frequently applied within literature studies. Whilst novel evidence for the link between cover crops and grape quality were presented when isolating cover crop pixels using high spatial resolution sUAS imagery. Finally, success was also demonstrated when using Narrowband Vegetation Indices (NVIs) generated from hyperspectral imagery in the 400-1000nm spectral region. However, the sensitivity of hyperspectral sensors to varying irradiance was a clear limitation of this methodology.
These findings are discussed in relation to their potential for a positive and feasible adoption of these remote sensing systems and approaches within the viticulture industry. Identifying how the developed methodologies can facilitate the monitoring of grape quality parameters yet to be explored within viticultural remote literature. This work provides a vital contribution and addresses an important knowledge gap, in global viticulture, which is timely for the emerging, and highly successful, UK viticulture market.
| Date of Award | Jun 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Julien Lecourt (Supervisor), Niall Burnside (Supervisor), Chris Joyce (Supervisor) & Matthew Brolly (Supervisor) |