Summary
The report tested the feasibility of using satellite imagery to map the area under vine, comparing vineyard boundaries digitised from imagery at three resolutions (0.6 m, 2.4 m and 4.8 m) against a known, accurate vineyard register. Vineyards could be identified with 96% accuracy at 0.6 m resolution, but coarser imagery (2.4 m and 4.8 m) could not reliably distinguish vineyards from other row crops. The report used these findings to estimate the cost of a national vineyard geographic dataset and recommended a staged approach, beginning with a scoping study to reduce the imagery capture area using existing data sources. An ongoing Wine Australia project provides updated information on this topic. Includes test data and accuracy comparisons from imagery of varying resolutions captured over the Adelaide Hills.
The Australian wine industry has indicated for a number of years that a national vineyard database could assist industry planning and management. In 2005, the Rural and Regional Affairs and Transport References Committee of the Australian Senate in its report, The Operation of the Wine-Making Industry (2005), recommended the establishment of a national register of vines.
An accurate record of national wine grape plantings (including geographic locations) could assist in pest and disease management, estimating grape supply, infrastructure planning and environmental management. The project objectives were:
- Identify the most cost effective resolution of satellite imagery that can be used to create an accurate database of vineyard boundaries
- Estimate the cost to create a national vineyard geographic database
The project was conducted in the Adelaide Hills in South Australia, which was chosen due to the variety of trellised row crops such as apples, pears, cherries and flowers in proximity to the vineyards.
Imagery was supplied by the DigitalGlobe, Quickbird satellite at two ground resolutions, 0.6 and 2.4 ms. The 2.4 m imagery was re-sampled to 4.8m. Vineyards were identified from satellite imagery and created manually using heads-up digitizing. Once a vineyard was identified, the block boundaries were manually digitized, ie drawn on screen and saved in a GIS vector file. The digitised vineyard blocks identified from the satellite imagery were compared to a highly accurate vineyard database.
Abstract
Research was undertaken to evaluate the effectiveness of satellite imagery to differentiate vineyards from other horticultural production. Imagery was captured at 3 spatial resolutions (0.6m, 2.4m and 4.8m) of the Adelaide Hills, a wine region where a high degree of diverse row crops is known to exist. Datasets of vineyard boundaries were created using manual heads-up digitising and the accuracy of each dataset (0.6m, 2.4m and 4.8m) was assessed against a known highly accurate vineyard dataset (The Phylloxera and Grape Industry Board of South Australia’s vineyard register). The study confirmed that an accurate vineyard dataset could be established using satellite imagery at a resolution of 0.6 m or better. Recommendations for creating a national vineyard dataset were made based on the findings of this study.
Downloads
Categories











