APPLICATION OF SATELLITE IMAGES TO LOCATE AND INVENTORY VINEYARDS IN THE DESIGNATION OF ORIGIN “BIERZO” IN SPAIN

摘要

With a view to contributing to the improvement of the current vineyard registers, this study presents a methodology for vineyard mapping based on satellite remote sensing systems. The procedure was validated for the Designation of Origin "Bierzo" in Spain. Different supervised classifications were performed based on two Landsat images acquired in the same year. The objectives of the present study were to determine which classification yielded the best results and to quantify the influence of different factors that affected the overall classification accuracy, such as the resampling method, the use of georeferenced mosaics, or the combination of the two images. The classification accuracy was determined by using confusion matrices that were applied to two ground truth images: one of the images consisted of a 700 × 700 m pixel grid, and another image was generated from validation sites. The percentages of agreement obtained exceeded 75%. Based on the results obtained, the area under vines at the municipal level was estimated by linear regression. A strong correlation (R 0.84) was observed between the official statistical data and the data obtained from image classifications