Using digital photogrammetry for evaluating soil erosion risk in olive orchard

The olives orchard is one of the most important crops in Spain. The long-term sustainability in marginal areas depends on good agricultural practices. One of them is the maintenance of grass cover in order to prevent the soil erosion, especially in areas of steep slopes. Therefore determining the percentage of grass cover in these areas it is essential. This research aims the grass cover determination in olive orchard using remote sensing techniques. To achieve this, frames of olive orchard were captured by the airborne sensor Ultracam D Vexcel. Where the selected area was an olive orchard of 32000 hectares in the Valle de los Pedroches in the South of Spain, where 64 digital aerial photographs and data of 173 agricultural plots were used. Object-oriented classification technique was applied in this frames, this means a previous image segmentation allowing isolating the grass cover in the olive orchard. Thus, it is allowed to work with the image at object level, which greatly expands the amount of information that can be extracted from it. The software used was eCognition (c). The resulting objects are the basis for subsequent classification. The results show that was possible achieving an overall accuracy of 96.10% and an excellent Kappa statistic (94.10%) for object-oriented classification. In conclusion this technique opens new perspectives for the determination of grass cover in olive orchards.