Relationships between remote-sensed data, spatial distribution of Ridoifio segetum and sunflower yield map, a preliminary approach

摘要

Sunflower is an important crop which is frequently infested by the umbeliferous weed Ridolfia segetum in the Mediterranean area. The objective of this study was to establish preliminary information on the relationship between crop yield, weed infestation and remote-sensed imagery. The study was conducted on a 42 ha farm named Matabueyes located in Cordoba province, southern Spain, naturally infested with R. segetum. Sunflower was planted early March and harvested mid-August. Harvesting was carried out by a Massey Ferguson harvester equipped with the Fieldstar system to generate a yield map. Conventional-colour (400-700 nm) and near-infrared (700-1100 nm) airborne images were taken at mid-May, mid-June and mid-July over the whole farm. Weed infestation was mapped using the Spectral Angle Mapper method (overall accuracy = 94%). Five percent of the total pixels (1.39 M) were selected at random for statistical analysis. NDVI index from the airborne images taken in mid-May turned out to be the most influential spectral factor to estimate sunflower yield distribution, which was also inversely affected by R. segetum infestation.