DETERMINATION of MAIZE PLANTATIONS on A LARGE-SCALE AGRICULTURAL BASIN with SENTINEL 2 SATELLITE IMAGES

This study focused on the determination of maize cultivated lands in Cukurova Basin, Turkey with the object-based classification of Sentinel 2 satellite images assisted with near real-time field campaigns. Cukurova Basin includes one of the biggest agricultural plains of Turkey and ranks first in the production of maize, soybean and pistachio. In addition, a considerable amount of cotton plantations are applied on the plain in addition to citrus farming, which makes the agricultural pattern of the plain too complex. This study uses a single-date image mosaic as input data for classification. Field campaigns provided several ground-truth parcel data with planted crop type, which is used to extract the spectral responses of crops through image bands. This information is used for defining a threshold-based classification schema. Results of the study indicate that, by designing a threshold-based hierarchical classification schema concerning spectral responses of the crops, over 90% accuracy can be achieved in the detection of maize planted parcels.