Abstract
Abstract. Detecting the seasonal agricultural crop pattern accurately is a vital part of the agricultural planning. In this extent, Cukurova Region that is located in Eastern Mediterranean Region of Turkey was evaluated on agricultural landscape pattern. This region is the most productive agricultural region of Turkey also crop variability and yield are higher than many parts of the world. The main agricultural part of the area is called Lower Seyhan Plane (LSP) and it has been formed by the Seyhan, Ceyhan and Berdan rivers. The purpose of the study was to define the wheat, corn and cotton crop pattern using multi-temporal Landsat satellite images and object based classification approach for 2007 and 2013 cropping years. Three main crop’s areal difference were evaluated and changes were monitored between 2007 and 2013. The accuracy of the classifications were obtained by the spatial kappa statistics. Overall kappa accuracy was derived to be 0.9. Classification results were shown that wheat areas were decreased 35% and corn and cotton areas were increased 49% and 69% respectively. Particularly, government subventions and market demands were impacted cropping pattern in the region significantly. In addition, multi-temporal Landsat images and object based classification were a great combination to define regional agricultural crop pattern with very good accuracy (>90%).
Highlights
Many field crops have various growing cycle in a season and mapping the agricultural fields are not an easy process because of temporal variability (Chen et al 2008)
For example; if a pixel size is 1 X 1 km, there should be more than 2 different land use/cover (LUC) in one pixel and traditional hard classification techniques are not effective in this condition
This paper presented a dominant agricultural crop pattern mapping and monitoring study in a complex Mediterranean Agricultural Basin
Summary
Many field crops have various growing cycle in a season and mapping the agricultural fields are not an easy process because of temporal variability (Chen et al 2008). Supervised classification techniques are applied to satellite images or aerial photos frequently in mapping stage in agricultural areas. Supervised object based classification approach is outshined in the literature recently (Şatır and Berberoğlu, 2012). Object based classification provides more accurate results especially in agricultural or urban areas because of pixel grouping based on spectral, textural and shape similarities of the pixels (Blaschke 2010). Agricultural mapping accuracy depends on the RS data type, classification scheme, available classifier and training samples (Cingolani et al 2004). Image classifications with coarse spatial resolution data have some uncertainties on mixed pixel effects. For example; if a pixel size is 1 X 1 km, there should be more than 2 different land use/cover (LUC) in one pixel and traditional hard classification techniques are not effective in this condition. Fuzzy techniques can be used to map mixture degree of each pixel (Ozdogan and Woodcock 2006)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
