The potentials of Sentinel-2 and LandSat-8 data in green infrastructure extraction, using object based image analysis (OBIA) method

Green infrastructure (GI) mapping and monitoring is crucial in urban areas, and remote sensing is widely used to accomplish the task. Improved moderate resolution Sentinel-2A (10 m) and LandSat-8 (15 m) images, in place of commercial satellite images, enable GI mapping with little to no cost. Considering so, the objective of this paper is to evaluate the potential of GI feature extraction of Sentinel-2A (S2) and LandSat-8 (L8) (freely available images) using the Object Based Image Analysis (OBIA) method. The advantage of using OBIA over pixel-based analysis has been investigated primarily with very high resolution images. Using OBIA, bottom up (i.e. Multiresolution) and top down (i.e. Spectral Difference) segmentation were implemented using eCognition to obtain image objects for both S2 and L8 images. Then, rule-based classification was performed to extract GI areas from the objects. NDVI, NDWI, NIR/R ratios were utilized in rule set development, after several trial and error process. Both S2 and L8 provided acceptable extraction of GI for urban areas. However, with an overall accuracy of 71.24%, S2 was more effective when extracting GI areas. Shadows along roads and high rise buildings caused some inaccuracy in classification.