Behind subjectivity: Factors affecting visual image interpretation for acquiring reference data for agricultural land monitoring

Visual interpretation of remotely sensed images is commonly applied to acquire reference data used for development and validation of large-scale land maps. Information resulting from visual image interpretation can be communicated in a form of a map or a reference dataset. There are many factors affecting the quality of visual image interpretation. This thesis aims to investigate factors affecting visual image interpretation for acquiring reference data for agricultural land monitoring. It was found that interpretation consistency is affected by individual aspects of interpreters, by analysed land cover, land use and land change type. Furthermore, consistency of visual interpretation is influenced by the adopted method of enhancing interpretation agreement, such as reviews and feedbacks. Presence of subjectivity in reference dataset cannot be ignored.