Abstract

Three decades ago, James Bogen and James Woodward argued against the possibility and usefulness of scientific explanations of data. They developed a picture of scientific reasoning where stable phenomena were identified via data without much input from theory. Rather than explain data, theories ‘save the phenomena’. In contrast, I argue that there are good reasons to explain data, and the practice of science reveals attempts to do so. I demonstrate that algorithms employed to address inverse problems in remote-sensing applications should be understood as attempts to identify phenomena by explaining the data. Thus, this paper furthers understanding of data-to-phenomena reasoning in science, and demonstrates theory may play a more central role in phenomena identification than previously recognized.