Exploratory geoplay in metaverse fieldwork: Avatar movement as learning data for geographic understanding

Geogames are increasingly recognized as playful systems based on real-world geographic data and spatial reasoning, yet open-ended geographic play remains underexplored. This study examines immersive virtual fieldwork as exploratory geoplay, where participants engage with geographic environments through autonomous observation, movement, and interaction rather than scores, competition, or fixed missions. We developed a multiplayer metaverse environment based on a life-scale 3D reconstruction of the Byobugaura Sea Cliff in Japan. Sixty-two participants explored it in two rounds: one unguided and one supported by light interpretive prompts. Avatar positions were recorded using the YAIBA behavior-logging system and analyzed through space-time clustering, PCA, trajectory-grid aggregation, and cluster-duration analysis. During unguided exploration, participants clustered around geomorphologically significant areas, while interpretive supports shifted visits toward breakwater and terrace zones. Cluster durations increased in the smaller sub-instance but remained stable in the larger main instance. Questionnaire responses reflected positive experiences and perceived behavioral changes. By framing avatar movement as learning data, while recognizing that movement does not directly measure knowledge acquisition, this study proposes exploratory geoplay as a model for studying locational attention and geographic inquiry within immersive environments.