Advances, challenges and prospective research when geographical information science meets digital twin

Digital twins, originating from industrial engineering, demonstrate considerable potential in the field of Geographical Information Science (GIS) in recent years, nevertheless, the absence of a unified research paradigm leads to significant disparities in geographical twinning frameworks, twinning technologies of geographical objects, data models of twinning objects, and visualisation in twin space as well. These discrepancies have hindered the development of geospatial digital twins. To address this issue, this study carries out a comprehensive review of 154 articles related to digital twins and GIS, published between 2019 and 2024, and describes their advances, potential challenges and future directions. The key insights are: 1) Spatiotemporal data models possess limited capacity for managing the virtual–real alignment of all geographical entities; therefore, an integrated and intelligent data model is required to replicate all geographical entities and realise virtual–real interactions within geospatial digital twins; 2) Twinning technologies that fail to consider dynamic behaviours constrain geographical scenario simulations; consequently, a geographical process twinning technology is expected to become a prominent research; 3) A collaborative visualisation scheme offers an optimal solution for addressing multimodal data, multi-user operations, and scenario simulations; 4) An integration strategy combining geospatial twin frameworks with geospatial artificial intelligence (GeoAI) models will become another prospective research. These findings contribute to understand the foundational theories and key technologies of geospatial digital twins, further promote its development.