A review of large language models in geomatics: integrating multimodal data, addressing challenges, and exploring synergies

The exponential growth of geospatial data from satellite imagery and LiDAR to crowdsourced and sensor-based streams poses challenges for traditional geomatics methods. This review examines the transformative role of large language models (LLMs) and vision language models (VLMs) in addressing these challenges by enabling the integration, interpretation, and automation of multimodal geospatial data. The study synthesizes applications across three domains: geospatial data analysis and management, environmental and urban planning, and remote sensing. Particular attention is given to domain-specific copilots and intelligent agents that enhance automation and decision support. The review also highlights advances in multimodal data fusion, autonomous GIS systems, and context-aware mapping tools, while critically assessing limitations such as computational demands, domain adaptability, and ethical concerns. By providing a structured overview of current applications and emerging research directions, the study underscores the potential of language-driven models to augment geomatics methodologies and reshape innovation, decision-making, and policy in spatial information science.