Remote sensing provides an unparalleled vantage point for observing the Earth's long-term surface evolution, demanding models to not only perceive land cover at isolated moments but also track changes, memorize evolution histories, and reason across time and space. Existing studies lack a systematic evaluation dissecting these distinct competencies. To address this, we introduce ChronoBench, a multidimensional benchmark that decomposes the task into four progressive cognitive levels: Land Cover Perception, Temporal Recognition, Long-Term Memory, and Spatio-Temporal Reasoning. ChronoBench comprises 12 sub-tasks and 17,689 rigorously validated QA pairs. Extensive evaluations reveal that mainstream MLLMs fall drastically behind human experts, with Long-Term Memory being the most critical bottleneck. Motivated by this, we propose GeoChrono, an MLLM with enhanced capabilities for tracing, memorizing, and reasoning about long-term geographic evolution. Leveraging the physical prior that geographic parcels remain spatially fixed while their semantics evolve, we design a Temporal Trajectory Encoder (TempEnc) to construct per-location temporal trajectories for dedicated land cover evolution modeling, and introduce a Coarse-to-Fine Token Compressor (C2FComp) that adaptively preserves dynamic regions while compressing the static background. To support training, we construct ChronoInstruct, a 104K-sample instruction-tuning dataset spanning all competency levels. GeoChrono achieves state-of-the-art performance on ChronoBench, surpassing leading commercial MLLMs by over 20%, while C2FComp reduces visual tokens by over 56% while retaining GeoChrono's 94.6% performance. The code and data will be available at GitHub.
Blogger's Review: GeoChrono significantly enhances long-term temporal understanding in the remote sensing domain through the introduction of ChronoBench and the innovative designs of TempEnc and C2FComp. Its breakthroughs in long-term memory lay a foundation for future applications in geographic information systems and environmental monitoring, showcasing the potential of deep learning in spatio-temporal reasoning.