Search for: All records

Quantifying physical mechanisms driving sea-level change—including global mean sea level (GMSL) and regional-to-local components (that is, sea-level budget)—is essential for reliable future projections and effective coastal management1,2. Although previous research has attempted to resolve China’s sea-level budget from the 1950s3,4, these studies capture short timescales and lack the long-term context necessary to fully assess modern sea-level rise in southeastern China5—one of the world’s most densely populated regions with immense socioeconomic importance6. Here we show that GMSL followed three distinct stages from 11,700 years before present (BP) to the modern day: (1) rapid early Holocene rise driven by the deglacial melt of land ice; (2) 4,000 years of stability from around 4200 BP to the mid-nineteenth century when regional processes dominated sea-level change; and (3) accelerating rise from the mid-nineteenth century. Our results arise from spatiotemporal hierarchical modelling of geological sea-level proxies and tide gauge data to produce site-specific sea-level budget estimates with uncertainty quantification. It is extremely likely (P ≥ 0.95) that the GMSL rise rate since 1900 (1.51 ± 0.16 mm year−1, 1σ) has exceeded any century over at least the past four millennia. Moreover, our analysis indicates that at least 94% of rapid modern urban subsidence is attributable to anthropogenic activities, with localized subsidence rates often exceeding GMSL rise. Such concurrent acceleration of global sea-level rise and rapid localized subsidence has not been observed in our Holocene geological record. 

Lithic Use-Wear Analysis (LUWA) using microscopic images is an underexplored vision-for-science research area. It seeks to distinguish the worked material, which is critical for understanding archaeological artifacts, material interactions, tool functionalities, and dental records. However, this challenging task goes beyond the well-studied image classification problem for common objects. It is affected by many confounders owing to the complex wear mechanism and microscopic imaging, which makes it difficult even for human experts to identify the worked material successfully. In this paper, we investigate the following three questions on this unique vision task for the first time:(i) How well can state-of-the-art pre-trained models (like DINOv2) generalize to the rarely seen domain? (ii) How can few-shot learning be exploited for scarce microscopic images? (iii) How do the ambiguous magnification and sensing modality influence the classification accuracy? To study these, we collaborated with archaeologists and built the first open-source and the largest LUWA dataset containing 23,130 microscopic images with different magnifications and sensing modalities. Extensive experiments show that existing pretrained models notably outperform human experts but still leave a large gap for improvements. Most importantly, the LUWA dataset provides an underexplored opportunity for vision and learning communities and complements existing image classification problems on common objects.