Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Oct;646(8086):856-864.
doi: 10.1038/s41586-025-09600-z. Epub 2025 Oct 15.

Modern sea-level rise breaks 4,000-year stability in southeastern China

Yucheng Lin  1   2 , Robert E Kopp  3   4 , Haixian Xiong  5   6 , Fiona D Hibbert  7 , Zhuo Zheng  8 , Fengling Yu  9 , Praveen Kumar  3   4 , Sönke Dangendorf  10 , Hailin Yi  11 , Yaze Zhang  12
Affiliations

Modern sea-level rise breaks 4,000-year stability in southeastern China

Yucheng Lin et al. Nature. 2025 Oct.

Abstract

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.

PubMed Disclaimer

Conflict of interest statement

Competing interests: The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. Spatiotemporal distribution of sea-level proxies and cities.
a, Spatial distribution of sea-level proxies and main cities along the China coastline, annotated with relevant river systems, provincial boundaries and tectonic plate configurations. Cities were selected for modern sea-level budget analysis based on the concurrent availability of sea-level data and contemporary VLM observations, listed from north to south: Changzhou (Jiangsu), Wuxi (Jiangsu), Suzhou (Jiangsu), Shanghai, Hangzhou (Zhejiang), Shaoxing (Zhejiang), Ningbo (Zhejiang), Wenzhou (Zhejiang), Fuzhou (Fujian), Xiamen (Fujian), Chaozhou (Guangdong), Shantou (Guangdong), Guangzhou (Guangdong), Foshan (Guangdong), Dongguan (Guangdong), Shenzhen (Guangdong), Zhongshan (Guangdong), Hong Kong, Haikou (Hainan) and Sanya (Hainan). Ocean bathymetry data are sourced from the GEBCO database (http://www.gebco.net/) and background topography is from Natural Earth relief data (https://www.naturalearthdata.com/). b, RSL reconstructions across study sites for early (11700–8200 BP, referenced to 1950 CE), middle (8200–4200 BP) and late (4200 BP to present) Holocene intervals, n = 292. RSL and chronological data presented with 2σ uncertainty intervals. c, Temporal distribution histogram of sea-level proxies. This sea-level database was sourced from previous publications,,,–.
Fig. 2
Fig. 2. Spatiotemporal distribution of Holocene China sea level.
a, Map showing expected linear-term (l(x, t)) contribution to VLM rate during the Holocene (positive indicates uplift and vice versa). Conditioning on the observations reduces the variance by at least 5% to the prior. Some of the main river channels are shown in blue. Note, the bottom-right legend applies for panels c–h. b, Reconstructed Holocene GMSL in comparison with previous studies (ANU, Creel et al., ICE-6G_C (ref. ), Bradley et al. and IPCC AR6 (ref. )). The blue lines indicate the mean from this study, the heavy and light blue shadings represent the 68% and 95% CIs from this study, respectively, and the light purple shading with dotted lines show the 90% CI from ref. . Grey lines represent further tested ice models (excluding ANU and ICE-6G_C) that inform prior GMSL distributions (see Methods). c–h, RSL and sea-level budgets evolution at six selected cities with locations shown in a. Sea-level budgets are separated into GMSL, GRD, linear, regional nonlinear and local nonlinear components (see Methods). Insets are zoomed-in views since 6000 BP. Previous GMSL reconstructions were compiled from refs. ,,,,.
Fig. 3
Fig. 3. Regional nonlinear term contribution to spatiotemporal sea-level changes.
a, RSL change rate associated with the regional nonlinear term (r(x, t)) between 4500 and 4400 BP (2550–2450 BCE). Conditioning on the observations reduces the variance by at least 15% to the prior. b,c, Same as a but for the periods between 2350 and 2250 BP (400–300 BCE) and 1500 and 1400 BP (550–450 CE), respectively. d, Detrended mean RSL change rates with 68% CI near Ningbo in the YRD (green) and Zhaoqing in the PRD (purple). The triangles represent notable palaeoflood events inferred from sediment stratigraphy,, (error bar indicates 95% age CI), with their locations marked in a and b. e,f, RSL change rates owing to ocean dynamic sea-level change, detrended to exclude climate-change signals and highlight climate variability, for the periods 1920–1980 CE (e) and 1950–2010 CE (f), smoothed with a 20-year moving average filter (source data from ref. ).
Fig. 4
Fig. 4. Modern China sea-level budget against its geological baseline.
a, Reconstructed GMSL change rate during the Holocene, with the inset showing a detailed view since 4000 BCE. The dashed line marks the last geological point with a 5% probability of centennial GMSL rising faster than post-1900 rates. Dark shading represents the 68% CI and light shading represents the 95% CI. b, Averaged sea-level budget across 20 main coastal cities (see Fig. 1a) in southeastern China over different periods, with RSL change rates shown in the centre and percentage contributions from each component listed. Geological estimates (left four components) are derived from mean model outputs, whereas instrument-based estimates are sourced from ref. , which do not incorporate local VLM estimates. OD, ocean dynamic effects. The local nonlinear term (cyan) is indiscernible, as it accounts for less than 1% variability in all periods. c, Comparison of geological contributions to VLM (linear kernel + GIA-induced VLM) with modern VLM from GNSS measurements at uplift sites (obtained from the Nevada Geodetic Laboratory at the University of Nevada; see Methods). d, Same as c but for subsidence sites. Error bars represent 1σ reconstruction uncertainty for geological VLM and one standard error for GNSS measurements, with colours indicating GNSS data coverage periods. R2, coefficient of determination weighted by reconstruction uncertainty; P, statistical significance test value, calculated from weighted Pearson correlation test. Note that all three metrics here were calculated by directly comparing geological VLM with GNSS data without any further coefficients.
Extended Data Fig. 1
Extended Data Fig. 1. Spatial distribution of non-China sea-level data used to determine GMSL change.
Triangles represent high-resolution sea-level proxy data, with blue indicating existing data from ref.  and red indicating newly incorporated proxy data. Orange circles mark locations with long-term tide gauge records from the Permanent Service for Mean Sea Level (https://psmsl.org/). Background topography is from Natural Earth Relief data (https://www.naturalearthdata.com/). This sea-level database was sourced from previous publications,,,–.
Extended Data Fig. 2
Extended Data Fig. 2. GMSL change related to spatially uniform covariance functions.
This covariance function (g(t)) primarily reflects contributions from processes that are not explained by GIAg(t), such as centennial GMSL fluctuations (as GIAg(t) typically has a temporal resolution of 500–1,000 years), GMSL variation associated with mountain glaciers and global thermosteric expansion. Dark shading, 68% CI; light shading, 95% CI.
Extended Data Fig. 3
Extended Data Fig. 3. Impact of land–ocean mask misclassification on GRD calculations in the YRD.
a, Modern topography. b,c, Reconstructed palaeotopography at 8500 BP after removing Holocene sediment thickness, with sediment coring locations shown as white dots, which is obtained from ref. . d,e, GRD signals at 8500 BP using modern and palaeotopography, respectively. f, Difference between modern and palaeotopography GRD signals. g, RSL predictions at Taizhou (location shown in d) using both topographies. h, GRD signals for Taizhou predictions. i, GRD signal difference at Taizhou. j, Linear rate of GRD overestimation at each sea-level data location. k,l, Linear term estimates before and after correcting for GRD overestimation.
Extended Data Fig. 4
Extended Data Fig. 4. Reconstructed natural VLM rate and GNSS measurement of VLM.
a, Comparison of natural VLM rate with GNSS data in eastern and southeastern China. Background here is based on OpenStreetMap (www.openstreetmap.org), highlighting the areas of urbanization. b, Same as a but near Hong Kong.
Extended Data Fig. 5
Extended Data Fig. 5. Reconstructed natural VLM rates and a modern VLM product.
a,e Mean (a) and 1σ uncertainty (e) of the natural/geological VLM rate (a combined signal of GIA and the linear component). b,f, Mean (b) and 1σ uncertainty (f) of the GIA-induced VLM rate. c,g, Mean (c) and 1σ uncertainty (g) of the VLM rate associated with the linear component. d,h, Mean (d) and 1σ uncertainty (h) of the modern VLM rate from ref. , which integrates data from tide gauges, satellite altimetry and GNSS measurements.
Extended Data Fig. 6
Extended Data Fig. 6. Posterior GIA model predictions of RSL at various locations outside China.
a, Great Barrier Reef, Australia. b, Mekong River Delta, Vietnam. c, South Sulawesi, Indonesia. d, Singapore. Note that these predictions are weighted only by different ice models, without applying weights to the Earth model, which may vary regionally. All RSL and age data were represented by 2σ uncertainty boxes. Dark shading, 68% CI; light shading, 95% CI.
Extended Data Fig. 7
Extended Data Fig. 7. Tenfold cross-validation results.
a, RSL observations versus predictions by our prior GIA model and hybrid GIA and GP model (GIA + GP). Horizontal and vertical error bars indicate ±2σ observational sea-level uncertainty and model prediction uncertainty, respectively. b, Histogram of weighted squared error between RSL data and model predictions, weighted by data uncertainty.
Extended Data Fig. 8
Extended Data Fig. 8. GIA Earth model calibration results.
Each figure is divided into upper and lower parts by an orange line, with the upper and lower parts indicating lithospheric thickness parameters of 96 km and 71 km, respectively. a, Calibration results for all China sea-level data. b, Calibration results for northern sea-level data between 25 and 35° N. c, Calibration results for southern sea-level data between 15 and 25° N.
Extended Data Fig. 9
Extended Data Fig. 9. Impacts of tidal variation and sterodynamic signals on sea level.
a, Impact of tidal variation on reference level bias in sea-level reconstruction. b, Impact of tidal variation on bias in indicative range uncertainty. c, Sterodynamic sea-level change owing to Qiantang River discharge anomalies under different weather conditions.
See this image and copyright information in PMC

References

    1. Frederikse, T. et al. The causes of sea-level rise since 1900. Nature584, 393–397 (2020). - PubMed
    1. Fox-Kemper, B. et al. in Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change 1211–1362 (Cambridge Univ. Press, 2021).
    1. Wang, H. et al. Regional characteristics of the effects of the El Niño-Southern Oscillation on the sea level in the China Sea. Ocean Dyn.68, 485–495 (2018).
    1. Chen, M., Yan, H., Jin, Y., Wang, T. & Yang, J. Observed sea level changes in the China seas since the 1990s. In Proc. 11th International Conference on Asian and Pacific Coasts 1023–1032 (Springer, 2024).
    1. Horton, B. P. et al. Mapping sea-level change in time, space, and probability. Annu. Rev. Environ. Resour.43, 481–521 (2018).

Publication types

LinkOut - more resources