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. ⁹, 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.