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 (ANU²⁰, Creel et al.¹², ICE-6G_C (ref. ¹⁹), Bradley et al.²⁶ and IPCC AR6 (ref. ²)). 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. ¹². 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. ^{2,12,19,20,26}.