Tide gauges within half a rupture length showed steps at 6 of 21 M7+ earthquake pairs; Tohoku 2011 subsidence reversed by about half in 10 years at a steady rate; other cases were mixed

Question: how often do tide gauges record a relative sea-level (RSL) step after large shallow earthquakes, and does the step reverse later? Approach: we paired PSMSL RLR monthly records (1,644 gauges) with USGS ComCat earthquakes of M 7 or more and depth of 70 km or less (1900-2025). For each pair we fitted a 15-year pre-event trend and seasonal model. The step S is the mean residual in post-event months 1-12. The later change is the residual in years 2-10 minus S. We compared each value with up to 40 placebo dates at the same gauge, with the same windows and rules. Results: within 0.5 rupture lengths (L, Blaser et al. 2010) of the epicentre, 6 of 21 usable pairs had a step with |z| of 3 or more (placebo rate 0.4%). At 1-5 L the rates were 2-4%, near the placebo rates. Tohoku 2011 subsidence reversed by 53% at Ofunato II (+770 mm step) and by 48% at Onahama (+508 mm) after 10 years, above all placebo values. At Ofunato the reversal rate was 45 mm/yr in years 2-6 and also in years 6-10. At Nezugaseki after Niigata 1964, RSL continued to rise by a further 50% of the step (p = 0.025). Adak after 1957 reversed by 57% (p = 0.038), but not with a 5-year pre-event window. Of 9 detected pairs (5 events), 3 reversed and 1 continued at p of 0.05 or less. Thus we found no general reversal fraction. Limitations: few cases (4 of 9 from Tohoku), epicentral distance instead of slip models, no ocean reference for great earthquakes, and a smallest placebo p-value of 0.024. This is an independent replication and extension of a sibling campaign study with a different design.

tide gauge · relative sea level · coseismic step · postseismic deformation · vertical land motion · PSMSL · ComCat · Tohoku 2011 · Niigata 1964 · placebo test · earthquake cycle · replication