Processing-induced uncertainty in MEMS-derived significant wave height estimates without attitude measurements

Introduction Significant wave height (SWH) derived from compact microelectromechanical systems (MEMS) is increasingly used for operational wave monitoring, but estimates can vary depending on the scientifically defensible processing choices applied to the same inertial record. Methods This study evaluates method-selection variability using a 31-day acceleration record from one operational buoy and one MEMS configuration for which attitude observations were unavailable. A common tilted-axis acceleration record was conditioned and processed through controlled spectral- and time-domain variants. Sensitivity was evaluated relative to a designated reference workflow, and the empirical between-method dispersion of a prescribed spectral ensemble was calculated for each observation epoch. The manufacturer’s time-aligned SWH product and a regional NDBC buoy were used only for contextual comparison, not as ground truth or for formal validation. Results For the investigated record, the prescribed ensemble produced a mean processing spread of 0.023 m, a median of 0.018 m, and a 95th percentile of 0.060 m; the corresponding 95th percentile relative spread was 6.42%. Among the tested configurations, modifications to time-domain stabilization produced larger reference-relative differences than substitution among the evaluated spectral estimators after common acceleration conditioning. Discussion These statistics quantify only the processing-method component of the prescribed ensemble and do not constitute a complete measurement-uncertainty budget. The numerical findings are specific to the investigated dataset, sensor configuration, processing choices, and sea states.