Abstract
Tidal data have been deemed to serve many important purposes, including establishing maritime boundaries, dredging sites, and navigation. Recently, even some building codes are dependent upon these data. We explore the efficacy of using harmonic analysis to model changes in water levels both in ocean and in protected sea environments. Typically, these data are correlated and have 'not-completely at random' missing observations. Our new analyses account for the presence of autocorrelation and missingness.