Correction to "North Adriatic tides: observations, variational data assimilation modeling, and linear tide dynamics".

A precision/round-off error has been discovered in the tidal analysis routines used in the paper "North Adriatic tides: observations, variational data assimilation modeling, and linear tide dynamics" by J. W. Book, H. Perkins, and M. Wimbush (2009, Geofizika, 26, 115-143). Tidal elevation phases for 12 of the 15 stations are, on average, too low by 3.9° for the diurnal constituents and 7.9° for the semidiurnal constituents in Tabs. 4 and 5. These tables have been corrected and are republished here. The error also had an effect on the input data used for the linear variational data assimilation model, and combined with a nearest neighbor interpolation scheme produced an approximate 15 minute forward shift in time for 6 of the 43 synthesized tidal records. The error produced final model solutions that had tidal elevation phases 3.5° too high for M2, 1.6° too high for K1, and similar matching phase shifts for other semidiurnal and diurnal constituents. The errors in the input data have been corrected, the interpolation scheme has been changed to a piecewise cubic spline method, and the model runs have all been redone. The new model results suggest a minor change in optimal friction parameter, which in turn alters model Q factors and dissipation. However, the original finding that these values are not well determined by this methodology remains true. Model and observational results originally shown in Figs. 5-7 and Figs. 13-14 have slightly changed and are republished here. The main conclusions from the original work regarding Kelvin waves and TRW dynamics for the North Adriatic basin remain unaltered by these corrections. [ABSTRACT FROM AUTHOR]