Error Analysis Procedures Used by the National Ocean Service to Compute Estimated Error Bounds for Tidal Datums in the Arctic Ocean.

Abstract

NOAA has an established National Water Level Observation Network (NWLON) along all U.S. coastlines. One purpose of the NWLON is to provide control for tidal datum determination at short-term water level stations installed for hydrographic and shoreline mapping surveys. There are significant gaps in NWLON coverage in Alaska. When short-term (1-12 months) water level stations are installed outside of an NWLON coverage area, a First Reduction (FRED) or arithmetic mean is used for datum determination instead of the preferred simultaneous comparison method that uses a nearby NWLON station to compute a 19-year equivalent National Tidal Datum Epoch (NTDE) datum. The datum error of a FRED is typically greater than the error computed by a simultaneous comparison procedure with an NWLON station. This report describes one method used by NOAA to establish error bounds on FRED tidal datums computed at short-term stations. The standard deviation of monthly Mean Tide level (MTL) at 29 operating and historical water level stations in Alaska with varying time series lengths was used to infer FRED datum errors within the study region. The combined results show that FRED datum errors decrease from 0.120 m, to 0.040 m and 0.008m (one-sigma) for 1, 12, and 228 month time series, respectively. Comparisons across the region show only minor statistical differences, supporting the use of combined values as representing FRED datum errors for the entire study area. These results will help facilitate better estimates of total tide-propagated error and better planning of required subordinate installation time series length in support of hydrographic and shoreline mapping surveys in Alaska.