Error analysis of the directional wave spectra obtained by the NDBC 3-m pitch-roll discus buoy

The National Data Buoy Center (NDBC) deploys a series of 3-m-diameter discus directional wave buoys around the coastline of the United States and its territories. These buoys make available a continuous set of hourly wind and directional wave data. This study estimates the accuracy of the significant sea height, mean sea direction, and the full directional wave spectrum, which are the major wave data parameters commonly derived from the NDBC buoy measurements. The errors are examined both on the basis of the uncertainty inherent in deriving results from the measurement of a random sea and through a statistical examination of a select section of data from a specific buoy. The errors in the derived parameters are found to be quite small, less than 3% in the sea height and usually less than 4/spl deg/ in the peak wave direction. However, errors in the directional wave spectra are found to be potentially significant. These errors are associated with the buoys only measuring sufficient data for the first two terms of the Fourier expansion of the directional spectrum. This measurement limitation is due to the operational requirements of the buoys and not due to any defects or design flaws. This limitation is examined in terms of its effect on the accuracy of the derived wave spectra. A method is developed in this study which allows for a maximum error estimate to be calculated for each hourly record of the buoy data with the goal of guiding the selection of those records which meet the accuracy requirements of a particular application utilizing the data. A select set of NDBC data, consisting of 1369 hourly records collected by Buoy 44 014 during August and September 1993, are analyzed to provide an overview of the size and variety of this error source of the measured directional spectrum. It is found that approximately 60% of the sample wave records give an upper bound rms error of less than 30% of the rms energy of the spectrum, which may be sufficiently accurate for many research requirements.