Quality Assessment of HF Radar–Derived Surface Currents Using Optimal Interpolation

Abstract

This study investigates the applicability of the optimal interpolation (OI) method proposed by Kim et al. for estimating ocean surface currents from high-frequency radar (HFR) in the northeastern Chukchi Sea, where HFR siting is dictated by power availability rather than optimal locations. Although the OI technique improves data coverage when compared to the conventional unweighted least squares fit (UWLS) method, biased solutions can emerge. The quality of the HFR velocity estimates derived by OI is controlled by three factors: 1) the number of available incorporating radials (AR), 2) the ratio of the incorporating radials from multiple contributing site locations [ratio of overlapping radial velocities (ROR) or radar geometry], and 3) the positive definiteness [condition number (CN)] of the correlation matrix. Operationally, ROR does not require knowledge of the angle covariance matrix used to compute the geometric dilution of precision (GDOP) in the UWLS method and can be computed before site selection to optimize coverage or after data processing to assess data quality when applying the OI method. The Kim et al. method is extended to examine sensitivities to data gaps in the radial distribution and the effects on OI estimates.

Surface Current Patterns in the Northeastern Chukchi Sea and Their Response to Wind Forcing

“…One‐dimensional radial surface currents, , from each HFR were used to estimate two‐dimensional surface currents, , following the optimal interpolation (OI) scheme of Kim et al (). Fang et al () investigated the performance of the OI method applied to the Chukchi HFR network. They used analytical stream functions to simulate spatially and temporally varying currents and found that for derived with a normalized skill 0.7 (0–1 scale), errors were ∼2° angle shift and <0.1% magnitude variation.…”

“…The were computed every cardinal hour, and the ROR at each grid point was determined. Fang et al () show that areas with 0.7 skill closely coincide to those with ROR < ∼5, thus grid points with ROR 5 were discarded. In addition, if the number of contributing in the search radius (35 km) of the grid point was <20, the OI‐output at the grid point was flagged.…”

“…If the angle between paired was not between 30° and 120°, the grid point was excluded from the analysis. This range was chosen in conjunction with other quality control factors reported in Fang et al () to allow optimal determination of .…”

Application of Empirical Orthogonal Function Interpolation to Reconstruct Hourly Fine Particulate Matter Concentration Data in Tianjin, China

“…e TPS method does not need to set parameters, while other RBF needs to set parameters [23]. Some statistical-based methods (e.g., Kriging interpolation, optimal interpolation (OI), and Kalman filter) are conventional and classical methods in geoscience [12,13,[24][25][26][27].…”

A Multifaceted Approach to Advance Oil Spill Modeling and Physical Oceanographic Research at the United States Bureau of Ocean Energy Management

“…This research effort identified the Chukchi Slope current-a subsurface current transporting Pacific-origin water westward over the upper continental slope-and quantified its seasonality [142,146,147]. Major surface current patterns in the northeastern Chukchi, and the wind conditions under which they occur, were derived and mapped [141,143]. The researchers also calculated the seasonal transport (and variations) at the head of Barrow Canyon [145].…”