Data-Driven Noise Estimation for Individual High-Frequency Radar Stations

Abstract

High-Frequency Radars (HFRs) provide continuous coastal current measurements, their reliability depends on proper quantification of measurement noise. In this work, we introduce an automated method to estimate noise levels in individual HFR stations offering a consistent and data-driven alternative to traditional noise estimation approaches. The approach adapts the k-sigma clipping technique to HFR radial velocity measurements, enabling the estimation of the noise standard deviation without requiring the use of external data or additional stations. The method provides daily noise estimates and, when data quality allows, it can be extended to instantaneous measurements. Results demonstrate that the estimated noise levels reflect both environmental variability and station performance. The proposed method is computationally efficient, easy to implement, and enhances the capacity for continuous monitoring and quality assessment of HFR systems.