Reducing Wave-Induced Microwave Water-Level Measurement Error with a Least Squares–Designed Digital Filter

Abstract

microwave water-level sensor, the Design Analysis model H-3611i, will soon enter service at tide stations operated by the National Oceanic and Atmospheric Administration?s Center for Operational Oceanographic Products and Services (CO-OPS) as part of the National Water Level Observation Network. CO-OPS tests include a multisensor deployment at the U.S. Army Corps of Engineers Field Research Facility at Duck, North Carolina, to evaluate microwave water-level measurement error over a wide range of Atlantic Ocean sea states. In situ precision and accuracy of processed (6-min average) water level is found to depend on sea state in addition to data processing methods and sensor operating mode. Estimates over selected 6-h measurement periods show that a degree-two polynomial successfully models the increase in sensor standard error with increasing zero-moment (Hm0) wave height but with differences in rate of error increase dependent on the application of a prefilter and choice of sensor operating mode. Prefiltering of 1-Hz ?fast mode? sensor output to remove variance at selected wind-wave frequencies can reduce standard error during extreme conditions (Hm0 ≈ 3 m) from approximately ±3 cm without prefiltering to about ±1 cm using a least squares?designed (LSD) digital filter with a 60-s cutoff period. When wave heights are elevated, skewed non-Gaussian distributions develop within the 1-Hz (360 s) sample domain wherein a 3σ outlier elimination process applied without prefiltering can introduce a negative bias of up to 5 cm in individual 6-min water-level averages.