[Paper Review] A Reanalysis of the October 2016 "Meteotsunami" in British Columbia with Help of High-Frequency Radars and Autoregressive Modeling

Baptiste Domps, Julien Marmain+1 more|arXiv (Cornell University)|Nov 14, 2020

earthquake and tectonic studies20 references18 citations

TL;DR

This study reanalyzes the October 2016 'meteotsunami' in British Columbia using high-frequency radar (HFR) data and a novel autoregressive modeling approach with maximum entropy method (MEM) to improve short-integration-time current estimation. The analysis reveals a sharp surface current front linked to an atmospheric front, disqualifying Proudman resonance and supporting a storm surge rather than a true meteotsunami as the cause of the event.

ABSTRACT

On October 14th, 2016, the station of Tofino (British Columbia, Canada) issued the first ever real-time tsunami alert triggered by a coastal High-Frequency Radar system, based on the identification of abnormal surface current patterns. The detection occurred in the absence of any reported seismic event but coincided with a strong atmospheric perturbation, which qualified the event as meteo-tsunami. We re-analyze this case in the light of a new radar signal processing method which was designed recently for inverting fast-varying sea surface currents from the complex voltage time series received on the antennas. This method, based on an Auto-regressive modeling combined with a Maximum Entropy Method, yields a dramatic improvement in both the Signal-to-Noise Ratio and the quality of the surface current estimation for very short integration time. This makes it possible to evidence the propagation of a sharp wave front of surface current during the event and to map its magnitude and arrival time over the radar coverage. We show that the amplitude and speed of the inferred residual current do not comply with a Proudman resonance mechanism but are consistent with the propagation of a low-pressure atmospheric front and wind vectors as revealed by satellite imagery. This indicates that the event that triggered a tsunami alert is more likely a storm surge than a true meteo-tsunami. Beyond this specific case, another outcome of the analysis is the promising use of oceanographic radars as proxy's for the characterization of atmospheric fronts.

Motivation & Objective

• To re-express the October 2016 coastal anomaly in British Columbia using advanced HFR signal processing.
• To resolve the ambiguity in the physical mechanism behind the tsunami-like event—whether Proudman resonance or storm surge.
• To validate the use of high-frequency radar as a real-time proxy for atmospheric front detection.
• To improve short-timescale ocean current estimation for early tsunami warning applications.
• To integrate HFR data with satellite and meteorological observations for multi-source event characterization.

Proposed method

• Applied a time-varying autoregressive (AR) model to HFR voltage time series to estimate surface currents at very short integration times.
• Combined AR modeling with the maximum entropy method (MEM) to enhance signal-to-noise ratio and spectral resolution.
• Used change point detection on residual currents to precisely identify the arrival time and amplitude of the current front.
• Mapped the propagation of the current front across the radar coverage using arrival time differences.
• Correlated radar-derived current front speed with atmospheric front speed from GOES-15 satellite imagery and GFS model data.
• Evaluated Froude number conditions to test for Proudman resonance compatibility, using long-wave celerity and front speed comparisons.

Experimental results

Research questions

• RQ1 What physical mechanism underlies the October 2016 'meteotsunami' event in British Columbia—Proudman resonance or storm surge?
• RQ2 Can high-frequency radar with advanced signal processing detect and characterize rapid atmospheric front impacts on ocean surface currents?
• RQ3 How does the speed and amplitude of the inferred surface current front compare with long-wave celerity and atmospheric front speed?
• RQ4 To what extent does the observed current front align with meteorological conditions such as wind shear and pressure drops?
• RQ5 Can HFR data serve as a reliable proxy for detecting and tracking atmospheric fronts in real time?

Key findings

• The HFR-derived current front propagated at approximately 65 km/h, significantly slower than the long-wave celerity of ~100 km/h, disqualifying Proudman resonance.
• The atmospheric front speed, estimated at ~75 km/h from GOES-15 and GFS data, closely matched the radar-inferred front speed, supporting a direct atmospheric forcing mechanism.
• The Froude number (Fr ≈ 0.87) fell outside the 'tsunamigenic' range (0.9 < Fr < 1.1), further rejecting Proudman resonance as the origin.
• A pressure drop of ~25 hPa, consistent with a 25 cm sea-level rise, was recorded by NOAA buoys and aligns with the observed 20 cm tide gauge oscillation.
• The current front's amplitude and timing are best explained by wind stress and Stokes drift in the surface layer, not by resonant wave amplification.
• The joint analysis confirms the event was a storm surge, not a true meteotsunami, highlighting HFR's value in distinguishing atmospheric from seismic tsunami signals.