Estimating ocean wave heights in Northeast Atlantic using land-based microseism data

Abstract

Various techniques are employed to monitor ocean Significant Wave Heights (SWH), such as insitu buoys, numerical ocean wave modeling, and satellite altimetry. Each method comes with its set of advantages and limitations in terms of spatial and temporal resolution. Buoys offer superior temporal resolution but are characterized by lower spatial resolution compared to numerical wave modeling and satellite altimetry. This study involves a comparative analysis of estimated wave heights from numerical ocean wave modeling using blown wind (HINDCAST) with measured Buoy data. Subsequently, these datasets (e.g. both buoy and HINDCAST data) are the utilized as inputs for our new approach to estimate SWH, employing an Artificial Neural Network (ANN) applied to terrestrial (microseism) data. This approach offers a combination of both spatial and temporal resolution of buoys and HINDCAST data, facilitating quasi-real-time monitoring under specific conditions. The ANN is trained using seismic amplitude of Irish National Seismic Network, buoy measurements, and HINDCAST data to estimate SWH at specific locations in the Northeast Atlantic. Preliminary comparisons between the ANN results and Buoy/HINDCAST data reveal a robust relationship between secondary microseism amplitudes recorded on land, and ocean seismic wave heights. The quality of the results varies with location in the ocean. Particularly for small and moderate wave heights, the estimated wave heights exhibit minimal discrepancies between the ANN and actual buoy data or between the ANN and HINDCAST ocean wave data. This promising finding underscores the potential accuracy of the ANN model in estimating wave heights from land based seismic data, under varying conditions.