Uncertainty in wave hindcasts in the North Atlantic Ocean

TL;DR

This study evaluates the uncertainty of numerical wave hindcast models in the North Atlantic, comparing them with buoy data, and highlights the need for model improvements to better predict extreme wave conditions for ship design.

Contribution

It quantifies model uncertainty in wave hindcasts by comparing four models with buoy data, especially under extreme conditions, providing insights for ship design applications.

Findings

Models show good accuracy for significant wave height but deviate during extreme conditions.

Underestimation of mean wave period $T_{m02}$ occurs in extreme waves.

Disagreement among models and buoys increases under extreme wave conditions.

Abstract

Accurate ocean surface wave knowledge is crucial for ship design. With the significant advancements of model physics and numerical resources, the recent numerical wave hindcast data has a potential to provide environmental conditions for wave load estimation in the ship design process. This study aims to quantify model uncertainty in the state of art numerical wave hindcast products to get insight into the application of wave hindcast for ship design. The model uncertainty is deduced based on the comparison with the wave buoys in the North Atlantic as well as the inter-model comparison with four wave model products. The multiple wave buoys distributed over the Northwest Atlantic and Northeast Atlantic are considered as wave buoy arrays and used for model evaluations. All the four models in general showed very good and similar accuracy in the estimation of the significant wave height $H_s$. The model $H_s$, however, deviates from buoys and also deviate among each other for the extreme wave conditions of $H_s$>10. It is also found that there is disagreement for the mean wave period $T_{m02}$ between numerical wave models and the wave buoys. More specifically, the models underestimate $T_{m02}$ for extreme wave conditions of $H_s$>10 in the Northeast Atlantic. The joint probability density function (JPD) of the significant wave height $H_s$ and mean wave period $T_{m02}$ for the extreme wave conditions are also derived. Among the JPDs, the total number of samples that satisfies the extreme wave conditions of $H_s$>10 and 8<$T_{m02}$<14 varied by 50% although the mean value of the four model products is close to the observation. This study indicates further model improvement is necessary to more accurately reproduce the extreme conditions. For the model improvement as well as the model evaluation, more measurements of the high sea states and their wave spectra information are warranted.