Variability and predictability of the North Atlantic wave climate

Abstract

[1] Wave climate across the ocean basins can be described using satellite altimetry; here, we concentrate on the North Atlantic region. Waves in the North Atlantic are strongly seasonal and peak in the winter season. The northeastern sector of the Atlantic and adjoining shelf seas also exhibit exceptionally high interannual variability in the winter, with monthly average significant wave height varying by up to a factor of 2 from one year to the next. The strength and geographical distribution of variability is broadly consistent throughout the winter months (December-March). A large fraction of these wave height anomalies is associated with a single pattern of pressure anomalies that resembles the North Atlantic Oscillation (NAO). A predictor based on NAO dependence is ''trained'' from relatively recent satellite data and then tested against earlier satellite and in situ data. The predictor is successful in large areas of the North Atlantic, confirming a robust relationship between wave height anomalies and the NAO over the last few decades. A substantial rise (up to 0.6 m) in monthly mean wave heights on the northeastern Atlantic during the latter part of the twentieth century is attributable to changes in the NAO. Substantial residual anomalies in wave heights exist after the influence of the NAO has been subtracted; these are partly explained by a second pair of North Atlantic patterns in wave height anomalies and sea level pressure anomalies. This ''East Atlantic'' pattern is particularly influential in midwinter and affects the southern part of the northeastern sector (including the region of Seven Stones Light Vessel).

Variability and trends in the directional wave climate of the Southern Hemisphere

“…Cotton and Carter (1994) established that only five valid calibrated satellite passes over a region of the ocean within a month were sufficient to obtain a reasonable accurate and unbiased estimate of the mean wave climatology in that month. Woolf et al (2002) indicate that this condition was generally satisfied each month by each satellite for all 2 × 2 grid squares within its orbit.…”

“…Woolf et al (2002) indicate that a reliable, satellitederived wave climatology requires calibration of the altimeter significant wave heights using reliable in situ measurements, data screening to remove erroneous data, and attention to the sparse-sampling problem. In this study, we have used the same methods described by Woolf et al (2002) to determine their satellite wave climatology, but the extra years and additional satellite missions extend this dataset. Owing to the problems previously mentioned with Geosat data, and their being no additional missions for this early period, the pre-1990 satellite wave climatology has been taken from the Woolf et al study.…”

“…The predominant focus of such wave studies has focussed on high shipping regions of the North Atlantic (e.g. Bacon and Carter, 1991;Swail and Cox, 2000;Woolf et al, 2002;Wolf and Woolf, 2006), and the Northern Hemisphere as a whole (e.g. Graham, 2003;Wang and Swail, 2001;Sasaki et al, 2005).…”

“…A strong relationship between the Northern Annular Mode, or North Atlantic Oscillation, and the wave climate in the North Atlantic Ocean was described by Woolf et al (2002), but little or no attention has been paid to effects of the SAM on the waves in the SH, and, in particular, the Southern Ocean. However, the variability of the SH wave climate has important implications for the global oceans.…”

“…Table I summarizes the calibration equations and the periods for which each satellite provides a validated climatology (Note that Woolf et al (2002) Sporadic gaps in the climatology occur for a few grid squares in occasional months due to the sparse sampling by the altimeter. In this study, as for the Woolf et al (2002) study, we are interested in the largescale variability of the surface ocean wave climate, and the analysis requires data for every grid square for all months. Therefore, data were interpolated and smoothed using a 5 cell × 5 cell digital filter approximating Gaussian smoothing with a radius (standard deviation) of 2.8°.…”