Increasing wave heights in the Gulf of Mexico driven by swell waves

George, Jesbin; Hiatt, Matthew; Willson, Clinton S.

doi:10.3389/fmars.2025.1720341

ORIGINAL RESEARCH article

Front. Mar. Sci., 12 January 2026

Sec. Physical Oceanography

Volume 12 - 2025 | https://doi.org/10.3389/fmars.2025.1720341

Increasing wave heights in the Gulf of Mexico driven by swell waves

Jesbin George^1*

Matthew Hiatt¹

Clinton S. Willson²

¹Department of Oceanography and Coastal Science & Coastal Studies Institute, Louisiana State University, Baton Rouge, LA, United States
²College of the Coast and Environment & Coastal Studies Institute, Louisiana State University, Baton Rouge, LA, United States

This study analyzes the climatology, spatial and temporal trends of wind speeds and significant wave height (H_s) of combined wind-seas and swells and H_s of separated swells and wind-seas on annual and seasonal scales in the Gulf of Mexico (GoM) using 50 years of atmospheric reanalysis data. The H_s data showed strong agreement with in-situ measurements, with minimal bias (+/-< 0.08 m) and high correlation (r > 0.9) across both deep and shallow regions, although a slight underestimation is noted for wind speed data. Wind speeds showed seasonal variability in the GoM, though they did not exhibit significant long-term increases, except for a notable decrease (~ -1.0 cm/s/yr) along the southern Florida region. Conversely, the significant wave height of combined wind-seas and swells (H_s,c) exhibits a significant increase, particularly in the western and central GoM (> 0.2 cm/yr). Seasonally, H_s,c peaks in winter and is lowest in summer, with notable increases across most parts of the GoM. The increase in H_s of swell waves closely mirrors that of H_s,c.. In contrast, H_s of wind-seas is primarily driven by local winds, does not significantly increase over time, instead showing a strong seasonal coupling with wind speed and its trends across all seasons, suggesting that swell waves are likely responsible for the increase in H_s,c. The study highlights that the significant increase in H_s,c and H_s of swell waves along the GoM are primarily driven by remote swell waves from the Caribbean Sea and westward-propagating waves generated within the GoM, underscoring the growing importance of swell waves in shaping the wave climate and coastal dynamics.

1 Introduction

Surface gravity waves (wind-waves, hereinafter waves) play a pivotal role in various engineering and environmental issues, both in the open ocean and nearshore zones, impacting ecosystems and communities globally. In the open ocean, waves can threaten the safety of offshore operations, marine structures, pipelines, mooring systems, renewable energy installations, maritime navigation (Nerzic and Mazé, 2013; Tsinker, 2004). In nearshore zones, wave-driven processes lead to coastal flooding and erosion (Castelle and Masselink, 2023; Toffoli and Bitner-Gregersen, 2017), directly affecting communities and coastal activities. Waves also regulate the exchange of momentum, heat, and mass across the air–sea interface (Villas Bôas et al., 2019; Young et al., 2011), thereby influencing global and regional climate systems (Hemer et al., 2012; Young and Ribal, 2019). Recent studies indicate increasing global wind speeds and wave heights associated with warming oceans (Kaur et al., 2021; Reguero et al., 2019; Young and Ribal, 2019). Even small long-term increases in wave heights can intensify extreme sea levels and coastal flooding (Tebaldi et al., 2021; Storlazzi et al., 2018), with broader implications for marine ecosystems and offshore operations (Cavaleri et al., 2012; Hemer et al., 2010). Understanding the long-term variability of ocean wind and wave climate is therefore essential for anticipating future risks and guiding coastal adaptation strategies.

Wave information can be obtained from in situ measurements, satellite altimetry, or model products, each with distinct advantages and disadvantages. In situ data is reliable but sparse; satellite altimetry offers wide spatial coverage but lower temporal resolution and requires careful calibration; model products provide large coverage but require regional validation against observations (Muhammed Naseef and Sanil Kumar, 2020; Sepulveda et al., 2015). Several global and regional scale climatologies have been developed using visual observations (Gulev et al., 2003; Gulev and Grigorieva, 2006), satellite remote sensing (Kumar et al., 2013; Young, 1999), and hind cast wave models (Appendini et al., 2014; Fan et al., 2012, 2014; Liang et al., 2019; Shanas et al., 2017; Sterl and Caires, 2005), with long-term trends also assessed using remotely sensed and hindcast models (e.g., Appendini et al., 2014; Gupta et al., 2015; Izaguirre et al., 2011; Young et al., 2011). Advances in data assimilation have enabled modern reanalysis that integrate in situ and satellite measurements, providing consistent long-term datasets for climate-scale analyses. ERA5, the fifth-generation reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF), offers enhanced spatial resolution, model physics, and extensive atmospheric and oceanic variables (Hersbach et al., 2020), and is widely used in global and regional studies of wave climatology and trends (Barbariol et al., 2021; George and Kumar, 2020, 2021; Semedo et al., 2011; Sreelakshmi and Bhaskaran, 2020; Stefanakos, 2021; Thomas et al., 2025).

The Gulf of Mexico, (GoM), a semi-enclosed marginal sea spanning ~1.6 million km², is bordered by the United States to the north and east, Mexico to the west and south, and Cuba to the southeast (Figure 1) serving as a vital source of natural resource for all three nations. The GoM is connected to the Atlantic Ocean by the Straits of Florida and to the Caribbean Sea (CS) by the Yucatan Channel (Figure 1). The GoM features a continental shelf of ~180 m in depth, covering nearly 38% of the entire gulf area (Guiberteau et al., 2012), and is relatively shallow with an average overall depth of ~1560 m. The GoM is also characterized as microtidal, with tide ranges less than 1 m, and the dominant tidal constituents vary throughout the basin (Stumpf and Haines, 1998). The major current surface in the GoM is the Loop Current, an integral part of the Gulf Stream, which enters the GoM through the Yucatan Channel. It has been previously reported that mean significant wave height (H_s) in the GoM is modulated by these currents by as much as 0.18 m and can exceed 1 m during storm conditions (Abolfazli et al., 2020). The waves in GoM are highly affected by frequent North Atlantic hurricanes and atmospheric fronts, with the former producing more energetic and devastating waves, and the latter generating more frequent waves with widespread regional impacts (Appendini et al., 2018; Guo et al., 2020). It has been reported that the H_s reached as high as 16.91 m during Hurricane Katrina in 2005 (Hu and Chen, 2011), and there is a rapid increase in H_s along the north GoM during cold front events (Guo et al., 2020). Appendini et al. (2014) demonstrated that cold front intrusions significantly influence the winter wave climate in the GoM, increasing the mean H_s and dominating the extreme wave conditions.

Figure 1

Map of the Gulf of Mexico depicting topography and monitoring stations. Red triangles show NDBC stations, yellow circles are reference locations, and blue circles indicate auxiliary locations. Depths are color-coded, ranging from five to six thousand two hundred fifty meters. Key locations include the Yucatan Channel and the Straits of Florida. Countries labeled are the United States, Mexico, and Cuba.

Figure 1. Study area across the Gulf of Mexico (GoM). Red triangles represent NDBC buoy locations used for comparing ERA5 wind speed and wave characteristics, yellow dots indicate reference locations used for analyzing distribution of peak wave periods and blue dots indicate selected auxiliary locations. Map is generated using Ocean Data View software (Schlitzer, 2015).

Previous studies of wave characteristics in the GoM have addressed several distinct aspects. Many works have focused on hurricane-generated waves, using numerical models to assess directional spectra (Hu and Chen, 2011), evaluate wind products and model configurations (Razavi Arab et al., 2024), and simulate storm surge, waves, and coastal inundation (Sheng et al., 2010). Other studies have emphasized the role of atmospheric cold fronts, which strongly influence the seasonal and interannual wave climate in the northern Gulf (Appendini et al., 2014; Guo et al., 2020). In addition, in situ observational studies have provided site-specific insights into wave variability on the shelf and shallow regions (Georgiou et al., 2005; Merrill et al., 2024). Abolfazli et al. (2020) conducted a decadal coupled modeling study to examine swell fractions, wave age, interannual variability, and the effects of currents on waves in the GoM. While these studies have advanced understanding of wave dynamics in the region, a comprehensive assessment of long-term changes in winds and waves requires multi decadal datasets. The Intergovernmental Panel on Climate Change (IPCC, 2013) recommends a minimum 30-year record to evaluate long term changes and their impacts, underscoring the need for extended analysis in the GoM.

Even though the GoM experiences highly variable climate conditions due to the influence of cyclones and atmospheric fronts, most climatology and trend studies in the region have been conducted as part of broader global analyses, which often overlook regional scale processes and impacts (Fan et al., 2012, 2014; Liang et al., 2019; Semedo et al., 2011; Young and Ribal, 2019; Zheng et al., 2022). Several recent efforts, however, have focused more directly on the GoM. Jamous and Marsooli (2023) used multidecadal buoy observations to assess mean and extreme wave climate along the U.S. East, Gulf, and West coasts, providing valuable long term in situ perspectives; their analysis indicates increasing H_s trends in central GoM, while other regions exhibited negative or insignificant trends. Appendini et al. (2014) examined wave climatology and trends in H_s and extreme wave heights (99th percentile) using a 30-year hindcast, while Appendini et al. (2018) analyzed the influence of cold fronts and projected changes under future emission scenarios, including implications for wave power. Most recently, Appendini et al. (2025) applied the COWCLIP framework with a regional climate model to assess the potential impacts of climate change on GoM wave climate. Collectively, these studies point to an overall increase in mean and extreme H_s across the GoM; however, they primarily analyzed the combined H_s and did not separate wind-sea and swell contributions and evaluate their spatial distribution or multi-decadal trends.

In semi-enclosed marginal seas, including the GoM, mesoscale and localized influences such as atmospheric fronts, coastal winds, land-sea breezes, low-level jets, currents, limited fetches, bottom topography and shoaling can modulate H_s (Abolfazli et al., 2020; Aboobacker et al., 2011; Appendini et al., 2014, 2018; Guo et al., 2020; Qian et al., 2020; Sanil Kumar and George, 2016). These factors result in distinct wind-sea and swell climates in regional and semi-enclosed marginal seas compared to the open ocean (Semedo et al., 2015). Even when two locations share similar bulk wave parameters (H_s or mean wave period), the dominant wave systems can differ markedly (Holthuijsen, 2007; Semedo et al., 2015). Thus, analyzing only combined H_s can obscure the underlying dynamics. For e.g., an increase in swell energy may offset a decrease in wind-sea energy, leading to a misleading interpretation of long-term trends in the combined H_s.

While previous studies have significantly advanced understanding of the GoM wave climate, they have been limited in scope with respect to separated wave components. Global ERA5-based assessments, such as Zheng et al. (2022), quantified large-scale swell and wind-sea variability but did not provide basin-specific climatology or explore the regional dynamics within the GoM. Recognizing that the GoM is influenced year-round by both locally generated wind seas and remotely propagating swells (Fan et al., 2014), the present study provides a synoptic scale, basin-wide assessment of winds and waves using ERA5 reanalysis data from 1974 to 2023, with a particular focus on climatology and trends of swell and wind-sea components on annual and seasonal scales. This distinction is critical because swell and wind-seas are driven by different forcing mechanisms and have different implications for coastal dynamics, offshore operations, and hazard assessment. A 50-year period was chosen to ensure statistically robust multidecadal trend estimates and to extend earlier GoM wave-climate studies, which typically analyzed only ~30 years of data (e.g., 1979–2008; Appendini et al., 2014). Using the full 1974–2023 ERA5 record allows us to assess long-term trends with greater confidence and aligns with the temporal coverage commonly used in historical wave-trend analyses. To the best of our knowledge, this is the first multi-decadal investigation in the GoM that explicitly analyzes both combined H_s and the separated swell and wind-sea components, providing a distinct contribution beyond prior studies limited to only combined H_s.

2 Data and methodology

The total H_s of combined wind-sea and swell waves (H_s,c), H_s of swell waves (H_s,s), H_s of wind-seas (H_s,w), U,V components of surface winds (10m wind), Peak wave period (T_p) and wave direction (DIR) were obtained at 1-hour intervals for the GoM over a 50-year period (1974-2023) from the ERA5 global atmospheric reanalysis produced by ECMWF (Hersbach et al., 2020). The atmospheric variables have a spatial resolution of 0.25° x 0.25°, while the wave model, based on the WAM model, has a spatial resolution of 0.5° x 0.5°. ERA5 employs a spectral partitioning scheme to separate the 2D wave spectrum into wind-sea and multiple swell systems. In the ERA5 WAM model, wind-sea refers to those spectral components that remain actively forced by the local wind, whereas swell denoted the remaining wave components that have propagated away from their generation region and are no longer wind-driven (ECMWF, 2016). Wind sea components are identified using a wave-age based forcing criterion (Equation 1), and a component is considered wind-sea when,

\begin{array}{l} 1.2 \times 28 (\frac{u *}{c (f)}) \cos (θ - \emptyset) > 1 & (1) \end{array}

where u_* is the friction velocity derived from surface atmospheric stress, c (f) is the phase speed as derived from the linear theory of waves and $\emptyset$ is the wind direction. Spectral components that do not satisfy this condition are classified as swell. The integral parameters (i.e., H_s, and DIR) are then computed separately for the wind-sea and total swell energy (ECMWF, 2016). ERA5 WAM model further adapted the Hanson and Phillips (2001) spectral partitioning algorithm to decompose the swell spectrum into up to three distinct swell systems. In this study, we did not apply any additional thresholds or user-defined criteria; instead, we directly used the wind-sea and swell partitioned outputs provided by ERA5.

The climatology of winds, H_s,c, H_s,s, and H_s,w along the GoM was determined based on seasonal and annual averages from 1974 to 2023. The seasons were defined as four-month spans: March-April-May (MAM), June-July-August (JJA), September-October-November (SON), and December-January-February (DJF). Linear regression analysis was employed to understand the trends in wind and wave parameters, with a statistical significance test applied and spatially mapped across the domain. The statistical significance of the trend values was calculated using the Mann-Kendall trend test at a 95% confidence level (Mann, 1945). Trends are reported in cm/yr; for reference, 0.1 cm/yr corresponds to 0.01 m/decade.

For the comparison of ERA5 wind data, in situ measurements from the National Data Buoy Center (NDBC) stations were selected (Figure 1, NDBC, 2024). The stations were West Gulf, TX (Station ID: 42002), Freeport, TX (Station ID: 42019), Orange Beach, AL (Station ID: 42012), West Florida Shelf South, FL (Station ID: 42023) and Mid Gulf (station ID: 42001). To compare ERA5 wave data with in situ measurements, five NDBC stations were selected across the GoM, spanning locations from deep water to shallow shelf areas (Figure 1). The chosen locations are Corpus Christi, TX (Station ID: 42020), Trinity Shoal, LA (Station ID: 42091), Southwest Pass, LA (Station ID: 42084), Egmont Channel Entrance, FL (Station ID: 42098), and Mid Gulf (Station ID: 42001) deployed at water depth of approximately 86m, 22m, 44m, 12m and 3200 m respectively (Figure 1). The time periods used for individual stations, along with their comparison statistics for both wind and wave parameters, are presented in Tables 1 and 2. The nearest grid points to these locations were extracted from the ERA5 dataset for analysis. To further assess the reliability of ERA5 spectral structure and its swell/wind-sea partitioning, we examined measured 1D and 2D spectra from the NDBC station ID: 42084. Spectra were analyzed under three representative conditions during June 2021: (i) a low-energy state (H_s,c< 0.3 m), (ii) a medium-energy state (H_s,c > 0.9 m), and (iii) a high-energy state associated with Tropical Storm Claudette (2021), where H_s,c exceeded 4 m.

Table 1

Table 1. Comparison statistics of ERA5 wind speed with measured wind speed data at different locations.

Table 2

Table 2. Comparison statistics between buoy measurements and ERA5 data at different locations for significant wave height, peak wave period, mean wave period, and mean wave direction.

The statistical parameters utilized for comparing ERA5 and measured data are bias, Root Mean Square Error (RMSE) and correlation coefficient (r), defined by the Equations 2–4 provided below.

\begin{array}{l} B i a s = \frac{1}{N} \sum_{i = 1}^{N} (A_{i} - B_{i}) & (2) \end{array}

\begin{array}{l} R M S E = \sqrt{\frac{1}{N} \sum_{i = 1}^{N} {(A_{i} - B_{i})}^{2}} & (3) \end{array}

\begin{array}{l} r = \frac{\sum_{i = 1}^{N} | (A_{i} - \bar{A}) (B_{i} - \bar{B}) |}{\sqrt{\sum_{i = 1}^{N} | {(A_{i} - \bar{A})}^{2} {(B_{i} - \bar{B})}^{2} |}} & (4) \end{array}

Where N represents the number of data points, A_i denotes the ERA5 data, B_i is the measured data and overbar is the mean value. For wave direction comparisons, circular statistics were applied to compute bias, RMSE, and correlation, following standard approaches (Fisher et al., 1993; Jammalamadaka and Sengupta, 2001).

3 Results

3.1 Comparison of ERA5 data with measured data

The comparison of wind speed and wave characteristics (i.e., H_s, T_p, mean wave period (T_m) and DIR) across different stations in the GoM is shown in Figure 2, and the corresponding statistical analyses are summarized in Tables 1 and 2. Wind speed comparison indicates a slight underestimation of ERA5 relative to in situ measurements across all locations, with biases ranging from –0.36 to –0.09 m/s, RMSE between 1.05 and 1.92 m/s, and r- values from 0.80 at shelf regions to 0.93 at deep-water locations (Figures 2a-e, Table 1).

In deep waters (~3200m), ERA5 H_s exhibited very slight underestimations relative to observations with an overall bias of -0.03m and strong correlation (r= 0.96, Figure 2j, Table 2). At ~86 m (Figure 2f), the underestimation was less than 1%, with an overall bias of -0.04 meters and an r-value of 0.95. At ~44 m (Figure 2h), however, a slight overestimation was observed (which remained below 5%), with an overall bias of 0.04 m and an r-value of 0.90. In shallow regions, ERA5 again slightly underestimated H_s. At a depth of ~22 m (Figure 2g), the underestimation was less than 6%, with an overall bias of -0.08 meters and an r-value of 0.95. At ~12 m (Figure 2i), the underestimation was under 8%, with an overall bias of -0.04 m and an r-value of 0.95. The RMSE values are 0.22, 0.24, 0.23, 0.18, and 0.14 at water depths of 3200m, 86m, 44m, 22m, and 12m, respectively (Table 2). The T_p shows an underestimation of less than 3% for all the locations except along the shallower location (~12m, Figure 2n). However, T_m (Figures 2p-t) shows overestimations throughout the observation period, where in the deep waters and shelf regions the overestimation even exceeds 15%. In the shallow regions, these overestimations are less than 5% (Figures 2l, n; Table 2). The DIR also shows overestimations along all the locations where RMSE exceeding 30 degrees (Figures 2u-y; Table 2). Figure 3 compares the measured and ERA5 derived 1D and 2D wave spectra under three representative wave-energy conditions during June 2021. Across all three energy conditions (Figures 3a–c), ERA5 consistently captures the dominant wave frequencies and the primary wave directions observed in the measured spectra. A slight bias in wave direction is evident, but the overall directional structure of the spectra is well represented. The larger scatter for T_m and DIR, previously noted in shallow zones of regional seas (e.g., George et al., 2020; Muhammed Naseef and Sanil Kumar, 2020), is attributed to the co-existence of wind-seas and swells (George et al., 2020). In mixed-sea conditions, where wind-sea and swell often have comparable energy but propagate from different directions, ERA5 may represent the swell component while the buoy captures the locally generated wind-sea direction, or vice versa (George et al., 2020). Additionally, short-lived wind fluctuations and sub-synoptic wind pulses that influence nearshore wave direction are often smoothed out in ERA5, contributing to the observed discrepancies. The present study uses ERA5 H_s data, where the bias is minimal, with over/underestimations less than 8% and r-values consistently above 0.9 across all locations along GoM.

Figure 2

Scatter plots showing comparisons of measured and estimated environmental data, including wind speed, wave height, and direction. Each subplot (a-y) displays bias, RMSE, and correlation coefficient values, with data points scattered around a diagonal line indicating perfect agreement. The plots assess the accuracy of estimates in various meteorological and oceanographic parameters.

Figure 2. Comparison of ERA5 and in situ observations from NDBC buoys. Panels (a–e) show time series of wind speed from ERA5 and measured data at different NDBC buoy locations. Panels (f–j) present scatter plots of significant wave height (H_s), (k–o) show scatter plots of peak wave period (T_p), (p–t) show scatter plots of mean wave period (T_m), and panels (u–y) show scatter plots of mean wave direction (DIR) for each buoy location.

Figure 3

Three sets of wave energy spectra and directional wave distributions are shown. Each set contains a line graph and two polar plots labeled ERA5 and 42084. Graph (a) shows data on 06/16/2021 with significant wave height of 0.24 meters. Graph (b) displays 06/19/2021 with 4.03 meters, and graph (c) presents 06/23/2021 with 1.04 meters. The line graphs compare data sets labeled ERA5 in blue and 42084 in red. Polar plots use a color scale from blue to red indicating wave energy intensity, aligned by date and label.

Figure 3. Comparison of 1D frequency spectra and 2D directional spectra from ERA5 and buoy measurement for three representative wave-energy conditions: (a) 16 June 2021, 12:00 UTC (Hs = 0.24 m), (b) 19 June 2021, 05:00 UTC (Hs = 4.03 m), and (c) 23 June 2021, 12:00 UTC (Hs = 1.04 m).

3.2 Annual climatology and trends

Annual climatology and trends of wind speeds, H_s,c, H_s,s, H_s,w are shown in Figure 4. Climatological wind speeds across the GoM range from 0.10 to 5.40 m/s, predominantly directed westward, with the strongest winds (> 4.5 m/s) observed near the Yucatan Peninsula and northern CS, and the weakest winds (< 2 m/s) in the northeastern GoM, where they are primarily directed westward (Figure 4a). Compared to the northeastern region, the northwestern GoM exhibits stronger winds (> 3 m/s), although the wind direction remains consistent, flowing towards the west-northwest. Climatological H_s,c (Figure 4c) across the GoM ranges from 0.38 to 1.2 m, with the western GoM and areas north of the CS experiencing higher waves (> 1.1 m) followed by the south-central GoM compared to other regions. The associated wave directions largely mirror the prevailing wind patterns. The shelf regions in the northern, southwestern, and eastern GoM exhibit comparatively lower H_s,c (< 0.9m), with the eastern GoM (west Florida shelf) showing the lowest H_s,c (< 0.7 m). H_s,s (Figure 4e) shows a similar distribution, extending through the Yucatan Channel into the GoM, with values ranging from 0.26 to 1 m with highest values (>0.85 m) in the northern CS and western GoM, and lowest (<0.4 m) along the eastern shelf. The corresponding wave directions generally mirror those of H_s,c across most regions, except along some parts of west Florida shelf where waves are predominantly directed eastward. In contrast, the H_s,w climatology differs from H_s,c and H_s,s in both distribution and magnitude, ranging from 0.12 m to 0.72 m (Figure 4g) with larger (>0.6 m) H_s,w in the northwest and southwest GoM and > 0.5 m across north, southwest and east shelf regions. The associated wave directions are mostly directed westward.

Figure 4

Eight-panel diagram showing annual climatology and trend maps for wind speed and wave height in the Gulf of Mexico. Panels (a) and (b) show wind speed climatology and trends, with vectors indicating direction and color scales indicating magnitude. Panels (c) and (d) display similar data for significant wave height $H_s$ with arrows for wave direction. Panels (e) and (f) present annual climatology and trends for storm-driven wave height $H_{s,s}$. Panels (g) and (h) depict significant wave height $H_{sw}$, with color bars indicating variations in m/s, m, and cm/year.

Figure 4. Annual climatology (left panel) and linear trends (right panel) for wind speeds (a, b), significant wave height of combined wind-sea and swell waves [H_s,c, (c, d)], significant wave height of swell waves [H_s,s, (e, f)], significant wave height of wind seas [H_s,w, (g, h)]. Dotted areas indicate regions where trends are statistically significant at 95% confidence level. Vectors indicate direction (coming from).

Trend analysis of wind speeds (Figure 4b) shows an overall modest increase across the GoM, though only the northeast of the CS and parts of the southwest GoM exhibit significant increases (> 0.6 cm/s/yr), while a significant decrease (< –1.0 cm/s/yr) is observed near the southern tip of Florida. Corresponding trends in H_s,c show a widespread modest increase in wave heights (> 0.2 cm/yr) in the western and central regions, while decreasing trend (< - 0.1 cm/yr) is observed in some parts of the eastern Florida shelf (Figure 4d). The H_s,s (Figure 4f) trends (~0.2 cm/yr) follows the same increasing pattern of H_s,c, extending from the northern CS to the southern GoM through the Yucatan Channel. H_s,w (Figure 4h) trends are more localized, with increase in wave heights (~ 0.1 cm/yr) in the southwest, central GoM, and northeastern CS, aligning with regions of increasing wind speeds (Figure 4b), whereas significant decreasing trends (< -0.10 cm/yr) are observed near the Florida shelf (Figure 4h).

3.3 Seasonal climatology and trends

3.3.1 Wind speed

Figure 4 shows the seasonal climatology of wind speeds and trends over the GoM. During MAM (Figure 5a), wind speeds range from 0.24 to 6.40 m/s, predominantly directed west-northwest in the central GoM. Strongest winds (>5 m/s) occur north of the CS and in the southwest near the Yucatan Peninsula, while weaker winds (<2 m/s) are observed along the eastern GoM. Winds in the southwest flow southward (>4 m/s), whereas in the north, they are northwestward with lower magnitudes (<4 m/s). A distinct shift in wind directions is evident during JJA (Figure 5c), transitioning from west to northwest, especially in the north-central and western GoM, reflecting clear seasonal variability. A reduction in the wind speed magnitude is observed in the central and northern regions relative to MAM, while the western GoM maintains stronger winds (>5 m/s). Additionally, the higher magnitude winds observed during MAM (Figure 5a) along the southwest GoM close to the Yucatan channel and northeast of the CS persist during JJA. In the eastern GoM, winds turn northward but remain weak (<2 m/s). During SON (Figure 5e), wind direction shifts to west-southwest, with reduced magnitudes in the western GoM (<3.5 m/s). However, the higher magnitude wind belt in the southwest and north of the CS persists with reduced intensity (<5 m/s), while the northern GoM sees moderately stronger winds (>2 m/s). In DJF (Figure 5g), wind speeds peak across the basin, exceeding 7 m/s in the west and 6 m/s in the northern and eastern GoM, with prevailing west-southwest directions.

Figure 5

Eight-panel chart showing wind speed climatology and trend data for different seasons in the Gulf of Mexico. Panels (a), (c), (e), and (g) display climatology data for MAM, JJA, SON, and DJF, respectively, with color gradients from blue to red indicating wind speed in meters per second. Panels (b), (d), (f), and (h) show trends for the same seasons with colors indicating changes from negative (blue) to positive (red) in centimeters per second per year. Arrows depict wind direction and intensity.

Figure 5. Seasonal climatology of wind speeds (left panel) and linear trends (right panel) during MAM (a, b), JJA (c, d), SON (e, f) and DJF (g, h). Dotted areas indicate regions where trends are statistically significant at 95% confidence level. Vectors indicate wind direction (coming from).

The trend analysis during MAM (Figure 5b) reveals a significant increase in wind speed (>1 cm/s/yr) along the southwest GoM and northeast of the CS, while the Florida shelf shows significant decreasing trend (< –1.0 cm/s/yr). In JJA (Figure 5d), trends strengthen further, with wind speeds increasing by >1–2 cm/s/yr in the southwest and near the Louisiana coast and decreasing (< –1.0 cm/s/yr) south of Florida and along the northern CS. SON (Figure 5f) exhibits minimal change, with a significant decreasing trend confined to the south Florida shelf. In DJF (Figure 5h), a significant increasing trend (>1 cm/s/yr) is observed in isolated areas of the northern CS and scattered regions across the central, western, and southern GoM, while a strong decreasing trend (< –1.5 cm/s/yr) continues along the Florida shelf and adjacent land areas.

3.3.2 Significant wave height of combined wind seas and swells

Figure 6 shows the seasonal climatology of H_s,c and trends over the GoM. During MAM (Figure 6a), H_s,c ranges from 0.5 to 1.25 m, with the highest values (>1.15 m) observed in the western GoM where the waves are predominantly directed westward, extending into the south-central basin and northern CS. Lower H_s,c values (<0.8 m) occur along the northern shelf, where waves are directed northwestward, along the eastern shelf with northward-directed waves, and in parts of the southern GoM where waves are oriented southwest to south. Conditions during the JJA (Figure 6c) are generally calmer, particularly in the eastern and northern regions (0.5–0.7 m) where the waves are directed northward, while the northern CS and western GoM exhibit higher values, exceeding 1.0 m and 0.9 m, respectively. H_s,c increases slightly during SON (Figure 6e), with values >1.0 m in the central and western GoM, extending northward along the CS. The most energetic conditions occur in DJF (Figure 6g), where H_s,c exceeds 1.4 m in the western GoM and 1.3 m in the central basin, with increased H_s,c values (>0.9 m) extending across the northern and eastern shelf regions, where waves are predominantly directed toward the southwest.

Figure 6

Seasonal climatology and trend maps for significant wave height ($H_{s,c}$) in the Gulf of Mexico. Panels show March-May (a, b), June-August (c, d), September-November (e, f), and December-February (g, h). Climatology maps (a, c, e, g) depict wave heights with color gradients from blue to red, indicating lower to higher values. Trend maps (b, d, f, h) use similar colors to show changes over time, with blue representing negative trends and red positive. Arrows depict wind direction.

Figure 6. Seasonal climatology of significant wave height of combined wind-sea and swell waves (H_s,c, left panel) and linear trends (right panel) during MAM (a, b), JJA (c, d), SON (e, f) and DJF (g, h). Dotted areas indicate regions where trends are statistically significant at 95% confidence level. Vectors indicate wave direction (coming from).

The trend analysis during MAM (Figure 6b) reveals that increasing trends in H_s,c dominates the western and central GoM, with peak values exceeding 0.3 cm/yr in the southwest, extending into the central basin. Moderate increases (> 0.1 cm/yr) are also observed in the northern CS and adjacent regions (Figure 6b), while the Florida shelf shows a decreasing trend. In JJA (Figure 6d), rising trends continue throughout the basin (0.2 cm/yr), with the southwest GoM showing stronger increases (>0.25 cm/yr). Trends during SON (Figure 6f) are generally weak and insignificant across the GoM. In contrast, DJF (Figure 6h) exhibits the strongest trend, with widespread increases (>0.2 cm/yr) across the western and central GoM and peak values exceeding 0.4 cm/yr in the northern CS. A significant decreasing trend (–0.10 cm/yr) is observed along the Florida shelf and surrounding areas during this season.

3.3.3 Significant wave height of swells

Figure 7 shows the seasonal climatology H_s,s and trends over the GoM. During MAM (Figure 7a), higher H_s,s values (> 0.9 m) are observed in the western GoM, extending into the central basin and through the Yucatan Channel. The northern CS also experiences larger H_s,s values (> 0.95 m), while the northern and eastern shelf regions exhibit lower H_s,s (< 0.7 m where the waves are predominantly directed northwest and northward. A slight decrease is noted in JJA (Figure 7c), although Hs,s remains relatively higher (> 0.70 m) along the western and central GoM and near the Yucatan Channel. The northern CS continues to exceed 1 m, while the values remain lower (< 0.6 m) along the northern and southern shelf regions, and below 0.5 m in the northeastern GoM. By SON (Figure 7e), H_s,s values< 0.85 m extend from the northern CS through the central and western GoM, while the lowest values (> 0.5 m) persist along the eastern and northern shelf regions, where the waves are predominantly directed south-southwest. A basin wide intensification is observed in DJF (Figure 7g), with the highest values (>1.1 m) in the southwestern GoM and northern CS, gradually decreasing towards the central, eastern, and northern shelf regions (<0.8 m). Along the eastern GoM, the waves are predominantly eastward directed, while in the northern GoM they are directed northwest, and in the central–southern regions they are directed southwest–south.

Figure 7

Series of maps showing significant wave height (H_s) climatology and trends in the Gulf of Mexico for different seasons. Panels (a), (c), (e), and (g) display climatology with wave heights indicated by colors from blue to red and arrows for wave direction. Panels (b), (d), (f), and (h) show trends with color gradients from blue to red. Latitude and longitude are marked on each map. Legends indicate wave height in meters and trends in centimeters per year.

Figure 7. Seasonal climatology of significant wave height of swell waves (H_s,s,left panel) and linear trends (right panel) during MAM (a, b), JJA (c, d), SON (e, f) and DJF (g, h). Dotted areas indicate regions where trends are statistically significant at 95% confidence level. Vectors indicate wave direction (coming from).

Trend analysis during MAM (Figure 7b) shows increasing H_s,s values (~0.3 cm/yr) in the southwestern GoM, gradually decreasing to ~ 0.2 cm/yr as they extend into the central and southern regions. Modest increasing trends (~0.1 cm/yr) are noted further east, while the Florida shelf exhibits a decreasing tendency. In JJA, significant increases in H_s,s are observed throughout the basin, with notable trends (~0.2 cm/yr) in the northwestern GoM, extending southward and eastward toward the Louisiana coast and central GoM (Figure 7d). Areas that previously showed decreasing trends during MAM, such as the Florida shelf, now exhibit an increasing trend. Positive trends observed on SON (Figure 7f), especially from the northern CS extending through the central basin and into the Yucatan Channel and western regions (> 0.2 cm/yr). In DJF (Figure 7h), the strongest H_s,s trends occur in the northern CS (> 0.3 cm/yr), followed by increases in the western and central GoM (> 0.2 cm/yr), while a declining pattern is observed in the eastern GoM.

3.3.4 Significant wave height of wind-seas

Figure 8 shows the seasonal climatology H_s,w and trends over the GoM. During MAM (Figure 8a), the strongest H_s,w values (>0.7 m) occur along the southwestern GoM near the Yucatan Peninsula, extending into the northwestern and central regions. In the eastern GoM, the waves are directed south–southwest, while along most of the remaining GoM they are predominantly westward. In JJA (Figure 8c), the overall intensity weakens, but H_s,w remains stronger (~0.5 m) in the southern and western GoM, with lower values (>0.3 m) along the northern and eastern regions. In the eastern GoM, the waves are directed south–southwest, while in the central basin they are predominantly westward, and in the northwestern GoM they are directed west–northwest. By SON (Figure 8e), the highest H_s,w values (>0.6 m) shift to the northwestern GoM, extending across the northern basin, central GoM, and northern CS, where the waves are predominantly directed west–southwest. The seasonal peak occurs in DJF (Figure 8g), with H_s,w exceeding 0.9 m in the western GoM and extending into the northern and central basin. Shelf regions in the northern, southern, and eastern GoM also experience relatively higher H_s,w during this season compared to others.

Figure 8

Eight maps showing climatology and trend data for significant wave height (Hs,w) in the Gulf of Mexico. Panels (a), (c), (e), and (g) depict the climatology for spring, summer, fall, and winter, respectively, with varying wave heights. Panels (b), (d), (f), and (h) illustrate corresponding trends in wave height changes over time for each season. Color scales at the bottom indicate wave height intensity and trend magnitude. Black arrows represent wave directions.

Figure 8. Seasonal climatology of significant wave height of wind-seas (H_s,w, left panel) and linear trends (right panel) during MAM (a, b), JJA (c, d), SON (e, f) and DJF (g, h). Dotted areas indicate regions where trends are statistically significant at 95% confidence level. Vectors indicate wave direction (coming from).

The trend analysis during MAM (Figure 8b), shows increasing H_s,w trends (~0.2 cm/yr) in the southern GoM near the Yucatan Channel and in isolated areas of the western GoM (0.18 cm/yr), while significant decreasing trends (–0.10 cm/yr) are noted along the Florida shelf (Figure 8b). In JJA (Figure 8d), H_s,w trends continue to increase across the southern and western GoM (~0.2 cm/yr), with moderate increases (0.10 cm/yr) extending from north of Cuba towards the Louisiana coast. A significant decrease is also noticed north of the CS. During SON (Figure 8f), the trends are insignificant. By DJF (Figure 8h), increasing trends dominate the GoM, with significant values (>0.2 cm/yr) in the northern CS, while the Florida coast exhibits a significant negative trend (–0.10 cm/yr).

4 Discussion

4.1 Wind - wave climatology and spatial trends in the GoM

The ERA5 reanalysis dataset forms the foundation of our climatological and trend analysis of wind and waves in the GoM. This dataset has been widely validated for global and regional wind - wave applications (e.g., Muhammed Naseef and Sanil Kumar, 2020; Parsons et al., 2018). In this study, we extended that validation by comparing ERA5 derived wind and wave parameters against in situ records, demonstrating robust agreement and affirming its suitability for long-term climatological and trend analyses in the GoM. To the best of our knowledge, this is the first multi-decadal study in the GoM that explicitly evaluates the long-term climatology and trends of swell and wind-sea components.

Our climatological analysis of wind speeds on both annual (Figure 4a) and seasonal scales (Figures 5a, c, e, g), reveals substantial regional variations across the GoM, with pronounced seasonal shifts in magnitude and direction. Annually, westward directed winds dominate the GoM, driven by the trade winds, while the northern and southwestern GoM exhibit directional shifts linked to seasonal forcing and orographic influences from surrounding mountain range (De Velasco and Winant, 1996). Wind speeds are weakest during JJA (Figure 5c), consistent with the seasonal reduction in frontal system activity (De Velasco and Winant, 1996). Stronger winds in SON and DJF, especially in the northern GoM, reflect increased cold front frequency (Roberts et al., 2015; Hiatt et al., 2019), are also evident in our analysis (Figure 5g). These findings are in good agreement with the detailed seasonal description provided by Zavala-Hidalgo et al. (2014), who noted that summer winds turn northward in the northwestern GoM due to the influence of the North Atlantic subtropical high and land heating, while in autumn - winter, winds strengthen under frequent cold front passages, shifting toward west and northwest. They further highlighted that in the southwestern GoM, blocking by Sierra Madre Oriental mountain range directs winds southward, a feature also consistent with our seasonal analysis.

Trend analysis reveals mostly insignificant increases in wind speeds across the GoM, except along the southern Florida coast, where a consistent and statistically significant decrease of ~ 1.0 cm/s/yr is observed annually and seasonally (Figure 4b; Figures 5b, d, f, h). These findings are consistent with the insignificant increasing trends along the GoM reported by Muller-Karger et al. (2015) and Lin and Oey (2020). However, our multidecadal analysis highlights a significant decline in wind speeds over the eastern GoM, particularly along the south Florida shelf.

Climatological analysis of H_s,c reveals a clear zonal and meridional (Figure 4c), with higher values in the western and central GoM and a northward decrease across the northern CS, consistent for all seasons (Figures 6a, c, e, g). These patterns reflect the influence of westward-directed winds generating long-fetch waves (Abolfazli et al., 2020; Appendini et al., 2014) and the Caribbean Low-Level Jet (CLLJ) associated with 925-hPa meridional wind anomalies (Appendini et al., 2014; Wang, 2007). Previous studies have also noted that wave events in the GoM are modulated by the CLLJ, which intensifies in February and July and can propagate into the western GoM through the Yucatán Channel (Appendini et al., 2014; Jamous and Marsooli, 2023). Lower H_s,c values particularly along the northern, southwestern and eastern shelf regions of the GoM (notably the west Florida shelf), are attributed to limited fetch (Abolfazli et al., 2020) and wave attenuation in shallow waters (Van der Westhuysen, 2010). In-situ measurements by Georgiou et al. (2005) further support this, showing that the shallow northern GoM typically experiences low-energy waves ranging from 0.5 to 1 m, consistent with our findings. Our analysis clearly reveals that the seasonal wind patterns (Figures 5a, c, e, f) strongly influence the wave climate in the GoM, particularly the spatial and temporal distribution of H_s,c with peak wave heights occurring during DJF (Figure 6f) and minimum during JJA (Figure 6c).

As previously discussed, although wind speeds are generally increasing across the GoM, these increases are not statistically significant (Figure 4b; Figures 5b, d, f, h). In contrast, H_s,c shows clear and significant, but modest increasing trends on annual and seasonal scales (Figure 4d; Figures 6b, d, h), except during SON (Figure 6f). Annually, most regions of the GoM exhibit increasing H_s,c trends, particularly in the western and central basins (> 0.2 cm/yr). These findings align with Zheng et al. (2022), who reported similar trends using ERA5 data, and with Appendini et al. (2014), who observed stronger trends from Louisiana to the Yucatan Channel. Additionally, our analysis reveals that these increasing trends extend southwestward into the western GoM and connect with increasing trends from the CS through the Yucatan Channel (Figure 4d). Previous long-term analyses by Appendini et al. (2014), showed an increase in H_s,c in the eastern GoM during September, and some parts of the western region, in November. In contrast, during October, their study observed an increase of 2–3 cm/yr in H_s,c, while our analysis indicates a lower trend of approximately 0.2 cm/yr for the entire season, though this is statistically insignificant. However, at a lower confidence level (85%), we observed significant increasing trends in most parts of the GoM, from the southern region extending toward the central GoM, aligning with similar magnitudes for the same season observed by Zheng et al. (2022). These discrepancies between our findings and those of Appendini et al. (2014) can likely be attributed to the period of analysis and the wind data sources used. Appendini et al. (2014) focused on 1979–2008 (30 years) and relied on the North American Regional Reanalysis (NARR; Mesinger et al., 2006), which is geographically limited to the North American region. Furthermore, Appendini et al. (2014) modeled only high-frequency waves within the Caribbean Sea and did not account for low-frequency swells propagating from the Atlantic into the Gulf, potentially underestimating the swell contribution to the total wave. In contrast, our study spans a longer temporal window (1974–2023) using ERA5, a globally consistent reanalysis dataset with improved spatial and temporal resolution that captures a broader spectrum of surface wave processes. Overall, our analysis confirms an increasing trend in H_s,c across the GoM over the last five decades. Notably, while H_s,c is rising, wind speeds are not significantly increasing, suggesting the influence of other mechanisms.

To better understand the drivers of these changes, we examined the climatology and trends of separated wind-sea and swell components across annual and seasonal scales. The annual and seasonal distributions of H_s,s (Figures 4e, 7a,c,e,g) reveal strong zonal and meridional gradients similar to those in H_s,c (Figures 4c, 6a,c,e,g), and the spatial pattern of their trends are also strongly aligned (Figures 4d, f; Figures 6b, d, f, h; Figures 7b, d, f, h), highlighting the influence of swell waves in shaping the overall H_s,c across the GoM. H_s,s is largest in the western GoM and northern CS, with lower values along the eastern shelf. These patterns are consistent with findings by Abolfazli et al. (2020), who reported persistent high swell energy in the southwestern GoM, supported by prevailing westward winds. Seasonal analysis shows that H_s,s increasing trends are more widespread than H_s,c, especially during SON (Figures 7b, d, f, h). Notably, SON (Figure 7f) exhibits a distinct increasing trend in H_s,s from the northern CS through the Yucatan Channel into the central and western GoM. Stronger winds (> 2 m/s) observed along the north-central GoM during SON, directed southwest-west and driven by cold front events, are typical during this season (Figure 5e). These ‘local swells’ generated within the north-central GoM propagate to the south-southwest (Figure 7e). Moreover, the magnitude of seasonal mean wind stress is greatest during SON along the Yucatan Basin in the CS (Pérez-Santos et al., 2010). The waves generated in this region (CS) likely propagating through the Yucatan Channel as ‘remote swells’ into the southern GoM, where they coexist with southwest–west traveling ‘local swells’ generated in the north-central basin. The presence of these distinct wave systems is evident from the wave rose distributions (point D and E in Figure 9), which indicate contributions from both the southeast and east. These coexisting wave systems contribute to enhanced H_s,s (Figure 9e) and increased trends (Figure 9f) as they propagate westward across the GoM. The climatology of swell energy fraction during SON, as observed by Abolfazli et al. (2020), also reflects this pattern, where they reported strong swell energy fractions along the western GoM that diminish but extend towards the central GoM and the Yucatan Channel. It is noteworthy that Jamous and Marsooli (2023), using long-term in situ records in the central GoM, also reported increasing H_s trends at a buoy located directly facing the Yucatán Channel. They attributed this signal to swells originating in the CS, which can propagate along great circle routes through the Yucatán Channel and into the central and northern GoM. Our analysis on annual and seasonal climatological assessments and trends for H_s,c (Figure 6) and H_s,s (Figure 7) noted similar patterns of stronger H_s,c and H_s,s, along with increasing trends observed around the Yucatan Channel, implying that these ‘remote swells’ from the CS reaching the south-central GoM modulates the H_s,c and H_s,s trends in the GoM.

Figure 9

Map of the Gulf of Mexico and Caribbean Sea showing compass rose diagrams labeled [A] to [E]. Each diagram represents wave period ranges, colored from light green for greater than eight to nine seconds, to dark blue for greater than twelve seconds. The areas include geographic labels such as Straits of Florida and Yucatán Channel.

Figure 9. Map showing the wave rose distribution of long period waves at four reference locations (B–E) in the GOM and one location (A) in the northern CS. Map is generated using Ocean Data View software (Schlitzer, 2015).

The presence and propagation pathways of these ‘remote swells’ are further supported by the distribution of T_p across reference locations (Figure 9). For this analysis, we separated the low frequency wave component, considering only waves with T_p > 8 s. Intermediate long-period waves (>8 s) approach predominantly from the southeast in the northern CS (Point A in Figure 9), where they occur more than 30% of the time during the analysis period. Along the southern GoM (Point B), these southeast-approaching waves persist and evolve into longer-period swells (>10 s). As they propagate into the central (Point D) and western GoM (Point E), these waves display longer periods and an increased frequency of occurrence from both the southeast (‘remote swells’) and the east (‘local swells’). In the northeastern GoM (Point C), the signatures of these ‘remote swells’ from the CS and south GoM are particularly evident, as they appear as longer period waves exceeding 12s approaching predominantly from the south.

The climatology and trends of H_s,w, both annually (Figures 4g, h) and seasonally (Figures 8a–h), differ notably from H_s,c and H_s,s in both distribution and magnitude. H_s,w patterns closely mirror regional wind speed distributions, with higher values concentrated in the northwest and southwest GoM, particularly near the Yucatan Peninsula, where stronger winds are observed (Figure 4a). This spatial correlation extends to trends where areas with increasing or decreasing wind speeds (Figure 4b) exhibit corresponding changes in H_s,w (Figure 4h), underscoring the strong wind-wave coupling in the region. Seasonal analysis further confirms this relationship where regions with largest H_s,w (Figure 8) align with zones of intensified wind speeds (Figure 5). This relationship is very common in semi-enclosed systems, where increased wind speeds lead to higher wave heights with relatively short-wave periods (Young and Verhagen, 1996). Overall, climatological and trend analysis of wind speeds and H_s,w in the GoM reveals that H_s,w is predominantly driven by local winds, consistent with global findings by Zheng et al. (2022), who reported a strong correlation (r > 0.8) between H_s,w and wind speeds.

The interpretation of long-term wave-height trends from reanalysis products must be approached with caution because their temporal homogeneity is influenced by changes in the observing system, particularly the introduction of satellite altimeters and scatterometry winds after the early 1990s (Aarnes et al., 2015; Muhammed Naseef and Sanil Kumar, 2020). Consistent with these studies, our sensitivity analysis, based on splitting the record into pre-satellite (1974–1990) and post-satellite (1991–2023) periods, shows that both the sign and magnitude of trends differ at several locations (Figure 1; Supplementary Figure S1), indicating that the observing-system updates can modulate long-term trend estimates. However, it is also important to recognize that wave-height trends themselves can evolve over time, and part of these differences likely reflects genuine interannual variability rather than solely inhomogeneities in the reanalysis products (Muhammed Naseef and Sanil Kumar, 2020). Even with these considerations, the full 50-year trends reported here remain statistically significant at the 95% confidence level and display strong spatial coherence across the basin. Moreover, buoy-based analyses in the GoM (Jamous and Marsooli, 2023) report trends of similar magnitude (−0.2 to 0.5 cm yr^-¹), confirming that the increases observed in ERA5 are consistent with independent observational datasets. Comparable modest but significant upward trends have also been documented in other global and regional assessments (Appendini et al., 2014; Young and Ribal, 2019; Zheng et al., 2022).

Overall, the GoM wave climate is governed by the interplay of local wind forcing, remote and local swells, and geographic constraints. While previous studies (e.g., Appendini et al. (2014); Jamous and Marsooli (2023)) have documented increasing H_s,c trends in the region, this study provides a new perspective by attributing these changes to an increase in swell waves, both generated within the GoM and remotely propagated from the CS.

4.2 Practical implications

The growing dominance of swell waves and the persistent rise in wave heights across the GoM have significant implications for coastal infrastructure, offshore operations and regional climate resilience. While swells typically have lower steepness than locally generated wind seas, their long periods and persistence enable them to propagate efficiently across the basin and exert substantial impacts nearshore. Although these gradual increases may not exceed the uncertainty range of measurements, their cumulative effect over multiple decades, when combined with sea-level rise and increasing storm activity, can amplify coastal flooding and erosion risks (Tebaldi et al., 2021; Storlazzi et al., 2018). Thus, even modest positive trends in H_s,c should not be disregarded in the context of long-term coastal impacts. This is particularly critical in the GoM, a major hub of offshore oil and gas production in both the United States and Mexico, where extensive platform infrastructure is concentrated in the north-central and southern sectors (BOEM, 2025; Day et al., 2024). The region also sustains a multi-billion-dollar marine transportation industry, with key shipping routes in the western and central GoM (BOEM, 2025). Increasing swell dominance in these sectors could compromise offshore platform integrity, accelerate coastal erosion, and pose risks to maritime safety, underscoring the need to adapt engineering design standards and navigational protocols to evolving wave conditions.

In addition to these offshore and industry related concerns, low-lying coastlines along the northern GoM are particularly vulnerable. Much of this coastline is geomorphologically sensitive, composed of marshes, mangrove swamps, and barrier islands (Day et al., 2024; Lamourie et al., 2025). These natural buffers are particularly susceptible to increasing wave heights, which can accelerate shoreline erosion, exacerbate flooding, and degrade ecologically significant habitats. Our multidecadal analysis of wave characteristics at selected sites in the northern GoM (Figure 1; see also Supplementary Figure S2) reveals substantial spatial and temporal variability. This variability underscores the need for site-specific assessments, as the impacts of rising wave heights are not uniform and may disproportionately affect certain regions and natural buffers. For example, the west Florida shelf exhibits a combination of low elevation, gentle slope, and active subsidence, making it highly vulnerable to sea level rise (Flocks et al., 2022; Toth et al., 2018). Our results show a significant long-term decline in wind speed over this region, which may reduce local wind-seas generation. However, we concurrently observe a moderate rise in swell wave activity during calmer summer months (Figure 7d) along the same shelf, suggesting that remote swell forcing is increasingly shaping the regional wave climate. Similarly, coastal Louisiana is undergoing extreme rates of subsidence (Allison et al., 2016; Flocks et al., 2022), further amplifying its exposure to wave-driven impacts. Together, these factors imply that changes in wave climate will disproportionately affect vulnerable sectors of the GoM, both offshore and along the coast. Moreover, projected increases in wave heights across the Southern Ocean and intensified swell propagation into adjacent basins, including the Atlantic (Casas-Prat et al., 2024), raise the likelihood of more frequent and intense remote swell events impacting the GoM, underscoring the need to integrate evolving wave climatology into infrastructure planning, hazard mitigation, and coastal management strategies.

4.3 Limitations and future perspectives

Using multidecadal ERA5 reanalysis data, we show that the observed increase in H_s,c is not solely driven by local wind-sea interactions but also reflects an enhanced contribution from swell waves, both generated within the GoM and remotely propagated from the CS. However, there are several limitations that must be acknowledged.

First, the interpretation of H_s,s and H_s,w trends are influenced by limitations in the ERA5 spectral partitioning scheme. Even though ERA5’s wave partitioning relies on wave–age forcing criterion, its coarse spatial resolution constrains its ability to resolve coastal refraction, mixed wind-sea–swell interactions, and short-lived wind fluctuations, the conditions common in the GoM. The ~31 km grid spacing also limits the model’s capacity to represent fine-scale bathymetric effects, fetch-limited wind-sea generation in shallow areas, and shoaling or turning of waves near the coast, all of which further contribute to directional discrepancies in nearshore regions. In addition, the swell partitioning itself is based purely on relative spectral energy and not on spatial coherence or source direction. As a result, the “first swell partition” at one grid point may not correspond to the same swell system at a neighboring location, and some low-energy components may remain unassigned. These issues are amplified in a semi-enclosed basin like the GoM, where restricted geometry and limited swell pathways reduce the persistence and clarity of distinct swell systems. Consequently, as demonstrated in our spectral validation, ERA5 successfully captures dominant frequencies and direction, however a slight bias in dominant direction and smoothing of multi-modal spectra is evident, particularly in nearshore environments. We therefore interpret swell and wind-sea trends with caution, while still benefiting from the ERA5 partitioning scheme, which captures the dominant separation between locally generated wind-sea and remotely generated low-frequency swell. However, these limitations do not invalidate the large-scale climatological trends but highlight the need for caution when interpreting fine-scale directional statistics or localized swell and wind-sea separation.

ERA5 also shows small but consistent underestimation of H_s in nearshore locations, though these biases remain within ~8% and display no temporal drift across the 50-year record, suggesting that these biases do not alter the long-term variability or trends. Because selective bias correction could introduce spatial inconsistencies, we did not apply bias adjustments and instead interpret nearshore trends cautiously. Similar regional studies (e.g., Muhammed Naseef and Sanil Kumar, 2020) have reported comparable nearshore biases but also demonstrated that ERA5 maintains high fidelity in representing temporal evolution and interannual variability. Based on this evidence and given that our objective is to assess basin-wide, multi-decadal changes rather than reconstruct short-term events, we did not apply a bias correction procedure. Even though the magnitude of the observed long-term trend in H_s,c (~-0.1–0.5xcm/yr) is smaller than the ERA5 bias (-0.08 to +0.04 m) and RMSE (0.15-0.22m) values, these metrics characterize short-term model-observation discrepancies rather than long term variability. The Mann–Kendall test confirms that the observed slopes are statistically significant at the 95% confidence level, and the spatial coherence of these trends across the basin further indicates that they represent a systematic multi-decadal signal. While the ~0.1–0.3 cm/yr increases in mean H_s,c are modest and fall within the uncertainty range of reanalysis biases and instrument accuracy, their cumulative effect over multiple decades, particularly when combined with sea-level rise, storm surges, and land subsidence, are likely to amplify coastal flooding and erosion risks.

Second, the study focuses on long-term mean and seasonal wave climate rather than extremes, even though extreme wave events generated by hurricanes, tropical storms, and strong cold fronts are often the primary drivers of immediate coastal damage and are critical for coastal hazard assessments. A dedicated investigation of extreme-event statistics is beyond the scope of this study; however, future work will extend this analysis to evaluate storm-driven H_s peaks, front-induced wave events, and associated return-period characteristics to better understand how extremes evolve alongside long-term climatological trends.

Third, the temporal variability, spatial extent and origins of swell waves still require further investigation. For e.g., the CLLJ intensifies during El Niño events and weakens during La Niña (Vega et al., 2020), while wave power in the GoM and CS has been shown to correlate negatively with the Atlantic Multidecadal Oscillation (AMO, Reguero et al., 2019). Taken together, these observations imply that large-scale climate oscillations such as ENSO and AMO may significantly modulate swell wave dynamics and overall wave energy distribution in the GoM. Understanding these links is essential for anticipating future wave climate variability in the region and will be an important focus of our future research.

Finally, although this study examined the long-term climatology and trends of wind speeds, H_s,c, H_s,s, and H_s,w across the GoM, a more thorough assessment of the spatial and temporal variability of wave characteristics, as well as their long-term behavior and distribution, is still required. While our preliminary site-based assessment in the northern GoM (Figure 1; Supplementary Figure S2) reveals considerable variability, it also highlights the need for detailed, region-specific studies. Future work should also incorporate wave directionality and spectral parameters to improve understanding of wave dynamics and their implications for coastal processes. These efforts are essential to fully characterize the evolving wave climate and are a key focus of our future research.

5 Conclusions

This study provides a synoptic analysis of the climatology and trends of wind speeds, significant wave height (H_s) of combined wind-seas and swells (H_s,c), H_s of swell waves (H_s,s) and H_s of wind-seas (H_s,w) across the GoM using 50 years of ERA5 reanalysis data. The comparison of ERA5 data with in-situ measurements of wind and wave parameters confirms the reliability of ERA5 in representing both wind and wave conditions in the GoM. Annually, the strongest winds occur in the southern GoM and northern Caribbean Sea (CS), however, wind speed trends are mostly insignificant (but increasing), except for significant decreases in the southern Florida region. H_s,c trends show widespread increases, particularly in the west and central basin. H_s,s display similar spatial patterns, especially from the northern CS through the Yucatán Channel. In contrast, H_s,w are more localized and closely tied to wind patterns, exhibiting only modest trend changes. Importantly, the separated analysis of wind-seas and swells reveals that although wind speeds across the GoM are not significantly increasing, wave heights are on a modest rise, primarily driven by the swell waves originating from remote locations and within the GoM. The analysis suggests that these waves are not solely propagated westward; instead, distinct wave systems are evident, with swell waves from the CS traveling through the Yucatán Channel and the westward-propagating waves generated in the east-central GoM, further intensifying as they reach the western GoM. This dynamic is clearly reflected in the annual and seasonal climatological fields and trends of both H_s,s and H_s,c, which consistently show stronger wave heights and widespread increases around the Yucatán Channel and across the broader GoM. The distribution of longer period waves across different reference locations further supports these swell propagation pathways. These findings highlight the growing influence of swell wave dynamics in shaping the GoM wave climate and provide a foundation for future investigations into remote forcing and basin scale wave variability.

Data availability statement

Publicly available datasets were analyzed in this study. This data can be found here: https://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-timeseries?tab=overview, and https://www.ndbc.noaa.gov/.

Author contributions

JG: Formal Analysis, Validation, Conceptualization, Writing – review & editing, Methodology, Data curation, Writing – original draft, Investigation, Visualization, Software. MH: Project administration, Resources, Writing – review & editing, Supervision, Funding acquisition, Writing – original draft, Conceptualization, Investigation. CW: Writing – review & editing, Supervision, Project administration, Writing – original draft, Funding acquisition.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the U.S. Department of Defense through the U.S. Army Engineer Research and Development Center (ERDC) under contract Nos. W912HZ2220005 and W912HZ2520012.

Acknowledgments

The authors thank the reviewers and the editor for their constructive comments, which significantly improved the quality of this manuscript. We also acknowledge the European Centre for Medium-Range Weather Forecasts (ECMWF) for providing the open access ERA5 reanalysis. All figures were generated using Python and Ocean data view.

Conflict of interest

The authors declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2025.1720341/full#supplementary-material

References

Aarnes O. J., Abdalla S., Bidlot J. R., and Breivik Ø (2015). Marine wind and wave height trends at different ERA-Interim forecast ranges. J. Climate 28, 819–837. doi: 10.1175/JCLI-D-14-00470.1