The Evaluation of ERA5’s Applicability in Nearshore Western Atlantic Regions During Hurricanes—“ISAIAS” 2020

Abstract

Hurricanes cause significant destruction, disrupting transportation, and resulting in loss of life and property. High-precision marine meteorological data are essential for understanding hurricanes. ERA5 provides high temporal resolution and global coverage of analytical data; however, the accuracy of the data during hurricanes is uncertain. To investigate the applicability of ERA5 during hurricanes, this study used buoy data as reference values and assessed the applicability of ERA5 sea-surface wind speed (WS), sea-surface temperature (SST), and sea-surface pressure (SSP) during the 2020 Atlantic hurricane “ISAIAS” through spatial distribution and error analysis. The results indicate that there is a positive correlation and consistency between the trends of ERA5 and reference values. The average correlation coefficients for SSP, WS, and SST are 0.953, 0.822, and 0.607, respectively. Nearshore topography has a significant impact on data accuracy, resulting in greater errors compared to open-water areas. This study provides a theoretical basis for the application of ERA5 data during hurricanes.

1. Introduction

Hurricanes primarily form over the surfaces of tropical or subtropical oceans and are highly destructive natural disasters. The strong winds, storm surges, and heavy rainfall they generate pose a severe threat to coastal regions, particularly nearshore areas [1,2,3]. The western Atlantic Ocean is a high-risk area for hurricane activity [4]. Its nearshore regions are densely populated and economically active, making precise monitoring and forecasting of hurricane paths, intensity, and associated marine meteorological elements (such as wind fields, waves, and sea-surface temperature) critically important [5]. With global warming, the intensity of hurricanes at landfall has increased, leading to disasters such as heavy rainfall, flooding, and landslides, threatening the safety of people and property [6,7]. High-precision and wide-range oceanic meteorological data are crucial for reflecting hurricane intensity, paths, and impact reports in hurricanes [8,9]. Buoys provide high-precision single-point real-time data; however, their limited spatial coverage makes it challenging to capture the spatial and temporal characteristics of hurricane advancement, such as the wind-speed gradient. Reanalysis data offer wide coverage and continuity, focusing on the limitations of buoy data. Consequently, they have become the primary source for analyzing hurricanes and other extreme weather events [10,11]. Atmospheric reanalysis data assimilate a large amount of satellite data and conventional observations from the ground and upper atmosphere, offering advantages such as long time series and high resolution. They are used for weather and climate analysis and can serve as driving fields for climate models [4,12,13]. Since the 1990s, the United States, Europe, and Japan have successively launched reanalysis products; U.S. reanalysis products include NCEP/NCAR as well as the recently developed CFSR and CFSV2. The Japan Meteorological Agency (JMA) and the Central Research Institute of Electric Power Industry (CRIEPI) jointly completed JRA25 and its next-generation product JRA55. The European Centre for Medium-Range Weather Forecasts (ECMWF) was one of the earliest institutions to conduct data reanalysis research, with its reanalysis data having undergone four generations: FGGE, ERA-15, ERA-40, and ERA-Interim [14]. The fifth-generation reanalysis product (ERA5) from the European Centre for Medium-Range Weather Forecasts features a spatial resolution of 0.25° × 0.25°, encompassing a 9-fold enhancement compared to its predecessor, ERA-Interim, which had a resolution of 0.75° × 0.75°. The temporal resolution has been improved by a factor of six. This improvement allows for a more detailed grid of hurricane-affected regions [15,16]. ERA5 relies on numerical prediction models and the assimilation of observations from various sources, which may result in errors in its values, particularly due to inversion bias of satellite microwave remote sensing in regions with heavy precipitation. This is particularly important in extreme weather conditions like hurricane eyewalls and nearshore land–ocean transition zones, which require verification of ERA5’s applicability and reliability [17,18]. In recent years, numerous scholars have conducted extensive research on the applicability of reanalysis data at global and regional scales. Monney [19] compared ERA-40, ERA-Interim, and NCEP/NCAR reanalysis data with observational data and found that ERA-Interim provided a better description of winter temperatures than ERA-40 and NCEP/NCAR. Decker [20] evaluated reanalysis data from GSFC, NCEP, and ECMWF, among others, and the results showed that ERA-Interim’s near-surface temperature, wind-speed, and precipitation data had higher correlations with observational data than NCEP/DOE and CFSR. Wang [21] compared ERA5 and ERA-Interim precipitation data using buoy precipitation observations of Arctic sea ice from 2010 to 2016, finding that ERA5 performs better in the Arctic. Tarek [22] used ERA5 precipitation as input for a hydrological model and found that it had the same accuracy as observed values. Compared to ERA-Interim, ERA5 precipitation data effectively reduced bias and enabled more accurate hydrological simulations in most of North America. Lv [23] used ECMWF wind field data to study the characteristics of wind and wave conditions in the Bohai Sea. This study indicated that ECMWF wind field data underestimated wind speeds over the Bohai Sea, but the overall trends and wind directions were consistent with observational data.

This study evaluates the accuracy of sea-surface wind speed, sea-surface pressure, and sea-surface temperature for ERA5 by analyzing spatial distribution characteristics and conducting error analysis. Nine buoy station observations during the transit of hurricane “ISAIAS” No. 9 in the Atlantic Ocean in 2020 provide reference values. This study aims to verify the applicability and reliability of ERA5 during hurricanes. It provides data support for hurricane forecasting and post-storm reporting. In addition, it provides a scientific basis for strengthening regional meteorological models and ensuring the safety of lives, property, and the ecological environment in coastal areas.

2. Materials and Methods

2.1. Study Area

This study examines the western Atlantic Ocean, specifically the area between 85° W–55° W and 10° N–40° N, as illustrated in Figure 1. Figure 1 displays the study area, with the dashed box outlining its boundaries. The yellow color denotes land, while blue signifies the ocean, which includes the majority of the Caribbean Sea, a portion of the Gulf of Mexico, and approaches the east coast of the United States near the Bahamas. It is a region where hurricanes are frequently generated and active and located in the tropical–subtropical transition zone and significantly influenced by the Gulf Stream, with surface-water temperatures in the summer ranging from 28 to 30 °C [24,25]. These temperatures produce the necessary thermal conditions for tropical cyclone formation, and a crucial threshold for hurricane development is an SST of more than 26.5 °C [26]. Differences in topography between the western shelf zone (water depth < 200 m) and the eastern deep-water zone (water depth > 3000 m) result in notable variations in the propagation characteristics of hurricane-induced storm surges [27,28]. The semi-enclosed topography of the Caribbean Sea and the Gulf of Mexico leads to slower hurricane movement, extending the impact of strong winds and heavy rainfall [29]. The National Hurricane Center (NHC) reports that the Atlantic Ocean averaged 19.4 named storms annually from 2015 to 2024. In 2020, there were 31 named storms, 12 of which affected the area. This statistic highlights the significance of analyzing the applicability of ERA5 data, which effectively represent the accurate monitoring of hurricanes.

2.2. Data

2.2.1. Buoy Data

Buoy stations monitor WS, SST, and SSP elements in real time and serve as a reference for evaluating the accuracy of ERA5 due to their high precision and performance. The buoy data were sourced from the National Data Buoy Center (NDBC). Nine buoys in the nearshore region of the western Atlantic Ocean were selected during the “ISAIAS” event: 41004, 41008, 41010, 41029, 42059, 42060, MISP4, SPGF1, and VAKF1. Their spatial distribution is illustrated in Figure 2. Figure 2 displays the station locations marked by green dots, while the adjacent figure highlights the SPGF1 station’s location.

Table 1. Coordinates of the buoy site.

Bouy	41004	41008	41010	41029	42059	42060	MISP4	SPGF1	VAKF1
Lat (°N)	32.502	31.400	28.878	32.803	15.300	16.434	18.090	26.704	25.731
Lon (°W)	79.099	80.866	78.467	79.624	67.483	63.329	67.939	78.995	80.162

provides the specific coordinate information for each station. This study examines the impact of nearshore topography on ERA5 accuracy, focusing on the nearest land station, SPGF1, situated on the west side of Grand Bahama Island, 13.1 km from Freeport. This study focuses on the transit period of “ISAIAS” from 29 July 2020 to 5 August 2020. The data encompass wind speed, sea-surface pressure, and sea-surface temperature, with details provided in

Table 2. Data information.

Data	Style	Source	Spatial Resolution	Time	Temporal Resolution
SST WS SSP	Bouy	NOBC	/	29 July–5 August	1 h
SST WS SST	ERA5	ECMWF	0.25 × 0.25°	15 July–15 August
Hurricane information	/	NHC	/

European Centre for Medium-Range Weather Forecasts (ECMWF), (https://cds.climate.copernicus.eu/, accessed on 1 September 2024). National Data Buoy Center (NDBC), (https://www.ndbc.noaa.gov/, accessed on 1 September 2024). National Hurricane Center (NHC), (https://www.nhc.noaa.gov/, accessed on 1 September 2024).

.

2.2.2. ERA5 Data

The ERA5 reanalysis data were sourced from the European Centre for Medium-Range Weather Forecasts (ECMWF), with a spatial resolution of 0.25° × 0.25° and a temporal resolution of 1 hour. This dataset is developed by integrating data from various sources, including satellite remote sensing, airborne observations, ground-based meteorological stations, and ocean buoys [30]. The study period covers the time before, during, and after the transit of “ISAIAS” (15 July 2020 to 15 August 2020). The ERA5 data include WS, SSP, and SST, with detailed information provided in Table 2.

Hurricane “ISAIAS” was the ninth hurricane of 2020, originating as a strong tropical wave near the coast of Africa on 24 July. It advanced westward, influenced by a subtropical high-pressure ridge, and developed into a tropical storm in the Lesser Antilles on 29 July. The tropical storm progressed west–northwest, first making landfall in San Pedro de Macris, located in the southeastern Dominican Republic, on 31 July. It subsequently made a second landfall in the Bahamas before weakening to a tropical storm on 1 August. The tropical storm made its third landfall in Andros Island, subsequently moving north–northwest across the warm waters of the Gulf Stream and making a fourth landfall in North Carolina on 4 August, with maximum sustained winds of 80 knots (41.16 m/s). It then moved rapidly north–northeast through Virginia, Maryland, Delaware, Pennsylvania, New Jersey, New York, and Vermont before transitioning into a temperate depression in southeastern Canada on 5 August, ultimately dissipating over Quebec. It’s movement path is shown in Figure 3.The trajectory of “ISAIAS” passed through the western Atlantic nearshore, featuring four landfalls and notable variations in intensity (ranging from a tropical storm to a Category 2 hurricane), leading to considerable economic damage in the affected region, several fatalities, and over 3 million homes experiencing power outages [31,32].

2.3. Methodology

In this study, the accuracy of the WS, SSP, and SST of ERA5 during “ISAIAS” was evaluated by using the correlation coefficient (CC), root mean square error (RMSE), and mean deviation (MD) as the reference buoy data. CC is the degree of linear correlation between the study variables, which is used to characterize the consistency between ERA5 and the reference values in terms of trends. The judgment criteria include very strong correlation (CC ≥ 0.9), strong correlation (0.7 ≤ CC < 0.9), moderate correlation (0.5 ≤ CC < 0.7), and weak correlation (CC < 0.5) [33]. Its calculation formula is shown in Equation (1):

$$\mathrm{Correlation} \mathrm{Coefficient}(\mathrm{CC})={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\left({\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}-{\overline{\mathsf{\eta }}}_{\mathrm{E}\mathrm{R}\mathrm{A}5}\right)\left({\mathsf{\eta }}_{\mathrm{obs}}-{\overline{\mathsf{\eta }}}_{\mathrm{obs}}\right)}{\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}-{\overline{\mathsf{\eta }}}_{\mathrm{E}\mathrm{R}\mathrm{A}5}{)}^2}\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({\mathsf{\eta }}_{\mathrm{obs}}-{\overline{\mathsf{\eta }}}_{\mathrm{obs}}{)}^2}}}.$$

(1)

In Equation (1), ${\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}$ and ${\mathsf{\eta }}_{\mathrm{obs}}$ are ERA5 and buoy values, ${\overline{\mathsf{\eta }}}_{\mathrm{E}\mathrm{R}\mathrm{A}5}$ and ${\overline{\mathsf{\eta }}}_{\mathrm{obs}}$ is the average of ERA5 and buoy values, and “n” is the total.

The RMSE quantifies the deviation between predicted and true values, specifically measuring the difference between ERA5 and observed values. A larger RMSE indicates lower accuracy and greater error in the ERA5 data. The acceptable range thresholds for the RMSE for the WS, SST, and SSP were 1.5–2.0 m/s, 0.5–1.0 °C, and 0.5–1.0 hPa [34]. Equation (2) presents the formula.

$$\mathrm{Root} \mathrm{Mean} \mathrm{Square} \mathrm{Error}(\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E})=\sqrt{{\displaystyle \frac{1}{\mathrm{n}}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}-{\mathsf{\eta }}_{\mathrm{obs}}{)}^2}.$$

(2)

In Equation (2), ${\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}$ and ${\mathsf{\eta }}_{\mathrm{obs}}$ are ERA5 and buoy values, and “n” is the total.

MD represents the mean error, calculated as the difference between the original data value and the true value. It indicates the overall error of ERA5, highlighting whether there is an overestimation or underestimation. When the MD is greater than 0, ERA5 exceeds the reference value; when it is less than 0, ERA5 falls below the reference value. An MD of 0 signifies that ERA5 is accurate. The acceptable range threshold for MD is −1.0–1.0 [35]. The positive and negative values of MD, along with the size of ERA5 accuracy, can be utilized to validate the accuracy of ERA5 [35,36,37]. The calculation formula is presented in Equation (3):

$$\mathrm{Mean} \mathrm{Deviation}\left(\mathrm{M}\mathrm{D}\right)={\displaystyle \frac{1}{\mathrm{n}}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\left({\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}-{\mathsf{\eta }}_{\mathrm{obs}}\right).$$

(3)

In Equation (3), ${\mathsf{\eta }}_{\mathrm{E}\mathrm{R}\mathrm{A}5}$ and ${\mathsf{\eta }}_{\mathrm{obs}}$ are ERA5 and buoy values, and “n” is the total.

A comprehensive assessment of the applicability of ERA5 in hurricane monitoring is made through the above indicators, combined with a visual analysis of the spatial distribution.

3. Results

3.1. Accuracy Assessment of WS During “ISAIAS”

3.1.1. Spatial Distribution of WS

The analysis of the spatial distribution of WS is conducted before (15 July), during (31 July and 3 August), and after (15 August) “ISAIAS” using ERA5 data, as illustrated in Figure 4. Figure 4a–d illustrate the wind-speed direction with vectors, while the solid line denotes the actual path of “ISAIAS”. The color gradient from blue to red indicates the increase in wind speed, ranging from 0 to 22 m/s.

Prior to the hurricane’s arrival on 15 July (Figure 4a), the stable subtropical high pressure resulted in a lower-wind-speed area (0–6 m/s), with dispersed wind directions and no notable gradient changes. The atmospheric environment exhibits stability, consistent with the characteristics of typical summer trade winds. On 31 July (Figure 4b), a hurricane developed in the southeastern Bahamas (75° W, 22.5° N), creating a significant area of high wind speeds. This area exhibits a circulation structure centered around the hurricane’s eye, which aligns with the actual path of “ISAIAS”. The maximum wind speed exceeded 22 m/s, with vector arrows indicating a counterclockwise direction. The atmospheric environment remained stable, reflecting typical summer wind characteristics. The vector arrows indicate a counterclockwise convergence, forming a spiral shape around the center, where wind speed increases from the inside to the outside. The center features low wind speed, while the surrounding wall of clouds represents the high-wind-speed area encircling the eye of the hurricane, illustrating a typical hurricane structure. On August 3 (Figure 4c), as the hurricane progressed northward, the center of the high-wind-speed zone shifted offshore east of Florida (78.5° W, 27.5° N). The wind speed in this zone increased to 14–21 m/s, while the wind speed within the eye area was measured at 6 m/s, closely encircled by a ring-like high-wind-speed zone. This observation aligns well with the actual path of hurricanes reported by the NHC. Following the hurricane’s dissipation on 15 August (Figure 4d), the area of high wind speeds vanished. Offshore waters of North Carolina experienced slightly elevated wind speeds of 8–11 m/s, while other regions reverted to a consistent low-wind-speed state of less than 8 m/s, resulting in a calmer wind field. The WS of ERA5 effectively captures the significant changes in the wind field induced by hurricanes and the core structural features on a spatial scale.

3.1.2. WS Error Analysis

The analysis of the WS of buoys and ERA5 during “ISAIAS” is presented in Figure 5. The solid line represents the ERA5 data, while the dotted line indicates the buoy data. The horizontal axis denotes time in hours (072913 is 29 July at 13:00), and the vertical axis shows wind speed in m/s. The time series in Figure 5 demonstrates that the changes in ERA5 and observed values align closely, illustrating the wind speed’s sudden increase–peak–decrease during the hurricane’s passage (stations 42059 and 41004). The fluctuation trends of both data types are largely synchronous with the observed values from station 41004, indicating a significant positive correlation between them.

This study examines that nearshore topography influences the accuracy of ERA5 by comparing the data from stations SPGF1 (nearshore) and 41004 (open ocean), as illustrated in Figure 5a,b. Figure 5a indicates that the ERA5 wind-speed profiles align closely with the observations during the hurricane, particularly throughout the hurricane transit, showing a peak deviation of less than 1.5 m/s. Additionally, there is a notable consistency in the trend and fluctuations of ERA5 changes. Figure 5b illustrates that as “ISAIAS” began, buoy data indicated a rapid decline from 21.3 m/s, which resulted in a steep descent curve. In contrast, the WS curve of ERA5 shows a gentle decrease, which underestimates the wind speed by 4 m/s at the time of hurricane landfall and is accompanied by a phase lag of 1–2 h.

Figure 5c indicates that merely 8 out of 30 h share the same value, highlighting a discrepancy between ERA5 and the observed values. The examination of wind-speed maxima indicates an error in the peak value of ERA5, as illustrated in Figure 5e. Using station 41004 as an example, the buoy experiences a rapid increase in speed from 10 m/s to a peak of 25 m/s within 2 h as the hurricane approaches. This rising process exhibits a steep curve, indicating a significant increase in wind influence. In contrast, the ERA5 data reflect a similar upward trend, but the peak reaches only 17 m/s in Figure 5, which is same as that in six stations.

Using buoy data as a reference, CC, RMSE, and MD analyses were performed on ERA5 WS, and the results are shown in

Table 3. Correlation analysis of WS of ERA5 with buoy data.

Station	CC	RMSE	MD
42059	0.85	1.32	0.45
SPGF1	0.52	2.92	−0.99
41010	0.82	2.14	1.12
41008	0.90	1.74	0.75
41004	0.90	2.50	−0.38
41029	0.94	1.96	−0.51

. Table 3 indicates an average CC of 0.82 across all stations, with CC values reaching or exceeding 0.83 for all stations except SPGF1. Station 41029 exhibits a CC of 0.94, reflecting a very strong correlation with ERA5 wind speed. The nearshore SPGF1 station exhibits a correlation coefficient of 0.52, indicating a weak correlation. In contrast, the open-water stations (41029, 41004) demonstrate correlation coefficients ranging from 0.90 to 0.94, reflecting a very strong correlation. Error analysis indicated that open-water station 42059 exhibited the lowest error (RMSE = 1.32 m/s, MD = 0.45 m/s), remaining within the acceptable error limit (RMSE < 2.0 m/s). In contrast, SPGF1 recorded the highest RMSE (2.92 m/s, surpassing the acceptable threshold) and MD (−0.99 m/s), suggesting that nearshore topography significantly impacts ERA5 accuracy, leading to increased error.

3.2. Accuracy Assessment of SST During “ISAIAS”

3.2.1. Spatial Distribution of SST

Analysis of the spatial distribution of SST was conducted before (15 July), during (31 July and 3 August), and after (15 August) “ISAIAS” using ERA5 data, as illustrated in Figure 6.

Figure 6a–d illustrate the wind speed direction with arrows, while the solid line denotes the actual path of “ISAIAS”. The color gradient from blue to red indicates the increase in SST, ranging from 27 to 31 °C. Prior to the hurricane’s arrival (15 July, Figure 6a), a warm tongue influenced by the Gulf Stream (>30 °C, red band) extends northward from the Florida Straits, creating a distinct temperature front with a cold water mass (<26 °C, blue zone), exhibiting a clear spatial gradient. On 31 July, during the hurricane development period, “ISAIAS” passes through the Caribbean high-temperature zone, reaching the southeastern Bahamas. The warm tongue vanishes in the eastern Florida waters, while a cooling zone emerges along the hurricane track in the northern Dominican waters. On 3 August, as “ISAIAS” progressed northward, it reached offshore eastern Florida (78.5° W, 27.5° N), as shown in Figure 6c. In comparison to Figure 6a, the high-temperature zone surrounding Florida was no longer present in the hurricane’s path, and a cooling zone developed along the hurricane track (20° N–32° N, 72° W–32° N). Following the dissipation of the hurricane on 15 August (Figure 6d), a low-temperature zone created by the surface cold water persisted in the southern waters of North Carolina. Meanwhile, the high-temperature zone associated with the warm tongue of the Gulf Stream gradually recovered and expanded northward along the U.S. east coast. ERA5’s SST data can reflect the downward trend in sea-surface temperature under the influence of hurricanes but cannot reflect the spiral spatial structure characteristics of hurricanes.

3.2.2. SST Error Analysis

The statistical analysis of SST trends from ERA5 and buoy values during “ISAIAS” is presented in Figure 7. Figure 7 displays the ERA5 data as a solid line and the buoy data as a dashed line. The horizontal axis represents time in hours, with 072902 indicating 29 July at 02:00, while the vertical axis shows temperature in degrees Celsius.

Figure 7 shows that ERA5 maintains a consistent trend. However, it exhibits a notable error, with buoy values consistently higher than ERA5 (ERA5 is 1–2 °C lower than observed values). For instance, at the VAKF1 station (Figure 7c), during the hurricane, the buoy SST dropped to 29.8 °C while ERA5 decreased to 28.4 °C, indicating an underestimation of 1.4 °C by ERA5. Figure 7 illustrates the significant fluctuations in buoy value during the hurricane, in contrast to the more gradual changes observed in the ERA5 data. The response of ERA5 exhibits a lag of 1–5 h. For instance, in Figure 7c, the buoy begins a continuous decrease at 19:00, whereas ERA5 shows a decrease starting at 21:00, marked by a single fluctuation, indicating a 2 h lag. Figure 7b illustrates an abrupt decrease in the buoy’ s temperature from 30 °C to 28 °C over a short duration, resulting in a total drop of 2 °C, while ERA5 shows only a decrease of 0.2 °C. ERA5 sea-surface temperature (SST) is underestimated, and there is a delay in SST changes, as shown in Figure 7.

The SST of ERA5 was analyzed for the CC, RMSE, and MD using the buoy data as a reference, with results presented in

Table 4. Correlation analysis of SST and buoy data for ERA5.

Station	CC	RMSE	MD
42060	0.56	1.04	−1.03
MISP4	0.49	1.22	−1.10
VAKF1	0.76	2.07	−2.03
41010	0.74	1.18	−1.17
41004	0.73	1.57	−1.55
41029	0.36	1.34	−1.23

. Table 4 indicates a significant error in SST, with the CC for all ERA5 stations varying between 0.36 and 0.76. Notably, station 41029 has a CC of 0.36, complicating ERA5’s ability to represent the actual changes during the “ISAIAS” period. The ERA5 data range and RMSE for each station during “ISAIAS” are between 1.04 and 2.07. The overall deviation of ERA5 is significant, surpassing the acceptable threshold of 0.5–1.0 °C, particularly at station VAKF1, which shows an RMSE deviation of 2.07 °C. In the hurricane period, the MD values of ERA5 and the SST reference values from each buoy were negative, with ERA5 deviating from the reference values by an average of 1.03–2.03 °C. The reduced SST values of ERA5 relative to actual values could affect the evaluation of hurricane damage and undermine the precision. The SST during “ISAIAS” requires a correction of 1–2 °C for applications related to ERA5 in the western Atlantic nearshore.

3.3. Accuracy Assessment of SSP During “ISAIAS”

3.3.1. Spatial Distribution of SSP

The spatial distribution of SSP was analyzed using ERA5 data before (15 July), during (31 July and 3 August), and after (15 August) “ISAIAS”, as illustrated in Figure 8. Figure 8a–d illustrates the wind speed direction with arrows, the solid line indicates the actual path of “ISAIAS”, and the color gradient from blue to red signifies the increase in SSP (range: 1000–1020 hPa). Prior to the hurricane’s arrival on 15 July (Figure 8a), the SSP exhibited an even distribution, characterized by high pressure (subtropical high pressure) north of the Tropic of Cancer and low pressure (tropical low pressure) south of the Tropic of Cancer. This configuration corresponds with the typical barometric distribution patterns observed in the northern hemisphere during summer [27,28,29]. Pressure varied between 1010 and 1020 hPa, exhibiting a uniform distribution of isobars. On 31 July, “ISAIAS” developed and traversed the Caribbean Sea, reaching the southeastern Bahamas. A low-pressure area exhibited a blue spiral distribution, aligning with the actual path of “ISAIAS”. The center of this low-pressure area, representing the eye of the hurricane, was in the sea north of Cuba. The low-pressure center, or eye of the hurricane, was in the northern Cuban waters. It exhibited an anticlockwise distribution and a distinct spatial structure of the SSP gradient. On 3 August (Figure 8c), a distinct low-pressure zone developed along the hurricane’s path, revealing a clear hurricane structure characterized by a dense spiral of isobars radiating outward from the low-pressure center. Following the dissipation of the hurricane on 15 August (Figure 8d), a low-pressure area influenced by the hurricane persisted in the offshore waters of North Carolina. Meanwhile, the SSP in surrounding waters gradually returned to pre-hurricane conditions, characterized by sparse and uniformly distributed isobars. The SSP of ERA5 accurately reflects the spiral distribution characteristics and spatial gradient characteristics of sea level pressure affected by hurricanes.

3.3.2. SSP Error Analysis

The statistical analysis of SSP trends from ERA5 and buoy values during “ISAIAS” is presented in Figure 9. Figure 9a–h presents the ERA5 data as a solid line and the buoy data as a dashed line. The horizontal axis represents time, with an example being 072902 (29 July 02:00), while the vertical axis indicates barometric pressure in hPa. Figure 9 illustrates a decline in the SSP during the “ISAIAS” period. The ERA5 data trend aligns with the actual values, showing a similar rate of decline, which reflects the typical characteristics of the SSP. The actual value changes consistently, with a consistent rate of decline and typical “V” and “W” type variations. The curves overlap in most periods, indicating that ERA5 effectively reflects the characteristics of sudden drops in barometric pressure during hurricane transits, demonstrating high data reliability. Except for the SPGF1 station (Figure 9d), the errors at the other stations are ≤1.47 hPa. In the open ocean, using station 41029 as an example (Figure 9h), during the hurricane, the SSP of the buoy data decreased from 1014 hPa to 1000.9 hPa and subsequently increased to 1016 hPa, creating a “V”-type fluctuation. ERA5 also reached its minimum simultaneously, with the decreasing trend aligning with the observed values. Using the SPGF1 station as a case study in the nearshore area (Figure 9d), in the “ISAIAS” period, both ERA5 and buoy data exhibited “V”-type fluctuations. However, the minimum value recorded by ERA5 was 3.8 hPa higher than that of the buoy, and the timing of ERA5’s minimum occurred 3 h earlier. The minimum of ERA5 occurred 3 h earlier than that of buoy data, and this was consistent across observations.

The analysis of SST for ERA5 included the CC, RMSE, and MD, as presented in

Table 5. Correlation analysis between SSP and buoy data for ERA5.

Station	CC	RMSE	MD
42060	0.96	0.81	0.52
42059	0.94	0.75	0.25
VAKF1	0.97	0.62	−0.45
SPGF1	0.90	2.01	1.27
41010	0.95	0.88	0.51
41008	0.93	0.90	0.53
41004	0.99	0.92	0.89
41029	0.98	0.78	−0.08

. The CC thresholds for each station ranged from 0.90 to 0.99, indicating a strong correlation with ERA5. The CC is greater than or equal to 0.93 at all stations except SPGF1, showing highly consistent trends, with the CC reaching up to 0.99 at station 41004, where the SSP data for ERA5 matched the observation level. The RMSE threshold for each station ranges from 0.62 to 2.01, with an acceptable RMSE of ≤0.92 (0.5–1.0 hPa), except for SPGF1 (RMSE = 2.01 hPa). The absolute value of the MD for each station is ≤1.27 hPa, while for all other stations except SPGF1 (MD ≤ 0.89 hPa, indicating a minimal degree of error and deviation. The ERA5 SSP data during “ISAIAS” clearly show the influence of nearshore topography; however, the errors remain minimal and within acceptable limits, with some stations achieving observation-level accuracy.

SSP data from ERA5 accurately represent the spatial and temporal structure of the hurricane low-pressure center, including the “V”-shaped fluctuation and spatial pressure gradient. It demonstrates the highest accuracy in open-ocean areas, with a 3–4 hPa overestimation nearshore. Additionally, the SSP of ERA5 effectively captures significant changes in hurricane pressure and structural characteristics, serving as crucial data support for disaster warnings.

4. Discussion

Hurricanes are destructive meteorological events, and accurate forecasts rely on high-precision meteorological data. This study examines hurricane “ISAIAS” No. 9 from 2020, utilizing buoy data as measured values to assess the applicability of WS, SST, and SSP from ERA5. The analysis focuses on spatial distribution and error evaluation during the “ISAIAS”. ERA5’s SSP and WS accurately reflect the structural characteristics of hurricanes, such as low-pressure centers and hurricane-affected areas, on a spatial scale. In contrast, SST only reflects overall changes in sea-surface temperature within the study area. The WS data from ERA5 indicate the presence of a spiral high-wind-speed zone (maximum wind speed > 22 m/s) and counterclockwise circulation features in the hurricane eyewall region. The distribution of the wind field aligns with the position of the low-pressure center of the SSP (Figure 8). The SST of ERA5 indicates cooling changes along the hurricane path, such as the disappearance of high-temperature areas in the waters surrounding Florida (Figure 6). The low-pressure system at the hurricane center primarily governs the SSP variations. ERA5 features a high resolution of 0.25° × 0.25° and utilizes advanced data assimilation techniques [15,38,39,40,41,42,43,44,45,46]. It accurately reflects the “V”-shaped fluctuation and spiral isobar distribution of the hurricane low-pressure center.

This is consistent with the actual path of the hurricane, which exhibited significant pressure gradient characteristics off the southeast coast of the Bahamas and the east coast of Florida (Figure 8). Concerning error analysis, ERA5 demonstrates the highest SSP precision (CC: 0.90–0.99), indicating a strong agreement in the open ocean (station 41029, CC = 0.98, Table 5), and a CC of 0.90 for the nearshore station (SPGF1), both showing a very strong correlation. The SSP meets the acceptable threshold (0.5–1.0 hPa) for open-ocean RMSE ≤ 0.92 hPa, and the absolute value of MD ≤ 1.27 hPa, indicating a minimal error. The WS accuracy of ERA5 ranks among the second (CC: 0.52–0.94), influenced by surface roughness, topographic friction, and turbulent exchange [42,43,44,45,46,47,48].

The correlation coefficient reaches CC = 0.94 in the open ocean (station 41029) and drops to CC = 0.52 at offshore station SPGF1, indicating a weak correlation. The RMSE of WS is ≤2.50 m/s in the open ocean, and the RMSE is 2.92 m/s at the offshore station SPGF1, which exceeds acceptable thresholds. The MD of ERA5 is underestimated by 0.99 m/s at the offshore station (SPGF1) and overestimated by 0.75 m/s in the open ocean (41008). In 41008, the overestimation was 0.75 m/s, and the WS peak of ERA5 lagged by 1–2 h at certain stations. The SST precision of ERA5 exhibited the lowest correlation coefficient (CC: 0.36–0.76), reaching a minimum of 0.36 at station 41029, indicating a poor consistency in the trend of changes between ERA5 and the observed values. The RMSE of SST ranged from 1.04 to 2.07 °C, indicating a significant deviation, while the MD values were consistently negative, averaging an underestimation of 1.03 to 2.03 °C (Table 4, Figure 7). The ocean mixing process triggered by hurricanes causes surface seawater to rise, leading to a decrease in SST and the disappearance of high-temperature areas (Figure 8) [10,47,48,49,50,51].

In strong current areas like the Gulf Stream, SST fails to effectively represent the spatial distribution characteristics of hurricanes. ERA5′s accuracy arises from the influence of various geographical mechanisms and physical processes on each parameter [43,52,53,54,55,56]. Variations exist in the applicability of various parameters at the same station, along with discrepancies in the correlations observed at station 41004 (WS = 0.90, SST = 0.73, SSP = 0.99, in Table 3, Table 4 and Table 5). The error in ERA5 data increases notably as proximity to land increases (station SPGF1). The variation in water depth in the nearshore area of the western Atlantic (ranging from shallows less than 200 m to depths exceeding 3000 m), along with the peninsular topography of the Florida Peninsula, results in nonlinear dynamics affecting hurricane paths. ERA5’s resolution remains inadequate for accurately representing small-scale topography and terrain features such as islands and narrow straits [57,58,59,60,61]. In Table 3, the WS correlation shows a comparison between open-ocean sites (42059, CC = 0.85) and nearshore sites (SPGF1, CC = 0.52), indicating that the complex topography of land amplifies the data error.

The ERA5 WS peaks exhibited a lag of 1–4 h (Figure 5). The timing and magnitude of the WS peak observed at each station varied, as did the onset of the increase. The values remained consistent for a brief duration (Figure 7 and Figure 9), attributed to the spatial-temporal discrepancy between the data assimilation period and the hurricane’s velocity. Future developments should focus on creating a higher-frequency assimilation scheme or incorporating real-time observations, such as drone and ship data, to enhance timeliness [14,62,63,64,65].

5. Conclusions

This study analyzed the applicability of ERA5 during hurricane ISAIAS, using 2020 buoy data as a reference benchmark. The results showed that ERA5 was consistent with the reference values in terms of trend. The average correlation coefficients for SSP, WS, and SST are 0.953, 0.822, and 0.607, respectively. Additionally, the accuracy of ERA5 data is higher in open seas than in coastal areas. SSP and WS can reveal the spatial structural characteristics of hurricanes, including spiral-shaped high-wind zones and low-pressure centers.

ERA5 provides high-resolution and global ocean data, addressing issues such as the scarcity of observational data. ERA5 reanalysis data provide essential data support for assessing the impact of disasters during extreme weather events. This study is limited to a single hurricane, “ISAIAS.” Future research should be expanded to include various tropical cyclones to verify the general applicability of the research results. The results of this study will provide a scientific basis for disaster risk assessment.