Tropical Cyclones Related Wind Power on Oceanic Near‐Inertial Oscillations

Wind power input to oceanic near‐inertial oscillations (NIOs) plays a crucial role in sustaining the global ocean conveyor belt. However, the impact of tropical cyclones (TCs) on wind power input to NIOs, despite being the most vigorous atmospheric dynamics capable of exciting NIOs, is often overlooked in global estimations due to their transient nature and a lack of observations. Utilizing hourly wind and ocean current records, we quantified the wind power on NIOs induced by TCs from 1990 to 2019. Our findings reveal that the wind power on NIOs due to TCs is estimated to be between 0.028 and 0.065 TW, which accounts for a significant proportion, that is, 8%–17%, of that over the globe. This study highlights the importance of incorporating the wind power induced by TCs when estimating the global wind power on NIOs, as its impact is non‐negligible. Our findings contribute to a better understanding of the global energy balance by improving the estimation of wind power on NIOs.

• Observed data shows tropical cyclones contribute 8%-17% of global wind power on near-inertial oscillations • Tropical cyclones could contribute up to 90% of wind power on near-inertial oscillations in their prone regions • Accounting for tropical cyclones is essential when estimating wind power on near-inertial oscillations

Supporting Information:
Supporting Information may be found in the online version of this article.
wind power on NIOs along the track of TCs, reaching a maximum of 8 mW/m 2 , surpassing the annually averaged wind power induced by winter storms, which reaches a maximum of 5 mW/m 2 (Sun et al., 2021).Furthermore, the surface near-inertial energy flux induced by TCs has been found to be particularly efficient in promoting interior mixing (von Storch & Lüschow, 2023).
Various studies have been conducted to estimate the wind power on NIOs, yielding a range of values from 0.26 to 1.5 TW (Alford, 2003(Alford, , 2020;;Jiang et al., 2005;Rimac et al., 2013;Watanabe & Hibiya, 2002).However, recent studies have suggested that wind power on NIOs may be lower than previously assumed (e.g., Alford, 2020;Liu et al., 2019).Liu et al. (2019) specifically used observational data to estimate wind power on NIOs and found that numerical models tend to overestimate this input.Consequently, the relative contribution of TCs in terms of wind energy input may be more important, which has gained significant attention (e.g., Sun et al., 2021).The worldwide wind power on NIOs is expected to be underestimated by 10%-33%, when the wind power on NIOs induced by TCs is overlooked (Alford et al., 2016;von Storch & Lüschow, 2023).
The wind power on NIOs induced by TCs is estimated to be 0.026-0.07TW, accounting for 5%-27% of the global wind power on NIOs (Alford et al., 2016;Liu et al., 2008;Nilsson, 1995).The variation in results could be attributed to differences in methodology, wind product, or parameterization used in the models (von Storch & Lüschow, 2023).It is important to note that the estimation of TC-induced wind power on NIOs is challenging due to limited observations.Consequently, models are often employed to calculate the wind power input.However, the choice of different models can significantly impact the results (von Storch & Lüschow, 2023).Liu et al. (2008) also proposed several shortcomings of coupled models in simulating currents, including inaccurately initialized oceanic conditions, the absence of air-sea heat flux, and entrainment parameterization.To avoid the deficiencies of models, further studies based on observational current data are necessary to accurately quantify the contribution of TCs to global wind power on NIOs.However, the scarcity of observations during TCs has hindered a comprehensive quantification of the associated wind power on NIOs induced by TCs.
Based on the velocity profile observed during the passage of TC Gilbert, the TC-induced wind power on NIOs is estimated to be approximately 0.74 TW (Shay & Jacob, 2006).This estimate is one order of magnitude larger than the former estimates, indicating that observations from a single TC may not accurately reflect the global wind power on NIOs induced by TCs.This is because TCs can vary significantly in intensity and translation speed, leading to variations in the ocean response to TCs (e.g., Zhang et al., 2020).Estimates of near-inertial wind power induced by TCs based on worldwide observations are thus necessary to improve the estimation and gain a better understanding of air-sea interactions.
Velocity data sets from drifters have indeed been widely utilized to investigate the sea surface current response to TCs (Chang et al., 2013;Fan et al., 2022;Wang et al., 2022).These data sets can also be applied to estimate the global wind power on NIOs (Liu et al., 2019).Using ocean current records derived from surface drifters and wind vectors from analysis products, the climatological wind power on NIOs over the global ocean is estimated to be 0.3-0.6TW (Liu et al., 2019).However, TC-induced wind power on NIOs has not been considered in the global estimation (Liu et al., 2019), leading to an underestimation of the overall wind power on NIOs (Sun et al., 2021).
To obtain a comprehensive understanding of wind power on NIOs and its global contribution, it is crucial to quantify the values of TC-induced wind power on NIOs.
In this study, ocean current records from drifters are used to estimate the wind power on NIOs induced by TCs.
The main objective is to determine the relative contribution of TCs to the wind power on NIOs using observed ocean current records.The structure of the paper is as follows: Section 2 provides a description of the methodology and data used in the study.Section 3 presents and discusses the spatial distribution and relative contributions of TCs to the wind power on NIOs.Finally, Section 4 concludes the paper by summarizing the findings.

Surface Drifter Observations
This study uses velocity data tracked by the GPS and Argos system (Elipot et al., 2016) from 1990 to 2019 to estimate the wind power on NIOs induced by TCs.To ensure data quality, we discarded records from single drifters that lasted less than 300 hr, resulting in 159,411,472 sample records (16,373 drifters) being retained.
The spatial distribution of samples is shown as the number of records in each 2° by 2° grid (Figure 1a).Most of the boxes between 60°S and 60°N were well-sampled, making it sufficient to compute averaged TC-induced wind power on NIOs.However, the convergence zones, such as the subtropical gyres, are more frequently sampled compared to divergence zones, that is, equatorial, polar, and coastal regions, where fewer records were found.

Wind Reanalysis Data
To avoid errors caused by temporal interpolation (Jing et al., 2015;Liu et al., 2019), we use hourly ocean surface wind data from the fifth generation European Centre for Medium-Range Weather Forecasts atmospheric reanalysis of the global climate (ERA5) in this study, which has a spatial resolution of 0.25°.

TCs Best Track Data
We obtain TC data from the best track data set of the International Best Track Archive for Climate Stewardship (IBTrACS), which provides the 6-hourly TC locations and intensity information (Knapp et al., 2010).To ensure consistency with the drifter and wind data, we interpolated the TC data onto hourly time intervals via linear regression.We excluded hourly TC data with a maximum wind speed less than 17 m/s from the study.

Computation of Near-Inertial Wind Power
Following Liu et al. (2019), we compute the wind power on NIOs (W I ) as follows: where τ is the surface wind stress and u I is the sea surface near-inertial current.For each drifter observation, we linearly interpolate the simultaneous ERA5 wind onto the drifter position to compute the wind stress.The surface wind stress is evaluated as: where ρ a is the air density (1.22 kg/m 3 ), C D is the drag coefficient based on field experiments in TCs (Powell et al., 2003), U 10 is the wind vector at 10 m, and u is the ocean current vector.
The near-inertial current is calculated from drifter data using a band-pass filter.The near-inertial band is defined as 0.75f-1.25f,where f is the Coriolis frequency.Following Liu et al. (2019), velocity records tracked by each drifter are broken into half-overlapping 300-hr segments.For each 300-hr segment, the value of f is set as the mean value.

TCs Induced Near-Inertial Wind Power
The drifters are divided into two groups: drifters under and without the influence of TCs.We use these groups to calculate the TC-related wind power (W tc ) and the climatology wind power (W climat ).For each hourly TC position and time, the drifters under the influence of TCs must meet the following spatial and temporal criteria: (a) the distance from drifters to TC center should be within 500 km; (b) the drifter sampling time should be within the range from 3 days before the passage of typhoons to 20 days after typhoons (see Text S1 in Supporting Information S1 for the selection of the threshold).A significant number of samples (1,704,590) were used to derive the wind power.
The proportion of drifters under the influence of TCs (P tc ) shows peaks in the main TC-prone regions (Figure 1b), reaching more than 20% in the North Pacific.The total wind power on oceanic NIOs in each grid can be estimated as follows: where W tc *P tc is the contribution of TCs, W climat *(1−P tc ) is the contribution of climatology, and (W tc −W climat )*P tc is the additional contribution of TCs (W contc ), representing the value of underestimate when TCs are overlooked (W total −W climat ).

Spatial Variation of Wind Power on Oceanic NIOs
The total wind power (W total ) on oceanic NIOs with the influence of TCs reveals significant values in the 30-60° zonal band across both hemispheres (Figure 2a).The global peak value is 4.7 mW/m 2 , appearing in the Southern Ocean (Figure 2a).This pattern is in consistent with previous works (Alford, 2020;Liu et al., 2019), due to the passages of winter storms in these regions.The quasi-global integral (60°S to 60°N) of W total is 0.33 TW, which is comparable with Liu et al. (2019), indicating the reliability of the data and methodology employed in this study.
Considering the influence of TCs, the W total in TC-prone region of western North Pacific exhibits values exceeding 1.5 mW/m 2 , which is of the same magnitude as the global peak value (Figure 2a).In comparision, the value of W climat in the TC-prone region of western North Pacific is smaller than 0.5 mW/m 2 (Figure 2b).Most of the wind power on NIOs induced by TCs is concentrated in the TC-prone region of North Pacific, especially in the western section in accordance with the TC tracks (Figure 2c).The peak value of W contc in the western North Pacific is 1.5 mW/m 2 , which is approximately 10 times larger than the corresponding value of W climat in the same area, accounting for more than 90% of the relative contribution (W contr /W total ) of wind work on NIOs.Lob et al. (2021) also found that the energy flux during the storm event is about 7.5 times greater than the background energy flux.Therefore, TCs can be considered as a significant source of wind energy input, especially in TC-prone regions.However, the peak value of W contc was still significantly smaller than the results obtained from CESM simulations ( Sun et al., 2021) and coupled model simulations (Liu et al., 2008).These studies reported peak values exceeding 8 mW/m 2 in CESM simulations and approximately 3 mW/m 2 along the TC tracks in the coupled model simulations, respectively.This discrepancy is attributed to the underestimation of wind speeds from ERA5 during TCs, which will be discussed further in Section 3.2.
Prominent enhancement of wind power in the TC-prone region is identified in this study, which is not present in Liu et al. (2019).This difference can be largely attributed to the dissimilarities in the wind products used.Specifically, in the CCMP data set, the intensity of TCs is weaker at 6-hr intervals compared to the 1-hr intervals used in this study.The TCs contribute an additional 0.028 TW (8% of the total) to the wind power on NIOs based on ERA5 reanalysis, larger than that from CCMP (0.018 TW, 5%).Consistently, the enhancement of wind power on NIOs in the TC-prone region is also captured in the wind data set from the Modern-Era Retrospective Analysis for Research and Applications (MERRA) (Rienecker et al., 2011) with a temporal resolution of 1 hr, as indicated in Supporting Information of Liu et al. (2019).It has been documented that the near-inertial band wind stress variation would be partially filtered out when linearly interpolating 6-hourly reanalysis winds into hourly intervals (Jing et al., 2015;Niwa & Hibiya, 1999).This artificial loss of near-inertial wind stress variation leads to a significant underestimation of the wind power on NIOs (Jing et al., 2015).Therefore, it is vital to utilize wind data with high temporal resolution to accurately capture the near-inertial wind power on NIOs.
The quasi-global integral contribution of TCs to the wind power on NIOs (0.028 TW) is comparable to theoretical and model results (Liu et al., 2008;Nilsson, 1995).Because the occurrence of TCs is limited and the TCs induced wind power was generally underestimated, the integrated value of W contr is an order of magnitude smaller than that of W climat ; thus, the wind power on NIOs induced by TCs is oftenly overlooked (Liu et al., 2019).This discrepancy is partly due to the fact that the peak value of W contc is only one-third of the global peak induced by wind storms.Moreover, the influence of TCs is considerably less widespread compared to that of wind storms, which typically covers the entire midlatitude zone (Figure 2b).In contrast, the impact of TCs is primarily localized in specific TC-prone zones (Figure 2c).

Discussion
The precise measurements of sea surface near-inertial currents and surface wind speed is essential for accurate estimation of wind power on NIOs induced by TCs, that is, following Equations 1 and 2. In previous studies, models have been used due to limited observations for conducting direct estimation of TC-induced wind power (e.g., Alford, 2001Alford, , 2003;;Jiang et al., 2005;Rimac et al., 2013;Watanabe & Hibiya, 2002).The slab model of Pollard and Millard (1970) has been widely used in these studies, but it may lead to an overestimation of global wind work since it fails to consider the role of NIOs in deepening the mixed layer (Alford, 2020;Plueddemann & Farrar, 2006).A model that takes into account mixed layer deepening, such as the Price-Weller-Pinkel model, has shown better agreement with observed estimates of wind power on NIOs (Plueddemann & Farrar, 2006).However, the Price-Weller-Pinkel model also has limitations due to initial oceanic conditions and model parameterization, which can result in an overestimation of global wind work (see Figure 10 of Plueddemann & Farrar, 2006).These model limitations is neatly illustrated by the substantial variability in the results of near-inertial currents obtained from different model results (von Storch & Lüschow, 2023).
In this study, observated sea surface current data from global drifter measurements were utilized, which is expected to provide more accurate information and consequently lead to a more precise estimation of wind power on NIOs induced by TCs.However, it should be noted that there was a limitation in the surface wind speed data used in this study.Figure 3a illustrates the comparisons between the maximum wind speed calculated from ERA5 reanalysis data and the IBTrACS data.It is observed that the maximum wind speed calculated from ERA5 is generally smaller, ranging from 3% to 66%, compared to the IBTrACS data.On average, there is a mean difference of 22% between the two data sets.This discrepancy is consistent with previous studies that have shown that atmospheric reanalysis data tend to underestimate wind speeds during TC events to some extent (e.g., Liu & Sasaki, 2019;Vincent et al., 2012).
To improve the accuracy of the estimated wind field during TCs from ERA5 data, a crude adjustment of wind speeds during TCs was conducted by multiplying the ERA5 values by a factor of the mean difference between ERA5 and the reference data set (ERA5 *1.22 adjustment method).The estimated wind power value obtained using ERA5 *1.22 adjustment method are 0.051 TW, accounting for 14% of the quasi-global integral (60°S to 60°N) of W total .To test the reliability of the adjustment method, another two adjustments were conducted based on the idealized vortex spatial structure proposed by Holland (1980) and Willoughby et al. (2006) as described in Supporting Information S1.The estimated wind power values obtained using Holland and Willoughby adjustment methods are 0.054 TW and 0.053 TW, respectively.These estimated wind power values both account for 15% of the quasi-global integral (60°S to 60°N) of W total .The fact that the values derived from different adjustment methods are comparable, indicating that the methods used are reliable.
Besides the underestimate wind speeds, the uncertainty of the estimates can also be influenced by sampling errors.The observed TC-related wind power exhibits a similar pattern to the sampling proportion of drifters (Figure 1b) and the influential time of TCs (Figure 3b).This suggests that the averaged wind power on NIOs induced by TCs is largely dependent on the occurrence of TCs.Former studies identified a striped enhancement of wind power along specific tracks of TCs (Sun et al., 2021;von Storch & Lüschow, 2023).However, these studies were limited to a 1 year simulation period, which may have resulted in a more localized and fragmented representation of TC-induced wind power on NIOs.In constrast, this study benefits from a longer period of 30 years of TC data, allowing for a more comprehensive and objective assessment of TC-induced wind power on NIOs in space.As a result, the estimated TC-related wind power on NIOs in this study exhibits a smoother pattern, capturing the overall spatial distribution of TC-induced wind power on NIOs more accurately.
However, it is important to acknowledge that drifter observations cannot cover the entire time and space, leading to inherent randomness and potential sampling bias under the passage of TCs.The proportion of drifters under the influence of TCs (P tc ) may not accurately reflect the actual influencing time of TCs (Figures 1b and 3b).
Comparing to T tc (Figure 3b), the P tc (Figure 1b) exhibits evident over-sampling in certain regions such as Arabian Sea, the temperate zone of the western North Pacific, and many nearshore areas.Conversely, some regions, notably the South China Sea, are under-sampling.Considering that the wind power on NIOs induced by TCs is largely dependent on the actual duration of TC influence, it is crucial to address the sampling bias by utilizing T tc instead of P tc .By accounting for the actual duration of TCs, an improved estimation for the contribution of TCs from ERA5*1.22 adjusted data yields a value of 0.061 TW, accounting for 16% of the global near-inertial power input (Figure 4b).Similarly, the values derived from Holland and Willoughby adjusted The relative contribution of TCs to wind power on NIOs is significant in TC-prone regions, exceeding 40% in majority of these regions (Figure 4d).In fact, TCs can contribute up to 90% of wind power on NIOs in many area, especially in the Northern Hemisphere. Figure 4a illustrates the meridional distribution of the zonally integrated wind power.The maximum value of TC-related wind power on NIOs from ERA5*1.22 adjusted data is 0.0032 TW at 17°N, which is comparable to the maximum climatology wind power on NIOs of 0.0035 TW in the Northern Hemisphere and much greater than the value of 0.0008 TW at 17°N (Figure 4a).Consequently, TCs contribute 80% of the zonal integrated wind power at 17°N (Figure 4c).Notably, TCs could contribute nearly half (49%) of the integrated wind power between 30°S and 30°N (Figure 4c).Indeed, the findings highlight the significant role that TCs play in enhancing wind power and their substantial contribution to zonally integrated wind power in regions with concentrated TC activities, emphasizing their importance in shaping the overall wind dynamics in the global ocean.

Conclusion
This study provides an objective estimation of the wind power on NIOs induced by TCs using global observed ocean velocity data.The findings reveal a significant contribution of TCs to the near-inertial power input in the global ocean, accounting for 8%-17% of the total.It is anticipated that this contribution will increase in the future due to the ongoing strengthening of global TCs (Wang et al., 2022).Previous research has already found a 16% increase in near-inertial energy input from TCs between 1984 and 2003 (Liu et al., 2008).Therefore, the energy input from TCs will play an even more crucial role in the future, highlighting the importance of conducting further research to enhance our understanding of the role of TCs in the wind power input on NIOs.
Furthermore, in localized regions prone to TC, TCs play a dominant role in the wind power input on NIOs.This dominance has the potential to influence the local climate system by affecting the mixed layer depth during TC seasons.Since NIOs contribute to the enhancement of shear at the base of mixed-layer (e.g., Whalen et al., 2020), the wind power on NIOs induced by TCs can significantly deepen the local mixed layer depth in TC-prone regions during TC seasons.These effects, in turn, can have implications for the local climate system.Overall, this study underscores the importance of utilizing accurate wind speed data and emphasizes the need for further research to enhance our understanding of the contribution of TCs to the wind power on NIOs.By gaining a more comprehensive understanding of these dynamics, we can improve climate modeling and prediction, leading to better-informed decision-making processes for managing the impacts of TCs on regional climate systems.

Figure 1 .
Figure 1.(a) Spatial distribution of drifter samples in 2° × 2° grids represented by the number (log 10 ) of samples.(b) Proportion (in percent) of drifters under the influence of TCs in each grid.

Figure 2 .
Figure 2. The spatial distribution of wind power (W/m 2 ) on oceanic NIOs (a) with the influence of TCs and (b) without the influence of TCs computed from the drifters and ERA5 data sets.(c) Contribution of TCs to near-inertial wind power.

Figure 3 .
Figure 3. (a) Comparisons of maximum wind speed (m/s) calculated from ERA5 with corresponding results of IBTrACS.(b) Proportion of influencing time of TCs (T tc ) calculated from IBTrACS data.

Figure 4 .
Figure 4. (a) Meridional distribution of zonally integrated wind power on NIOs with (ERA5*1.22,dashed red) and without (Climatology, solid black) the influence of TCs.(b) Additional contribution of TCs to wind power (W/m 2 ) on NIOs (W contc ).(c) Meridional distribution of percentage of TC-related wind power on oceanic NIOs (W contc /W total ).(d) Percentage of TC-related wind power on oceanic NIOs (W contc /W total ).The value of W contc is estimated by considering the influence of the actual duration of TCs, based on the wind fields from ERA5*1.22 adjustment.