 Research article
 Open access
 Published:
Wind stress curl as a driving force of annual waves in the upper ocean for interpreting energetics at all latitudes
Progress in Earth and Planetary Science volume 11, Article number: 29 (2024)
Abstract
The dynamics of waves and eddies in the upper ocean plays an important role in the climate variation of tropical and subtropical regions. Previous diagnoses for annual Rossby waves in oceanic model outputs manifested zonally alternating signals (ZASs) in the timeaveraged distributions of wind input as well as pressureflux divergence terms in the budget equation of wave energy. This is the case when the annual mean of the wind input is estimated as the inner product of simulated velocity vector and wind stress vector in previous studies. The present study proposes a new mathematical expression for estimating the wind input that is analogous to one derived from the quasigeostrophic potential vorticity equation. Namely, the wind input is estimated as the negative of the product of pseudostreamfunction and wind stress curl, the latter of which is associated with the horizontal divergence of Ekman velocity. This can be interpreted as replacement of kinetic energy input with gravitational potential energy input. Pseudostreamfunction in the present study is inverted from Ertel’s potential vorticity anomaly and is seamlessly available at all latitudes. This contrasts with the quasigeostrophic streamfunction which is singular at the equator. The new expression enables reducing ZASs in the horizontal distributions of both wind input and pressureflux divergence terms, without harming the qualitative advantage of energy flux vectors to indicate the group velocity of waves at all latitudes.
1 Introduction
Variations of surface currents in the ocean are influenced by both intrinsic dynamics and wind forcing. The input of kinetic energy through wind forcing plays a significant role in driving ocean currents and turbulent mixing within the surface mixed layer (Ferrari and Wunsch 2009). The spatial variability of winds results in Ekman pumping and suction, which affect the vertical motion of water below the surface mixed layer of the ocean. On the other hand, temporal variations in wind forcing generate both inertiagravity waves and planetary waves within the ocean. The budget of mechanical energy in the upper layer of the global ocean was estimated in Lueck and Reid (1984). They considered the sources and sinks of energy, including wind input, heat forcing, and viscous dissipation. Their analysis revealed that the wind input is 100–500 times greater than the input of available gravitational potential energy resulting from heat forcing and precipitation/evaporation.
Previous studies put forward two mathematical expressions for estimating the energy input from wind forcing. One is the classical expression (Wunsch 1998) which is based on the budget of kinetic energy to calculate the wind input by taking the dot product of surface velocity vector and wind stress vector. Using this expression, Scott and Xu (2009) estimated that the global energy input from wind to geostrophic flows at the sea surface was 0.85–1.0 TW (1 TW = 10^{12} watts). This may be compared with the result of Alford (2001) who estimated that the energy input from wind forcing to nearinertial waves in the upper ocean amounted to about 0.29 TW. Wang and Huang (2004) estimated the wind energy input to the Ekman layer as 3 TW. Xu et al. (2016) used the classical expression to estimate the wind input to mesoscale eddies in the ocean, revealing a significant suppression of oceanic eddies by atmospheric winds. Another approach for estimating the wind input involves quasigeostrophic dynamics. In this expression, the wind input is calculated as the negative of the product between the quasigeostrophic streamfunction (whose vertical variation is associated with buoyancy) and the wind stress curl (which captures the horizontal divergence of Ekman velocity). This second expression was used in some studies such as the paper by Berloff and McWilliams (1999a). They investigated the spatial and temporal variability of midlatitude oceanic gyres generated by steady winds in 1.5layer and 2layer quasigeostrophic models to study a marginal stability problem, highlighting the instability of steady western boundary currents at moderate Reynolds numbers.
As a first step, the present study focuses on climatological wind forcing on ocean currents. Many studies noted the effect of wind stress curl on annual baroclinic Rossby waves (RWs) originating from eastern boundary regions at midlatitudes. Krauss and Wuebber (1982) suggested that wind stress curl in the eastern boundary regions of the North Atlantic Ocean effectively drives RWs. Cummins et al. (1986) investigated how RWs are influenced by the annual cycle of wind stress curl in the North Pacific Ocean. They noted clear wave crests and troughs in the anomaly field of sea surface height (SSH) in regions between 20° and 44° N, indicating that RWs originated from the eastern boundary regions. While the crests and troughs of SSH near the eastern boundary were aligned parallel to the meridional direction, a southeastnorthwest inclination was exhibited further away from the boundary. As noted by Schopf et al. (1981), the alternation in direction could be attributed to variations in wave speed with latitude depending on the deformation radius change. The presence of high meridional wavenumbers contributes to the generation of meridional group velocity. In the Southern Hemisphere, Reason et al. (1987) demonstrated using a numerical experiment that annual RWs are generated by the maxima of wind stress curl in three regions at 25° S or 38° S around the eastern boundary.
We note that the estimation of wind input rates for oceanic annual waves, as demonstrated in the Indian Ocean by Li and Aiki (2020), exhibited zonally alternating signals (referred to as ZASs in the present manuscript) that can complicate the understanding of the annual mean energy budget. These ZASs occur near the eastern boundary when the period of RWs is close to that of wind forcing, which is often unavoidable in model experiments using annual wind forcing. The objective of this study is to investigate the presence of ZASs in wind input across other oceans, explore the factors contributing to ZASs, and propose a method to mitigate their effects for a more comprehensive understanding of upper ocean dynamics. This investigation will be valuable for future studies aiming to gain insight into airsea interactions from an energy perspective. The structure of this manuscript is as follows. In Sect. 2, we present the results of a model experiment utilizing a shallowwater equation system associated with the first baroclinic mode forced by climatological wind forcing in the Pacific Ocean. In Sect. 3, we examine mathematical expressions for estimating wind input, modifying the seamless diagnostic scheme to facilitate the analysis of annual wave energy transfer and budget in the upper layer of the Pacific Ocean. Finally, Sect. 4 provides a summary of our findings and highlights directions for future research.
2 Materials and methods
We have performed a 20year experiment in the Pacific Ocean using a shallow water equation system associated with the first baroclinic mode. Investigation for the second and greater baroclinic modes should be done in a future study following the result of the present study. A total of 2400 snapshots have been stored in 20 years with an interval of 3.04 (= 365/120) days. The model integration has reached a climatological equilibrium within 20 years, and all 120 snapshots in the last year have been used in the following investigation. The present study has used the monthly climatology of wind forcing with subtracting a time mean component, as in Li and Aiki (2020), to investigate the dynamics of annual waves. We shall show for the Pacific Ocean that ZASs appear in the energy input by wind forcing.
2.1 Model setup
The model domain of the present study is from 50° S to 60° N and from 120° E to 70° W with realistic coastlines. The model equations are discretized in spherical coordinates using Arakawa’s Cgrid with a halfdegree resolution in both the zonal and meridional directions. Basic equations in spherical coordinates are shown in Appendix A. All equations in the main body of the present manuscript are written in Cartesian coordinates for simplicity, where the variables \(x\) and \(y\) denote the zonal and meridional directions, respectively, with values increasing eastward and northward. We have simulated the time evolution of zonal velocity \(u\), meridional velocity \(v\), and layer thickness \(h\) using a shallowwater equation system to read,
where the symbol \(f\) is the Coriolis parameter. The symbols \(F^{u}\) and \(F^{v}\) in Eqs. (1a) and (1b), respectively, are eddy viscosity terms as represented using the Smagorinsky et al. (1965) scheme. The symbols \(\tau_{{{\text{wind}}}}^{x}\) and \(\tau_{{{\text{wind}}}}^{y}\) in Eqs. (1a) and (1b) represent climatological wind stress anomalies in the zonal and meridional directions, respectively, in the Pacific Ocean. These have been derived from the monthly climatology of ECMWF ORAS3 wind stress in 1959–2009. Here we have subtracted the time mean component of wind stress to yield the model forcing. The symbol \(F^{h}\) in Eq. (1c) represents forcing for layer thickness and is set to zero in the model setup. This thickness forcing is revisited later in the manuscript to indicate expressions for the source/sink of gravitational potential energy.
Gravity wave speed \(c\) in Eqs. (1a) and (1b) has been estimated to be 2.71 m/s for the first baroclinic mode using the vertical profiles of annual mean salinity and temperature from the World Ocean Atlas (WOA). In Eqs. (1a)–(1c), the symbol \(H_{1}\) = 1107.5 m is the reference thickness associated with the first baroclinic mode, estimated as \(H_{1} = \frac{{\sqrt {H_{{{\text{mix}}}} H_{{{\text{bottom}}}} } }}{\alpha }\) based on the vertical mode decomposition. The nondimensional parameter, \(\alpha\) = 0.39, has been used to represent the windinduced momentum input to the mixedlayer for the first baroclinic mode. See Appendix A in Li and Aiki (2020) for the details of the vertical mode decomposition. The climatological annual mean of the mixedlayer depth, \(H_{{{\text{mix}}}}\) = 45.5 m, has been estimated based on a temperature difference of 0.2 °C using Argo data in the tropical Pacific Ocean, and the symbol \(H_{{{\text{bottom}}}}\) = 4100 m is the reference depth from the sea surface to the bottom of the tropical Pacific Ocean.
The zonal phase velocity of equatorial waves on an equatorial \(\beta\)plane may be written as
where \(\omega\) is wave frequency, \(k\) is zonal wavenumber, \(n\) is meridional mode number, and \(\beta = \frac{\partial f}{{\partial y}}\) which is Rossby parameter (Matsuno 1966). Substitution of \(n =  1\) to Eq. (1d) yields \(\omega /k = c\) representing equatorial Kelvin waves (KWs). Substitution of a set of \(n = 1\) and \(\omega = k = 0\) to Eq. (1d) yields \(\omega /k =  c/3\) representing long Rossby waves in the first meridional mode. This corresponds to the observational result of oceanic RWs (Chelton and Schlax 1996). Let \(L = 15,000\) km be the zonal distance of the Pacific Ocean at the equator from 145° E to 80° W. It takes \(\frac{L}{c}\) and \(\frac{3L}{c}\) for equatorial KWs and long equatorial RWs, respectively, to travel eastward or westward in the Pacific Ocean basin. The sum of these periods, \(\frac{L}{c} + \frac{3L}{c}\), is referred to as the basin mode period and is estimated to be 0.70 years for the first baroclinic mode.
2.2 Basic results
We explain the seasonal sequence of simulated quantities in the last year of the climatological experiment. Figure 1 shows snapshots in the middle of January, April, July, and October. The signals of zonal velocity (color shading in the left panels of Fig. 1) deflect toward lower latitudes as RWs propagate westward. The zonal wind stress (contours in the left panels of Fig. 1) is much stronger in January and July than in April and October, indicating the significant influence of the monsoon in Southeast Asia and Northeast Australia. In January, the wind stress is westward at 0°–20° N and 40°–20° S, and eastward at 20°–50° N and 20° S–0°, while the wind direction is reversed in July.
Geopotential is written as,
whose signals are more widely spread in the meridional direction of the Pacific Ocean (as shown by color shading in the right panels of Fig. 1) than that of zonal velocity. This indicates that gravitational potential energy is greater than kinetic energy in mid and highlatitude regions, that is consistent with the quasigeostrophic dynamics. The ZASs of geopotential in the eastern boundary regions at midlatitude are generated by wind stress curl corresponding to the simulation result of Cummins et al. (1986). The variations in the group velocity depending on the deformation radius lead to the meridional propagation of geopotential (Schopf et al. 1981). In January, a negative and strong geopotential anomaly area is generated in lowlatitude regions, as indicated by the red box in Fig. 1b. The negative signals propagate to the western boundary regions in July.
Potential vorticity anomaly may be written as
Contours in the right panels of Fig. 1 represent pseudostreamfunction which is derived by inverting \(q\) using the following equation,
where \(\nabla^{2}\) is the horizontal Laplacian operator. This equation has been solved using the Successive OverRelaxation (SOR) method (Perrin 1997) with a coastal boundary condition of \(\varphi = 0\). Note that pseudostreamfunction \(\varphi\) is available at all latitudes, in contrast to quasigeostrophic streamfunction which may be written as \(\frac{p}{f}\). The signals of pseudostreamfunction in Fig. 1 are positively correlated with that of geopotential in the Northern Hemisphere and negatively correlated in the Southern Hemisphere. The significant anomalies of pseudostreamfunction are concentrated in the western equatorial regions.
Figure 2 shows a set of Hovmöller diagrams in both midlatitude and equatorial regions. Zonal velocity (color shading in the left panels of Fig. 2) exhibits clear westward propagating signals at both 35° N and the equator, indicating the influence of RWs. While zonal velocity at 35° N manifests a seasonal cycle, it takes about 2 years to travel from 130° to 140° W. On the other hand, zonal wind stress (contours in the left panels of Fig. 2) at 35° N as well shows a seasonal cycle. Zonal wind stress is positive from November to April, and negative from May to October. These are revisited later in the manuscript. At the equator, the regions of greater magnitude in zonal velocity appear on the western side of the basin where zonal velocity is eastward from June to November and westward from December to May. Geopotential (color shading in the right panels of Fig. 2) as well shows westward propagating signals. At 35° N, geopotential and pseudostreamfunction (contours in the right panels of Fig. 2) both show a seasonal cycle with a similar period as zonal velocity. At the equator, the regions of greater values in pseudostreamfunction anomaly are found in the western Pacific Ocean. These are consistent with the model results of Qiu (2003). We note that correlation between geopotential and pseudostreamfunction is not found in equatorial regions where geostrophy is not robust. Pseudostreamfunction at the equator may capture the signals of mixed Rossbygravity (Yanai) waves associated with the semiannual component of wind forcing, to be explored in a future study following the spectral analysis approach of Song and Aiki (2021).
3 Results and discussion
We investigate the budget of wave energy in this section. All results in this section have been derived by taking a time average of 120 snapshots (except for those otherwise noted) through the last year of model integration.
3.1 Classical expression
Manipulation of shallow water Eqs. (1a)–(1c) yields an equation for the budget of wave energy to read,
where the second line is pressureflux divergence, the third line is wind input rate, and the last line is dissipation rate (as well as buoyancy production rate). In this classical expression, the wind input is estimated as the inner product of simulated velocity vector and wind stress vector, as explained in Sect. 1 of this manuscript. The third line of Eq. (3a) can be interpreted as the input of kinetic energy through wind forcing. This expression of the wind input has often been used in previous studies (Wunsch 1998; Scott and Xu 2009; Ferrari and Wunsch 2009; Zhai et al. 2012; Renault et al. 2016). See Aiki et al. (2016) for the separate budget equations of kinetic energy and gravitational potential energy to validate expressions for the forcing for each.
We have plotted the timeaveraged distribution of some terms in Eq. (3a) to investigate the energy budget of annual waves in the Pacific Ocean. Despite representing the annual mean, Fig. 3a, b manifests signindefinite signals in the wind input and the pressureflux divergence, respectively, that we refer to as ZASs. The patterns of ZASs in both hemispheres converge closer to the equator as moving from east to west. The dissipation rate in Fig. 3c exhibits greater values in the western equatorial region. We note that the classical expression of the wind input yields distinct ZASs in regions near the eastern boundary around 30°–40°N and 40°–30° S in the Pacific Ocean. The results correspond to McCreary et al. (1987) that the annual RWs vanish around 40°N in the North Pacific Ocean and the observations (Qiu et al. 1997). In the Northern (Southern) Hemisphere, the isolines of these signals are in the northeastsouthwest (southeastnorthwest) direction that is symmetric about the equator. In the Indian Ocean as well, ZASs appear in the wind input when estimated using the classical expression (Li and Aiki 2020). Li and Aiki (2020) did not conduct an indepth investigation into the origin of ZASs, apart from their general speculation that ZASs in the Indian Ocean were believed to result from a phase relationship between ocean flows and wind forcing. The present study focuses on these signals that may disturb the understanding of the energy budget of annual waves.
To investigate the origin of these signals, we plot a Hovmöller diagram in Fig. 4a at 35° N between 140° and 130° W, representing a typical region for ZASs. The time series of the classical wind input rate show westward shifting with a seasonal cycle. The pattern of the wind input originates from a phase relationship between ocean flows and wind forcing, as illustrated in Fig. 2a for the zonal component. Taking the time average of signals in the red box areas of Fig. 4a yields positive values. Likewise, taking the time average of signals in the green box areas of Fig. 4a yields negative values. These lead to ZASs in the annual mean field of the wind input.
3.2 The modified (phaseresistant) expression of the present study
As explained in Sect. 1 of this manuscript concerning the quasigeostrophic dynamics, wind input may also be estimated as the negative of the product of streamfunction and wind stress curl (Scott and Straub 1998; Berloff and McWilliams 1999a, 1999b). Streamfunction in the quasigeostrophic dynamics is associated with (the vertical integral of) temperature anomaly (through the hydrostatic balance). Wind stress curl is associated with the horizontal divergence of Ekman velocity (and thus the vertical convergence of velocity through the incompressibility condition). The product of streamfunction and wind stress curl is closely linked to the budget of gravitational potential energy, understanding vertical integration by parts. This should not be confused with whether an injected energy is efficiently stored in a kinetic energy reservoir or a gravitational potential energy reservoir according to relationship between wavelength and the deformation radius.
With an intent to introduce the idea of the quasigeostrophic expression mentioned above, we propose to modify the wind input term as in the third line of the following equation to read,
where \({\text{curl}}_{{{\text{tau}}}} = \partial_{x} \tau_{{{\text{wind}}}}^{y}  \partial_{y} \tau_{{{\text{wind}}}}^{x}\) is wind stress curl. Pseudostreamfunction \(\varphi\) has been defined in Eq. (2b) of the present manuscript, and \(\varphi_{x} = \partial_{x} \varphi\) and \(\varphi_{y} = \partial_{y} \varphi\). Pressureflux divergence term at the second line of Eq. (3b) has also been modified from that in Eq. (3a) to maintain mathematical consistency. The third line of Eq. (3b) represents the modified version of the wind input term, following the expression in the quasigeostrophic literature. In contrast to quasigeostrophic streamfunction which may be written as \(\frac{p}{f}\), pseudostreamfunction \(\varphi\) is available at all latitudes that is a cornerstone in the present study.
Figure 5a, b shows the modified wind input and the modified pressureflux divergence of annual wave energy estimated by the modified expression in Eq. (3b). The modified wind input (Fig. 5a) becomes more concentrated with small anomaly areas at the equatorial region. The modified pressureflux divergence (Fig. 5b) exhibits a similar pattern to the modified wind input. It can be said that the new (modified) expression provides a phaseresistant feature for estimating both the wind input and the pressureflux divergence.
Terms on the third line of Eq. (3b) may be better explained if we decompose the horizontal component of velocity as,
where superscripts “ekm”, “geo”, and “igw” stand for velocity components associated with Ekman flow, geostrophic flow, and inertiagravity waves, respectively.
One way to define the Ekman velocity is to use the reference thickness to read,
which is directed perpendicular to wind forcing vector. The result is that there is no wind input to the Ekman component of oceanic velocity in a singlelayer framework,
which is consistent with a classical framework with vertical profiles wherein wind input to Ekman spiral flow at the sea surface (Wang and Huang 2004) is canceled in sharp by the vertical integral of dissipation associated with turbulent viscosity, as detailed in Appendix B of the present manuscript.
Geostrophic velocity is defined in the present study using pseudostreamfunction to read,
which is available at all latitudes. Geostrophic velocity satisfies the following,
where the second line has been derived using Eq. (6a), the third line has been derived using \(\varphi \approx \frac{p}{f}\), the last line has been derived using Eq. (5a). This makes it possible for the sum of additional terms in the modified expression to offset the wind input rate in the classical expression in such a way as to reduce ZASs in the annual mean estimation. The last line of Eq. (6b) is analogous to the last term of Eq. (3b), allowing for
to be interpreted as a variant of thickness forcing \(F^{h}\) which leads to the source/sink of gravitational potential energy.
On the other hand, velocity associated with inertiagravity waves, coastal KWs, and equatorial KWs is characterized by no perturbation of potential vorticity as well as pseudostreamfunction,
where \(q\) has been defined in Eq. (2a). Substitution of Eq. (7a) to the third line in Eq. (3b) yields,
which retrieves the classical expression of wind input.
To summarize, it is only for geostrophic velocity that the modified expression of wind input can be considered as representing the input of gravitational potential energy while canceling out the input of kinetic energy. To investigate the elements of reducing ZASs, we plot Fig. 4b, the Hovmöller diagram of
which is the sum of additional terms of the modified wind input expression with respect to the classical expression. The sum of additional terms in the modified expression manifests evident consistency with the westward shifting pattern of the classical expression, except that they have opposite signs.
To investigate the details of the ZAS reduction in the Pacific Ocean, we have estimated the magnitude of the wind input associated with the classical expression using
as shown in Fig. 6a where the areas of greater orders in magnitude near 30°–40°N and near the coastlines of the North American continent, corresponding to regions where ZASs are significant. We have also estimated the magnitude of the modified expression using
as shown in Fig. 6b. Both terms (9a) and (9b) are in a unit of W/m^{2}. Signals in Fig. 6b are smaller by one to two orders of magnitude than that in Fig. 6a in the mid and lowlatitude regions of the Pacific Ocean, especially in the regions of ZASs. To summarize the phaseresistant feature is given by cancelation within each of the second and last terms on the lefthand side of Eq. (6b) as shown by its first equality. Such remedy is not required for the product of streamfunction and wind stress curl.
3.3 Validation of the modified energy flux
We have plotted energy fluxes to help better understand the annual mean budget of wave energy in the Pacific Ocean. There are three schemes for illustrating the transfer routes of wave energy based on the group velocity, as explained in Appendix C of the present manuscript. Below we show the distributions of the energy fluxes based on the three schemes, as their performance in the Pacific Ocean has not been compared in the previous studies.
Figure 7 shows energy flux vectors associated with the pressureflux scheme (Cummins and Oey 1997), and the OrlanskiSheldon scheme (Orlanski and Sheldon 1993, hereafter OS93 scheme) following expressions (16a) and (16b), respectively, in Appendix C of the present manuscript. The energy fluxes of both schemes are compatible with the wave energy budget Eq. (3a). Figure 7a for the pressureflux scheme manifests westward flux at the equator and eastward flux in regions between 8° S and 8° N. This scheme assumes only inertiagravity waves but not midlatitude RWs. On the other hand, Fig. 7b for the OS93 scheme is able to display the westward transfer of wave energy by offequatorial RWs. The distinguishing feature of the OS93 scheme is the lack of values in the vicinity of the equator and coastlines, associated with equatorial and coastal KWs, respectively. Figure 7 illustrates the lack of validity in the pressureflux and OS93 schemes for determining the group velocity of annual waves in the Pacific Ocean. This result is in line with similar diagnoses for annual waves in the Indian Ocean (Li and Aiki 2020) and the Atlantic Ocean (Song and Aiki 2020).
Figure 8a shows the annual mean distribution of the energy flux as estimated by the AikiGreatbatchClaus scheme (Aiki et al. 2017, hereafter AGC17 scheme), following expression (16c) in Appendix C of the present manuscript. The AGC17 energy flux is compatible with the classical wind input in the wave energy budget Eq. (3a). The AGC17 energy flux (Fig. 8a) can reveal both the westward flux of energy associated with midlatitude RWs and the eastward flux of energy associated with equatorial KWs. In the Pacific Ocean, equatorial KWs traverse toward the eastern boundary, undergo reflection, and subsequently diverge into two packets of westward RWs. These wave packets then coalesce into a single packet of westward propagating waves at 160° W as indicated by Toyoda et al. (2021, 2023). We need to modify the AGC17 energy flux in such a way as to be compatible with the modified wind input in the wave energy budget Eq. (3b). This is referred to as the modified AGC17 scheme (hereafter, AGC17M scheme), following expression (16d) in Appendix C of the present manuscript. Figure 8b shows the annual mean distribution of the AGC17M energy flux. The striking resemblance between Fig. 8a, b implies that modifying both the wind input and the energy flux vectors has minimum impact on the qualitative superiority of the AGC17 scheme in terms of seamless connection between equatorial and offequatorial regions.
Figure 9 shows the time evolution of zonal flux and wind input rate during the last year of the model experiment at both 8° N and the equator. This figure is designed to facilitate a more comprehensive evaluation of the disparities between the AGC17 energy flux (left panels of Fig. 9) and the AGC17M energy flux (right panels of Fig. 9). Zonal flux (color shading) in all panels of Fig. 9 exhibits the evident signals of westward transfers associated with RWs. Figure 9c, d exhibits the weak signals of eastward transfers in the eastern domain associated with equatorial KWs. There is almost no difference between the zonal flux of wave energy (color shading in Fig. 9) in the two schemes. This confirms that the new expression in the present study continues the superiority of the AGC17 scheme in the Pacific Ocean. At both 8° N and the equator, the wind input rate of the classical expression (contours in Fig. 9a, c) shows a much stronger value than the modified wind input rate (contours in Fig. 9b, d) generally. At 8° N, significant values in the modified wind input are distributed in the central and eastern basins. At the equator, the modified wind input exhibits significant negative values in the western boundary regions, indicating that layer thickness anomalies are dumped by the horizontal divergence of Ekman velocity.
4 Conclusions
The present study has focused on the annual mean energy budget in the upper layer as a basis for understanding the dynamics of largescale waves. In the classical expression of the energy budget, wind input is estimated as the inner product of horizontal velocity vector and wind stress vector. Application of this expression to the energy diagnosis of annual waves yields ZASs in the annual mean wind input as shown by Li and Aiki (2020) for the Indian Ocean and the present study for the Pacific Ocean. We have investigated the reason why ZASs appear in the classical expression. Velocity associated with oceanic annual RWs contains both temporal and zonal variations while wind stress varies mainly in time to represent a seasonal cycle. The phase relationship between velocity and wind stress causes ZASs which disturb the understanding of the annual mean wind input.
We note that, in quasigeostrophic dynamics, wind input is determined as the negative product of streamfunction and wind stress curl. Geostrophic streamfunction is associated with (the vertical integral of) temperature anomaly (through the hydrostatic balance), while wind stress curl is associated with the horizontal divergence of windinduced Ekman velocity (which may be rewritten as the vertical convergence of velocity using the incompressibility condition). The product of streamfunction and wind stress curl is closely linked to the budget of gravitational potential energy (after vertical integration by parts). This quasigeostrophic expression of the wind input is singular at the equator, so the present study has modified it using pseudostreamfunction which is obtained from inverting the anomaly of Ertel’s potential vorticity. The modified expression of wind input takes the form of representing the input of gravitational potential energy with canceling out the input of kinetic energy. Because it uses pseudostreamfunction, the modified (phaseresistant) expression of the wind input is applicable to waves at all latitudes.
To maintain the mathematical consistency between equations for the budget of wave energy, modification of the wind input affects the expression of the wave energy flux. The AGC17 energy flux is compatible with the classical wind input in the wave energy budget Eq. (3a), while the AGC17M energy flux is compatible with the modified energy input in the wave energy budget Eq. (3b). The striking resemblance between Fig. 8a, b implies that modifying both the wind input and the energy flux vectors has minimum impact on the qualitative superiority of the AGC17 scheme in representing the group velocity of waves at all latitudes.
The result of the present study may be applied to the investigation of interannual variations in such a way as to reduce the complexity associated with a seasonal cycle in oceanic flows. We plan to utilize a model with more realistic setup and higher vertical resolution in the future. The realistic model would enable us to separate geostrophic flow and Ekman flow. Whether it is still effective to reduce ZASs will be checked.
Goddard and Philander (2000) pointed out that sea surface temperature (SST) and available potential energy are closely correlated, implying a connection between the wind input of wave energy and El Niño events. Toyoda et al. (2021, 2023) diagnosed an ocean reanalysis dataset in terms of the temporal evolution of wave energy associated with ENSO. They provided an energetic view for investigating ENSO which had often been understood in terms of raw quantities including SST, SSH, and thermocline depth in previous studies. In future studies, the energy perspective is expected to be used more extensively to enhance our understanding of the role of waves in tropical climate variations.
Availability of data and materials
Readers can download the European Centre for MediumRange Weather Forecasts Ocean Reanalysis/analysis System (ECMWF ORAS3) dataset from http://apdrc.soest.hawaii.edu:80/dods/public_data/Reanalysis_Data/ORAS3/1x1_grid. The temperature and salinity data which are used to calculate gravity wave speed are from http://apdrc.soest.hawaii.edu/dods/public_data/WOA/WOA13/1_deg/annual. The numerical simulation code and output data used or analyzed during the current study are available from the corresponding author on reasonable request.
Change history
01 August 2024
A Correction to this paper has been published: https://doi.org/10.1186/s4064502400640z
Abbreviations
 ZAS:

Zonally alternating signal
 RW:

Rossby wave
 SSH:

Sea surface height
 STCC:

Subtropical countercurrent
 WOA:

World Ocean Atlas
 KW:

Kelvin wave
 ECMWF ORAS3:

The European Centre for MediumRange Weather Forecasts Ocean Reanalysis System 3
 SOR:

Successive overrelaxation
 OS93 scheme:

Orlanski–Sheldon scheme
 AGC17 scheme:

Aiki–Greatbatch–Claus scheme
 AGC17M scheme:

The modified Aiki–Greatbatch–Claus scheme
 SST:

Sea surface temperature
 ENSO:

El Niño Southern Oscillation
References
Aiki H, Zhai X, Greatbatch RJ (2016) Energetics of the global ocean: the role of mesoscale eddies. In: Behara SK, Yamagata T (eds) IndoPacific climate variability and predictability. World scientific series on AsiaPacific weather and climate, vol 7. World Scientific, Singapore, pp 109–134. https://doi.org/10.1142/9789814696623_0004
Aiki H, Greatbatch RJ, Claus M (2017) Towards a seamlessly diagnosable expression for the energy flux associated with both equatorial and midlatitude waves. Prog Earth Planet Sci 4(1):1–18. https://doi.org/10.1186/s4064501701211
Aiki H, Fukutomi Y, Kanno Y, Ogata T, Toyoda T, Nakano H (2021) The energy flux of threedimensional waves in the atmosphere: exact expression for a basic model diagnosis with no equatorial gap. J Atmos Sci 78(11):3745–3758. https://doi.org/10.1175/JASD200177.1
Alford MH (2001) Internal swell generation: the spatial distribution of energy flux from the wind to mixed layer nearinertial motions. J Phys Oceanogr 31(8):2359–2368. https://doi.org/10.1175/15200485(2001)031%3c2359:ISGTSD%3e2.0.CO;2
Berloff PS, McWilliams JC (1999a) Largescale, lowfrequency variability in winddriven ocean gyres. J Phys Oceanogr 29(8):1925–1949. https://doi.org/10.1175/15200485(1999)029%3c1925:LSLFVI%3e2.0.CO;2
Berloff PS, McWilliams JC (1999b) Quasigeostrophic dynamics of the western boundary current. J Phys Oceanogr 29(10):2607–2634. https://doi.org/10.1175/15200485(1999)029%3c2607:QDOTWB%3e2.0.CO;2
Chelton DB, Schlax MG (1996) Global observations of oceanic Rossby waves. Sci 272(5259):234–238. https://doi.org/10.1126/science.272.5259.234
Cummins PF, Oey LY (1997) Simulation of barotropic and baroclinic tides off northern British Columbia. J Phys Oceanogr 27(5):762–781. https://doi.org/10.1175/15200485(1997)027%3c0762:SOBABT%3e2.0.CO;2
Cummins PF, Mysak LA, Hamilton K (1986) Generation of annual Rossby waves in the North Pacific by the wind stress curl. J Phys Oceanogr 16(7):1179–1189. https://doi.org/10.1175/15200485(1986)016%3c1179:GOARWI%3e2.0.CO;2
Ferrari R, Wunsch C (2009) Ocean circulation kinetic energy: reservoirs, sources, and sinks. Annu Rev Fluid Mech 41:253–282. https://doi.org/10.1146/annurev.fluid.40.111406.102139
Goddard L, Philander SG (2000) The energetics of El Niño and La Niña. J Clim 13(9):1496–1516. https://doi.org/10.1175/15200442(2000)013%3c1496:TEOENO%3e2.0.CO;2
Krauss W, Wuebber C (1982) Response of the North Atlantic to annual wind variations along the eastern coast. Deep Sea Res Part I Oceanogr Res Pap 29(7):851–868. https://doi.org/10.1016/01980149(82)900504
Li Z, Aiki H (2020) The life cycle of annual waves in the Indian Ocean as identified by seamless diagnosis of the energy flux. Geophys Res Lett 47(2):e2019GL085670. https://doi.org/10.1029/2019GL085670
Li Z, Aiki H (2022) The 1994 positive Indian Ocean Dipole event as investigated by the transfer routes of oceanic wave energy. J Phys Oceanogr 52(3):459–473. https://doi.org/10.1175/JPOD210189.1
Li Z, Aiki H (2024) Interpreting negative IOD events based on the transfer routes of wave energy in the upper ocean. J Phys Oceanogr 54(1):95–113. https://doi.org/10.1175/JPOD220267.1
Li Z, Aiki H, Nagura M, Ogata T (2021) The vertical structure of annual wave energy flux in the tropical Indian Ocean. Prog Earth Planet Sci 8(1):1–19. https://doi.org/10.1186/s40645021004329
Lueck R, Reid R (1984) On the production and dissipation of mechanical energy in the ocean. J Geophys Res Oceans 89(C3):3439–3445. https://doi.org/10.1029/JC089iC03p03439
Matsuno T (1966) Quasigeostrophic motions in the equatorial area. J Meteorol Soc Jpn Ser II 44(1):25–43. https://doi.org/10.2151/jmsj1965.44.1_25
McCreary JP, Kundu PK, Chao SY (1987) On the dynamics of the California Current System. J Mar Res 45(1):1–32. https://doi.org/10.1357/002224087788400945
Ogata T, Aiki H (2019) The pathway of intraseasonal wave energy in the tropical Indian Ocean as identified by a seamless diagnostic scheme. Sci Online Lett Atmosp 15:262–267. https://doi.org/10.2151/sola.2019047
Orlanski I, Sheldon J (1993) A case of downstream baroclinic development over western North America. Mon Weather Rev 121(11):2929–2950. https://doi.org/10.1175/15200493(1993)121%3c2929:ACODBD%3e2.0.CO;2
Perrin CL (1997) Numerical recipes in Fortran 90: the art of scientific computing, Volume 2. J Am Chem Soc 119(37):8748–8748. https://doi.org/10.1021/ja965936f
Qiu B (2003) Kuroshio Extension variability and forcing of the Pacific decadal oscillations: responses and potential feedback. J Phys Oceanogr 33(12):2465–2482. https://doi.org/10.1175/15200485(2003)033%3c2465:KEVAFO%3e2.0.CO;2
Qiu B, Miao W, Müller P (1997) Propagation and decay of forced and free baroclinic Rossby waves in offequatorial oceans. J Phys Oceanogr 27(11):2405–2417. https://doi.org/10.1175/15200485(1997)027%3c2405:PADOFA%3e2.0.CO;2
Reason CJC, Mysak LA, Cummins LA (1987) Generation of annualperiod Rossby Waves in the South Atlantic Ocean by the wind stress curl. J Phys Oceanogr 17:2030–2042. https://doi.org/10.1175/15200485(1987)017%3c2030:GOAPRW%3e2.0.CO;2
Renault L, Molemaker MJ, Gula J, Masson S, McWilliams JC (2016) Control and stabilization of the Gulf Stream by oceanic current interaction with the atmosphere. J Phys Oceanogr 46(11):3439–3453. https://doi.org/10.1175/JPOD160115.1
Schopf PS, Anderson DL, Smith R (1981) Betadispersion of lowfrequency Rossby waves. Dyn Atmos Oceans 5(3):187–214. https://doi.org/10.1016/03770265(81)900117
Scott RB, Straub DN (1998) Small viscosity behavior of a homogeneous, quasigeostrophic, ocean circulation model. J Mar Res 56(6):1225–1258. https://doi.org/10.1357/002224098765093625
Scott RB, Xu Y (2009) An update on the wind power input to the surface geostrophic flow of the World Ocean. Deep Sea Res Part I Oceanogr Res Pap 56(3):295–304. https://doi.org/10.1016/j.dsr.2008.09.010
Smagorinsky J, Manabe S, Holloway JL (1965) Numerical results from a ninelevel general circulation model of the atmosphere. Mon Weather Rev 93(12):727–768. https://doi.org/10.1175/15200493(1965)093%3c0727:NRFANL%3e2.3.CO;2
Song Q, Aiki H (2020) The climatological horizontal pattern of energy flux in the tropical Atlantic as identified by a unified diagnosis for Rossby and Kelvin waves. J Geophys Res Oceans 125(2):e2019JC015407. https://doi.org/10.1029/2019JC015407
Song Q, Aiki H (2021) Horizontal energy flux of winddriven intraseasonal waves in the tropical Atlantic by a unified diagnosis. J Phys Oceanogr 51(9):3037–3050. https://doi.org/10.1175/JPOD200262.1
Song Q, Aiki H (2023) Equatorial wave diagnosis for the Atlantic Niño in 2019 with an ocean reanalysis. Ocean Sci 19(6):1705–1717. https://doi.org/10.5194/os1917052023
Song Q, Aiki H, Tang Y (2023a) The role of equatorially forced waves in triggering Benguela Niño/Niña as investigated by an energy flux diagnosis. J Geophys Res Oceans 128(4):e2022JC019272. https://doi.org/10.1029/2022JC019272
Song Q, Tang Y, Aiki H (2023b) Dual wave energy sources for the Atlantic Niño events identified by wave energy flux in case studies. J Geophys Res Oceans 128(7):e2023JC019972. https://doi.org/10.1029/2023JC019972
Toyoda T, Nakano H, Aiki H, Ogata T, Fukutomi Y, Kanno Y, Urakawa LS, Sakamoto K, Yamanaka G, Nagura M (2021) Energy flow diagnosis of ENSO from an ocean reanalysis. J Clim 34(10):4023–4042. https://doi.org/10.1175/JCLID200704.1
Toyoda T, Urakawa LS, Aiki H, Nakano H, Shindo E, Yoshimura H, Kawakami Y, Sakamoto K, Yamagami A, Ushijima Y, Harada Y, Kobayashi C, Tomita H, Tozuka T, Yamanaka G (2023) Effective generation mechanisms of tropical instability waves as represented by highresolution coupled atmosphere–ocean prediction experiments. Sci Rep 13:14742. https://doi.org/10.1038/s41598023411595
Wang W, Huang RX (2004) Wind energy input to the Ekman layer. J Phys Oceanogr 34(5):1267–1275. https://doi.org/10.1175/15200485(2004)034%3C1267:WEITTE%3E2.0.CO;2
Wunsch C (1998) The work done by the wind on the oceanic general circulation. J Phys Oceanogr 28(11):2332–2340. https://doi.org/10.1175/15200485(1998)028%3c2332:TWDBTW%3e2.0.CO;2
Xu C, Zhai X, Shang XD (2016) Work done by atmospheric winds on mesoscale ocean eddies. Geophys Res Lett 43(23):12174–12180. https://doi.org/10.1002/2016GL071275
Zhai X, Johnson HL, Marshall DP, Wunsch C (2012) On the wind power input to the ocean general circulation. J Phys Oceanogr 42(8):1357–1365. https://doi.org/10.1175/JPOD1209.1
Acknowledgements
We thank Prof. Joji Ishizaka and Dr. Yoshihisa Mino for their advices in our study. KY also thanks Dr. Zimeng Li for the kind suggestions in the experiment.
Funding
This study is funded by Japan Society for the Promotion of Science KAKENHI Grant 22H00176.
Author information
Authors and Affiliations
Contributions
KY analyzed the data, carried out the experimental study, and constructed the manuscript. HA proposed the topic, gave constructions of the study, and modified the manuscript. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The original online version of this article was revised: Some equations, reference and minor grammer related errors have been corrected.
Appendices
Appendix A: Equations in spherical coordinates
All equations in the main body of the present manuscript are written in a Cartesian coordinate system for simplicity. Both the model code of shallow water equations and the diagnosis code of energy quantities have been written in a spherical coordinate system, which is explained below using \(\lambda\) and \(\theta\) as longitude and latitude coordinates, respectively.
The shallowwater Eqs. (1a)–(1c) in Sect. 2.1 may be expressed in spherical coordinates to read,
where \(R\) is the radius of the Earth. Manipulation of Eqs. (10a)–(10c) yields an equation for the budget of wave energy to read,
which corresponds to Eq. (3a). The pressureflux divergence and windforcing terms in Eq. (11a) may be rewritten in such a way as to yield a term associated with wind stress curl to read,
which corresponds to Eq. (3b) and \({\text{curl}}_{{{\text{tau}}}} = \frac{1}{R\cos \theta }\frac{\partial }{\partial \lambda }\left( {\tau_{{{\text{wind}}}}^{y} } \right)  \frac{1}{R\cos \theta }\frac{\partial }{\partial \theta }\left( {\cos \theta \tau_{{{\text{wind}}}}^{x} } \right)\) is understood. The inversion equation of pseudostreamfunction may be written in spherical coordinates to read,
which corresponds to Eqs. (2a) and (2b) in Sect. 2.2.
Appendix B: Case of Ekman spiral velocity
Equation (5b) shows that there is no wind input to Ekman flows in a singlelayer framework. This is revisited in this section using a depthdependent framework that can represent the Ekman spiral velocity. Let the symbol \(z\) represent the vertical coordinate and increase upward. Let the symbols \(U^{{{\text{ekm}}}}\) and \(V^{{{\text{ekm}}}}\) be the zonal and meridional components of Ekman spiral velocity in deep water, respectively, that are defined by the balance between the Coriolis term and the vertical viscosity term in momentum equations to read,
where \(A\) denotes to the coefficient of turbulent viscosity and subscript \(z\) denotes partial differentiation in the vertical direction. At the sea surface (\(z = 0\)), the boundary condition is \(\rho_{0} AU_{z}^{{{\text{ekm}}}} = \tau_{{{\text{wind}}}}^{x}\) and \(\rho_{0} AV_{z}^{{{\text{ekm}}}} = \tau_{{{\text{wind}}}}^{y}\). At deeper depths (\(z \to  \infty\)), the Ekman spiral velocity decays in amplitude to zero (\(U^{{{\text{ekm}}}} = 0,V^{{{\text{ekm}}}} = 0\)).
Multiplying Eqs. (13a) and (13b) by \(U^{{{\text{ekm}}}}\) and \(V^{{{\text{ekm}}}}\), respectively, and summing them yields
where the lefthand side is clearly zero. Equation (14) may be rewritten using vertical integration by parts to read
which indicates that wind input to the Ekman spiral velocity at the sea surface is canceled in sharp by the vertical integral of dissipation associated with turbulent viscosity.
Appendix C: Diagnostic schemes for energy fluxes
For diagnosing the horizontal flux of wave energy, based on the direction of group velocity in linear wave theory, three schemes were proposed in previous studies. The pressureflux scheme may be used in the studies of inertiagravity waves, that is written as,
where the edged brackets indicate a vector in the horizontal direction. Application of this scheme to simulated velocity and simulated geopotential in the model result of the present study yields Fig. 7a. This scheme cannot indicate the group velocity of midlatitude RWs, to be consistent with the results of Li and Aiki (2020) for the Indian Ocean and Song and Aiki (2020) for the Atlantic Ocean.
The OS93 scheme may be used to investigate the energy flux of midlatitude RWs, that is written as,
where subscripts \(x\) and \(y\) indicate partial differentiation in the zonal and meridional directions, respectively. Application of this scheme to the result of the model experiment yields Fig. 7b. This scheme is singular at the equator, and can indicate neither the energy flux of equatorial waves nor that of coastal KWs. These problems were resolved in the AGC17 scheme which is seamlessly applicable to waves at all latitudes with the following expression,
where \(\varphi\) is defined by Eq. (2b) in the present manuscript. Application of this scheme to the result of the model experiment yields Fig. 8a. The AGC17 scheme was used for investigating waves in the Indian Ocean (Ogata and Aiki 2019; Li and Aiki 2020, 2022, 2024; Li et al. 2021), the Atlantic Ocean (Song and Aiki 2020, 2021, 2023; Song et al. 2023a, b), the Pacific Ocean (Toyoda et al. 2021, 2023), and the atmosphere (Aiki et al. 2021). In Toyoda et al. (2021, 2023) and Song and Aiki (2023), the horizontal variation of gravity wave speed \(c\) was considered in their calculation of the pseudostreamfunction.
To be consistent with the modified (phaseresistant) expression of the wind input in the present study, we propose to modify the expression of the AGC17 energy flux as follows,
which is referred to as the AGC17M scheme. The zonal component of the AGC17M energy flux is shown in Fig. 8b. Comparing Figs. 8a, b and 9, we confirm that the modification of the energy flux maintains the qualitative superiority of the AGC17 scheme both in the horizontal distribution and time evolution through the year.
Apart from the four schemes mentioned above, Fourier analysis and ray theory were used in previous studies for tracing the life cycle of waves. These classical approaches do not provide the horizontal distribution of energy fluxes in the presence of different types of waves.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ye, K., Aiki, H. Wind stress curl as a driving force of annual waves in the upper ocean for interpreting energetics at all latitudes. Prog Earth Planet Sci 11, 29 (2024). https://doi.org/10.1186/s40645024006301
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s40645024006301