Near 5-Day Nonisostatic Response to Atmospheric Surface Pressure and Coastal-Trapped Waves Observed in the Northern South China Sea

Ruixiang Zhao State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou, China

Search for other papers by Ruixiang Zhao in
Current site
Google Scholar
PubMed
Close
,
Xiao-Hua Zhu State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, and Ocean College, Zhejiang University, Hangzhou, China

Search for other papers by Xiao-Hua Zhu in
Current site
Google Scholar
PubMed
Close
, and
Jae-Hun Park Department of Ocean Sciences, Inha University, Incheon, South Korea

Search for other papers by Jae-Hun Park in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Oceanic, nonisostatic responses to near 5-day Rossby–Haurwitz atmospheric pressure waves have been observed in open oceans; however, such responses based on observations in marginal seas such as the South China Sea have not been reported, owing to the limited ocean bottom pressure P bot records. The P bot measurements from pressure recording inverted echo sounders (PIESs) at sites in the northern South China Sea revealed a nonisostatic-like response near 5 days, although the coastal-trapped waves (CTWs) appeared to obscure it because their broadband periods include the near 5-day band. Cross-spectral analysis revealed that the PIES P bot records and the sea level (SL) records of Hong Kong all correlate strongly with the atmospheric pressure and winds over the East China Sea. This is indicative of remotely forced CTWs. The PIES P bot records showed higher coherence near 5 days with the zonal low-pass wavelength filters applied to the atmospheric pressure, and the phase analysis results strongly suggest nonisostatic oceanic responses to the westward-propagating Rossby–Haurwitz waves. Effective separation of CTWs and the nonisostatic responses from the P bot records at the near 5-day period was achieved. The oceanic responses to the Rossby–Haurwitz waves in the northern South China Sea were nonisostatic; a 1-mbar change in air pressure resulted in a 1.58-mbar change in P bot with a phase lag of 14.8°. The mean phase speed of CTWs from Hong Kong to station P3 was estimated to be 9.9 m s−1.

Denotes content that is immediately available upon publication as open access.

© 2017 American Meteorological Society.

Corresponding author: Xiao-Hua Zhu, xhzhu@sio.org.cn

Abstract

Oceanic, nonisostatic responses to near 5-day Rossby–Haurwitz atmospheric pressure waves have been observed in open oceans; however, such responses based on observations in marginal seas such as the South China Sea have not been reported, owing to the limited ocean bottom pressure P bot records. The P bot measurements from pressure recording inverted echo sounders (PIESs) at sites in the northern South China Sea revealed a nonisostatic-like response near 5 days, although the coastal-trapped waves (CTWs) appeared to obscure it because their broadband periods include the near 5-day band. Cross-spectral analysis revealed that the PIES P bot records and the sea level (SL) records of Hong Kong all correlate strongly with the atmospheric pressure and winds over the East China Sea. This is indicative of remotely forced CTWs. The PIES P bot records showed higher coherence near 5 days with the zonal low-pass wavelength filters applied to the atmospheric pressure, and the phase analysis results strongly suggest nonisostatic oceanic responses to the westward-propagating Rossby–Haurwitz waves. Effective separation of CTWs and the nonisostatic responses from the P bot records at the near 5-day period was achieved. The oceanic responses to the Rossby–Haurwitz waves in the northern South China Sea were nonisostatic; a 1-mbar change in air pressure resulted in a 1.58-mbar change in P bot with a phase lag of 14.8°. The mean phase speed of CTWs from Hong Kong to station P3 was estimated to be 9.9 m s−1.

Denotes content that is immediately available upon publication as open access.

© 2017 American Meteorological Society.

Corresponding author: Xiao-Hua Zhu, xhzhu@sio.org.cn

1. Introduction

The inverse barometer effect is an isostatic oceanic response to the local atmospheric surface pressure Patm, that is, an increase (decrease) in Patm leads to a decrease (increase) in sea level (SL), which gives rise to an unchanging ocean bottom pressure Pbot (Wunsch and Stammer 1997; Mathers and Woodworth 2001). The inverse barometer effect is expected to hold in the open ocean for periods of 2 days to 1 month (Brown et al. 1975; Ponte 1993). However, observations (e.g., Luther 1982; Woodworth et al. 1995; Hirose et al. 2001; Park and Watts 2006) show an exception for the near 5-day period band. Observational and numerical studies (e.g., Ponte 1997; Hirose et al. 2001; Park and Watts 2006) suggest that the oceanic nonisostatic response of this period was driven globally by the Rossby–Haurwitz wave (Madden and Julian 1972). In the tropical open ocean, scattered tide gauge data showed that the oceanic response to Patm at the near 5-day band does not comply with the inverse barometer effect or is nonisostatic because a 1-mbar increment (decrement) in Patm does not lead to a 1-cm decrement (increment) in SL (Luther 1982). Thus, Pbot is expected to display a significant near 5-day signal. Scarce Pbot observations from the tropical Atlantic (Woodworth et al. 1995) have confirmed that Pbot is phase locked to the local Patm at this period.

The behaviors of nonisostatic responses at the near 5-day period vary among locations because they are controlled by the shape and depth of the ocean basin; thus, more complex behaviors are expected in the Pacific and in the marginal seas (Ponte 1993, 1997). Thus far, Pbot observations in the marginal seas are quite limited. The few studies conducted on the Japan/East Sea (e.g., Park and Watts 2005, 2006; Xu et al. 2007) do not focus on the near 5-day oscillation associated with the Rossby–Haurwitz wave. For other important marginal seas, such as the South China Sea, Pbot data are rarely accessible. Thus, the complex behaviors of the nonisostatic oceanic response to the Patm in the South China Sea have never been verified and studied from observation. In this paper, we will demonstrate and study this near 5-day nonisostatic oceanic response in the northern South China Sea by using Pbot measurements from five pressure recording inverted echo sounders (PIESs) deployed from October 2012 to July 2014 (Zhu et al. 2015; Zhao et al. 2016; Zhao and Zhu 2016).

The South China Sea is the world’s largest marginal sea in the tropical region. Located between the Pacific and the Indian Oceans (Fig. 1a), it is surrounded by the Chinese mainland to the north, the Luzon Islands to the east, the Indonesian Islands to the south, and the Indo-China Peninsula to the west. It is connected to the Pacific Ocean mainly through the Luzon Strait with a maximum sill depth of about 2400 m, and it contains a broad continental shelf shallower than 100 m situated in the northern South China Sea.

Fig. 1.
Fig. 1.

(a) Map of the South China Sea. Black triangles indicate the PIES stations. The black box shows the region of (b). (b) Map of the northern South China Sea and the southern East China Sea. The five black triangles indicate PIES stations P1, P2, P3, P4, and P5 from north to south, respectively. The red dots represent the tide gauge stations used in this study.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

The topographic features of the northern South China Sea allow the propagation of continental shelf waves. Over the broad continental shelf, significant coastal-trapped waves (CTWs) were reported in previous research with periods of 1.6 to 32 days and phase speeds of 4.7 to 22.1 m s−1 (Chen and Su 1987; Li 1993; Ding et al. 2012; Li et al. 2015). It has been suggested that synoptic weather systems such as winter storm bursts in the Taiwan Strait and the East China Sea generate the CTWs that propagate to the northern South China Sea along the China mainland coast (Chen and Su 1987; Li et al. 2015). The frequency of the nonisostatic oceanic response examined in the present study falls into the frequency range of CTWs in the northern South China Sea. Previous studies from Chen and Su (1987) even suggested the existence of CTWs with a near 5-day period of 4.7 and 5 days. Thus, it is appropriate to propose that the Pbot data from PIESs deployed in this study at the near 5-day period likely contain both signals from the nonisostatic response forced by the global-scale, atmospheric, Rossby–Haurwitz wave and the signals from the propagated CTWs forced by remote synoptic weather systems. In this study, we separate the two processes from PIESs Pbot data by using data of several tide gauges, and we give some quantified results.

The rest of the paper is organized as follows: Section 2 describes the data used for this study and processing methods. Section 3 demonstrates cross-spectral analysis results providing the evidence for Rossby–Haurwitz waves and CTWs. Section 4 displays the processes for the separation of nonisostatic responses and CTWs from the Pbot data and gives quantitative results of the two processes.

2. Data and methods

During October 2012 to July 2014, five PIESs were deployed in the northern South China Sea (Fig. 1b). The PIESs were located in the gap between Hainan Island and the Xisha (Paracel) Islands, roughly perpendicular to the local bathymetry contours and the continental shelf break. The PIESs were bottom mounted at depths of 642, 1842, 2091, 1348, and 1030 m for stations P1, P2, P3, P4, and P5 from north to south, respectively. Apart from their basic function to measure the round-trip travel times of sound pulses from the seafloor to the sea surface, the PIESs were equipped with high-resolution pressure gauges (Paroscientific Inc., Redmond, Washington) to measure the Pbot. The interval of Pbot measurement was 1 h. The recorded time of the Pbot measurements was converted to UTC time to facilitate the combined analysis with the tide gauge SL data and the reanalysis Patm data. The Pbot data were windowed, detided, despiked, and dedrifted following the procedures by Kennelly et al. (2007).

The sea surface Patm data and the 10-m wind data used in this study, obtained from European Center for Medium-Range Weather Forecasts (ECMWF), had a spatial resolution of 0.75° and a time interval of 6 h.

We downloaded the SL data from several tide gauges (Fig. 1b) in the South China Sea in Hong Kong and in Zhapo, in the Taiwan Strait in Xiamen, and in the East China Sea in Kanmen. These stations are located northeast of PIES stations, where the upper regions were selected for CTW propagation to facilitate investigation of the possible origin of the CTWs observed by the PIES. The data were documented and processed by the Joint Archive for Sea Level (JASL). It should be noted that the SL data of only Hong Kong were available during the PIES observational period. These data were segmented during the PIES observational period for joint analysis with PIES data; the SL time series of all tide gauges, including those in Zhapo, Hong Kong, Xiamen, and Kanmen, were segmented from 1988 to 1989 to validate the CTW propagation at the near 5-day period. The hourly SL data used for this study were first detided by harmonic analysis (Pawlowicz et al. 2002), and a third-order Butterworth bandpass filter was then applied forward and backward to avoid phase shift. The bandpass filter had cutoff periods of 4 and 6 days.

Here, we applied the cross-spectral analysis method in which all time series were detrended and segmented with 50% overlap points, each containing 512 data points for 6-hourly time series. The hourly time series were subsampled to match the sampling interval. In addition, we applied a Hamming window to each segment. Following Emery and Thomson (2001) and Park and Watts (2005), we used block averaging in the frequency domain to smooth the spectra.

The complex empirical orthogonal function (CEOF) method was applied to extract the Rossby–Haurwitz waves from the Patm. The CEOF is a complex form of the well-known empirical orthogonal function (EOF); that is, its spatial pattern and temporal variations for each mode are in complex forms. The CEOF method first constructs a dataset U with its real part as a given dataset u and its imaginary part given as the Hilbert transform of u. A covariance matrix is calculated between spatial locations (i, j). Then, the spatial pattern Bn(x) (n is for each mode) is obtained as the complex eigenvectors of Cij, and the temporal variation An(t) is obtained as the associated principal components. The spatial pattern Bn(x) and the temporal variation An(t) include information of phase and amplitude, defined as
e1
e2
e3
and
e4
The asterisk superscript denotes the conjugate. More detailed information of the CEOF has been reported by Barnett (1984).

3. Evidence for nonisostatic responses and CTWs

The five Pbot time series exhibited quite similar patterns for the subtidal frequency band (Fig. 2a) and were almost identical for the near 5-day period band (Fig. 2b). The power spectral analysis of the hourly Pbot showed clear peaks at 4.3 days (Fig. 2c), which suggests that the signals of this frequency band are strong and significant. Because of the high consistency of Pbot, particularly in the frequency band of our focus, we used data from only station P3 for further analysis. The power spectrum of Hong Kong displayed a similar shape with those of PIES stations (Fig. 2c), especially in the near 5-day period band. This result suggests that oceanic oscillations of this frequency are prominent in the northern South China Sea. It should be noted that the oceanic oscillation was greatly attenuated offshore at P3 because the Hong Kong SL time series was multiplied by a factor of 0.2 for better joint illustration.

Fig. 2.
Fig. 2.

(a) Time series of the detided ocean bottom pressure Pbot for all PIES stations (gray lines except Pbot of station P3, which is shown by the black line). (b) Bandpass filtered Pbot for all PIES stations (red lines except Pbot of station P3, which is shown by the blue line). (c) Variance-preserving power spectrum density estimation of all PIES stations and Hong Kong. The gray line represents the 95% confidence level.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

Following Park and Watts (2006), we conducted cross-spectral analysis at the near 5-day period (4.6 days) for Pbot from station P3 with Patm and 10-m wind data of ECMWF (Figs. 3m–o). It should be noted that the Pbot data were not bandpass filtered and were subsampled at 6-h intervals to match the time intervals of ECMWF data. Park and Watts (2006) showed no significant coherence between Pbot and local Patm at midlatitudes. They showed an inverted barometer response, although high values of the coherences squared (hereafter referred to as coherences) between the midlatitude Pbot and remote Patm near the equator were noted. Our results differ significantly from those of theirs because the region of the maximum coherence was located in the East China Sea and its neighboring western Pacific Ocean at approximately 12°–32°N, 120°–140°E. The Pbot also showed high coherence with the 10-m wind data in the East China Sea. These results indicate that the atmospheric conditions, including both air pressure and wind, in the East China Sea were likely to drive the near 5-day oscillation remotely in the northern South China Sea. In addition, we found that the SL in Hong Kong did not show high coherence with its local Patm (Fig. 3j). Instead, the SL of Hong Kong showed high coherence with regions in the East China Sea for Patm and winds (Figs. 3j,l). This result can be attributed to the CTWs propagating from the East China Sea.

Fig. 3.
Fig. 3.

Cross-spectral analysis results at the near 5-day period band for the tide gauges at (a)–(c) Kanmen, (d)–(f), Xiamen, (g)–(i) Zhapo, and (j)–(l) Hong Kong, and (m)–(o) the PIES station P3. The first two columns represent coherence and phase including SL or ocean bottom pressure Pbot lags from Patm in degrees, where the contour line interval is 30°. The last column represents coherence for the eastward and northward wind components at 10 m; however, only the larger value is plotted. The red dot for each subfigure denotes a tide gauge station or PIES station. Coherence and phase were omitted in mapping if the coherence was lower than the 95% confidence level.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

To validate our hypothesis, we used the sea level data from the tide gauges from the East China Sea to the South China Sea along the China mainland coast during 1988 to 1989, which is consistent with the temporal span of PIES observation. Cross-spectral analysis was conducted between the SL of Hong Kong and SLs of other tide gauges. The SLs of Kanmen, Xiamen, and Zhapo all showed coherences above the 95% confidence level of 0.23 with the SL of Hong Kong at the near 5-day period (Figs. 4a,d–f). The phase distribution (Fig. 4b) shows that these coherent signals propagated from the East China Sea to the South China Sea, which is consistent with the CTW propagating direction. The mean phase speed was estimated to be roughly 7.9 m s−1, which is consistent with the CTW phase speed obtained from previous studies (Chen and Su 1987; Li 1993; Ding et al. 2012; Li et al. 2015), if the CTWs ideally propagated exactly along the lines connecting adjacent stations. The coherence function between Hong Kong and P3 (Fig. 4c), Zhapo (Fig. 4d), Xiamen (Fig. 4e), and Kanmen (Fig. 4f) showed significant coherence in a broad band of frequencies including the 4–6-day periods with peaks at 4.3 and 5.3 days. This result demonstrates the broad bandwidth nature of CTWs. Although the coherences between Hong Kong and other tide gauges decreased with distance because a myriad of processes can alter the CTW signals during their propagation, the results show that the SL of Hong Kong was affected significantly by the CTW originating from the East China Sea.

Fig. 4.
Fig. 4.

Cross-spectral analysis results at the near 5-day period band for the tide gauges in Hong Kong during 1988 to 1989. The black star indicates the Hong Kong tide gauge; the black triangle indicates the P3 station; and colored dots indicate the tide gauges for joint analysis with Hong Kong, with colors showing their (a) coherences or (b) phase lags. The lower panels show the coherence function between Hong Kong and (c) P3, (d) Zhapo, (e) Xiamen, and (f) Kanmen. The red lines show the 95% confidence level.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

The Rossby–Haurwitz waves are atmospheric waves of large scale that can be obscured by synoptic weather systems of relatively smaller scales in cross-spectral analysis. To check whether the Pbot in the South China Sea was affected by the atmospheric Rossby–Haurwitz waves characterized by very long zonal wavelength with a wavenumber of −1 (which means that the wavelength is one zonal circle for each latitude and the signal is westward propagating), we applied zonal low-pass filtering with a cutoff wavenumber of 300° on the ECMWF Patm and 10-m winds. The gains displayed in Fig. 5 are defined as Patm/Pbot or Patm/SL, or, more specifically, Gain(f) = G1(f)/G2(f), where f is frequency and G1(f) and G2(f) are one-sided autospectrum estimates for Patm and Pbot (or SL), respectively. The gains are defined as such to facilitate discussion of the structure of the Patm, which is assumed to be the driving agent. Interestingly, the coherence between Pbot at P3 with the local Patm increased significantly from 0.07 to 0.39 after the Patm was filtered (Fig. 5q). A high-coherence band appeared in the tropic region extending from about 30°N to 30°S, which is similar to the patterns reported by Park and Watts (2006). Coherence with the wind was reduced (Fig. 5t) compared with earlier results, which suggests that the wind effects were negligible at this spatial scale. Perhaps the most conspicuous results are that the phase distribution of Patm showed a consistent westward propagation at all latitudes. We identified a near 30° phase difference, as indicated in Fig. 5r by contour lines with intervals of 30°, for a near 30° longitudinal distance of the same latitude. This strongly indicates typical characteristics of the Rossby–Haurwitz wave. We also observed less gain in Patm near the equator but more poleward (Fig. 5s). This pattern is also consistent with the theoretical pattern of the Rossby–Haurwitz wave, which shows a meridionally symmetric structure with a smaller amplitude of 0.5–1.0 mbar in the tropics and a larger amplitude of 1–2 mbar in extratropical regions. One discrepancy from Park and Watts (2006) is that the phase of local Patm over P3 was not equal to that of Pbot; a difference of about 70° was noted. This phenomenon appears to be related to phase shifts caused by the CTW signals in Pbot that were not previously removed. The separation of CTWs and the nonisostatic oceanic responses will be discussed subsequently.

Fig. 5.
Fig. 5.

As in Fig. 3, but that a 300° zonal low-pass filter was applied to the atmospheric surface pressure Patm, and the third column inserted shows the gain, defined as Patm/SL or Patm/Pbot.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

The maximum coherence regions identified are east of the South China Sea and northeast of Australia, showing a nearly symmetric pattern around the equator; however, the local coherence was also high. This strange pattern might have occurred because the local CTWs were not removed from Pbot, and the filtered Patm might also contain noise other than the Rossby–Haurwitz waves. Otherwise, the Pbot would have constant coherence with the Patm in every location if Patm contained only the Rossby–Haurwitz wave. The pattern could also be explained by the narrowband stochastic nature of atmospheric forcing of oceanic Rossby waves, which can produce a nonlocal maximum of coherence (Brink 1989). Nevertheless, the increased high local coherence with Patm on the disadvantageous condition that the CTWs of Pbot were not removed, the meridionally symmetric pattern of the analyzed variables, and the westward phase propagation with a wavenumber of −1 all provide robust evidence for the existence of near 5-day nonisostatic oceanic responses to the atmospheric Rossby–Haurwitz wave in the northern South China Sea.

Figure 6 shows time series of SL data from Hong Kong (Fig. 6a), Pbot from station P3 (Fig. 6b), and Patm data from ECMWF interpolated to P3 (Fig. 6c). Figure 6c was plotted with data unfiltered in space, and Fig. 6d was plotted with data after the zonal 300° wavelength low-pass filtering. The SL of Hong Kong and the Pbot of P3 clearly showed high coherence, at r = 0.50. We determined that the Pbot of P3 had little relevance with its local Patm (Fig. 6c), at r = 0.07, but showed very high coherence with the Patm over the East China Sea (Fig. 6e), which peaked at r = 0.78. This indicates that the main signals of Pbot were likely driven remotely from the East China Sea. The Pbot showed higher coherence with the Patm, at r = 0.39, when a zonal 300° wavenumber filter was applied to the latter (Fig. 6d). We noticed that during January to June 2014, when the filtered Patm was weaker, the SL of Hong Kong and the Pbot of P3 showed remarkably better coherence, reaching 0.78; a Hamming window containing 128 data points was applied instead to maintain the confidence level. This close agreement may have occurred because the Rossby–Haurwitz wave-forced nonisostatic response acted as noise in our analysis of the CTWs. Therefore, when the nonisostatic response is reduced, we can expect better coherence between the SL of Hong Kong and the Pbot of P3. These results support the existence of a near 5-day nonisostatic response and CTWs in the northern South China Sea.

Fig. 6.
Fig. 6.

Bandpass filtered time series with cutoff periods of 4 and 6 days for the (a) SL of Hong Kong; (b) ocean bottom pressure Pbot of station P3; (c) atmospheric surface pressure Patm of P3; (d) Patm of P3, where the Patm was applied with a 300° zonal low-pass filter; and (e) Patm in the East China Sea, the coherence of which was the best with the Pbot of P3. All of the SL, Pbot, and Patm data are shown in blue, red, and black lines, respectively. The shaded areas represent 1 Jan to 15 Jun 2014, the period in which the Patm applied with a 300° zonal low-pass filter was weak and the coherence between the SL of Hong Kong and the Pbot of P3 was higher. It is noted that the ordinate ranges of these time series varied from ±1 to ±10.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

To verify the CTWs in P3, we calculated the phase lag from Hong Kong to P3, utilizing the cross-spectral analysis including hourly data of SL and Pbot. We found that the Pbot of P3 lagged the SL of Hong Kong about 30.8° near the 5-day period. Assuming the Pbot was dominated by the CTWs, which tended to propagate along the shelf, the phase speed of CTWs can be roughly estimated, assuming that the phase of SL in Hong Kong is equal to a location A in the outer shelf break (Fig. 7). The direction from A to Hong Kong was perpendicular to the orientation of the continental shelf, although the direction from A to P3 was almost parallel. The distance from A to P3 is about 462 km; therefore, the mean phase speed was about 12.5 m s−1, if we neglect the oceanic response to Rossby–Haurwitz waves in P3. The mean phase speed is in the CTW phase speed range of 4.7 to 22.1 m s−1, as reported in previous research (Chen and Su 1987; Li 1993; Ding et al. 2012; Li et al. 2015). This serves as strong evidence of the existence of energetic CTWs in P3.

Fig. 7.
Fig. 7.

Illustration of the calculation of the CTW phase speed between the Hong Kong station and station P3; these stations are shown in red triangles. The red dashed line is assumed to have the same phase as that of Hong Kong at the 5-day period. The red solid line is assumed to be the propagation path to P3. The red arrow indicates the propagation direction. The results observed when the nonisostatic response was removed (not removed) from the ocean bottom pressure Pbot of P3 are denoted in red (blue) color, respectively.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

4. Discussion

The basic aim of this study is to effectively separate the nonisostatic response signals and the CTW signals mixed in the Pbot time series at the near 5-day period band. To achieve this, several reasonable assumptions should be made. First, the Pbot contains signals forced by the Rossby–Haurwitz wave and those from the CTWs; thus,
e5
if other signals are negligible near 5-day periods. The SL of Hong Kong contains CTW signals and other signals, which have no relevance with the atmospheric Rossby–Haurwitz wave. Figure 5 shows that the local coherence with the zonal filtered Patm is quite low. Second, a simple linear relationship exists for a specific location; that is, there is a scaling constant a and a phase offset constant b for P3 so that
e6
where Patm_RHW(b) indicates the local atmospheric Rossby–Haurwitz wave signal with a phase offset of b at the 5-day period. The phase offset is introduced for the following reasons. First, the spatial scale of the South China Sea is small, whereas the forcing scale of the Rossby–Haurwitz wave is quite large; thus, no local oceanic responses are likely in the South China Sea. The oceanic responses propagate from the Pacific Ocean to the South China Sea, resulting in a mismatch of phases between the local oceanic response and the atmospheric Rossby–Haurwitz wave. Second, resonances of oceanic nonisostatic responses within the South China Sea Basin are possible, which can also lead to phase mismatches. Finally, we expect that when the Pbot_RHW is totally removed, the Pbot_CTW should have the best coherence with the SL of Hong Kong. This assumption is reasonable because the large-scale Rossby–Haurwitz wave signal is correlated with neither the CTW signal nor the other signals in the SL of Hong Kong, as indicated in Fig. 5m by r = 0.11. In addition, Hong Kong and P3 are in relatively close proximity, which appears to result in limited modulation of CTWs during their propagation. Therefore, proper parameters a and b must be found so that PbotaPatm_RHW(b) has the best coherence with the SL of Hong Kong.

To achieve the above requirement, extraction of the exact Rossby–Haurwitz wave in the atmosphere is key. Because the Rossby–Haurwitz wave is characterized by its long wavelength and its near 5-day period, we first applied a third-order Butterworth bandpass filter with cutoff periods of 4 and 6 days for each spatial grid of ECMWF Patm. Then, a zonal wavelength low-pass filter was applied with cutoff wavelengths of 300°. This cutoff wavelength was chosen because it showed good performance in noise reduction. The global Patm showed a more symmetric pattern around the equator that was closer to the theoretical Rossby–Haurwitz waves.

Although bandpass filters in both time and space were applied to the Patm to filter out signals other than the Rossby–Haurwitz waves, significant noise remained in the Patm. Here, cross-spectral analysis was applied between the global Patm and that over 0°, 180° (figure not shown). The energy poleward of 30°N and 30°S was significantly higher than expected because the theoretical structure of the Rossby–Haurwitz wave is such that the amplitude is about 1–2 mbar in extratropical regions and 0.5–1.0 mbar in the tropics (Madden and Julian 1973; Park and Watts 2006). This means that the global Patm gain, defined as the ratio of the global Patm and the Patm over 0°, 180° at 5-day period [Gain(f) = G1(f)/G2(f), where G1(f) and G2(f) are one-sided autospectrum estimates for Patm at each grid point and the Patm over 0°, 180°] should be no larger than 2/0.5 = 4. However, the maximum gain reached about 100 at the poles. Another phenomenon worth noting is that the coherences poleward of 30°N and 30°S were sharply decreased, indicating the existence of significant noise. For high latitudes, this large noise may be attributed to the energetic atmospheric systems. In addition, the zonal wavelength filter showed poor performance in filtering out synoptic weather systems near the poles because the zonal circles are shorter in those regions. For midlatitudes, noise can arise from the strong westerly bands in that region, which are closely related to atmospheric pressure; we identified much higher energy in the Southern Hemisphere. On the basis of these considerations, we selected Patm only between 30°N and 30°S for the following analysis. After implementation of the aforementioned processes, the Patm were expected to contain little noise for contaminating the Rossby–Haurwitz waves; their signals were not reduced by filtering.

The CEOF analysis method is an effective tool for identifying and extracting the Rossby–Haurwitz waves in the Patm, as demonstrated below. Figure 8 shows that the leading mode can explain up to 77.5% of the total variance, leaving the other modes negligible. The amplitude structure of the leading mode was nearly symmetric around the equator, and the amplitudes increased moderately poleward. The most prominent features were its westward propagation and a wavenumber of 1. A nearly 1° phase difference corresponded to a nearly 1° longitudinal distance of the same latitude, which resembled the spatial structure shown in Fig. 5r. These features all agree quite well with the characteristics of theoretical Rossby–Haurwitz waves (Madden and Julian 1972; Ponte 1997; Park and Watts 2006). Thus, we determined that the leading mode of Patm in fact demonstrated the signals of Rossby–Haurwitz waves. The time-varying Rossby–Haurwitz wave signals for each spatial grid of ECMWF were obtained as the real part of the CEOF leading mode, that is, X1(x, t) = Re[A1(t) × B1(x)], where B(x) is the spatial pattern, A(t) is the temporal variation, and the subscript 1 denotes the first mode of CEOF. The Patm of the Rossby–Haurwitz wave for each location was obtained by interpolation. For our case, Patm_RHW(b) was obtained by simple spatial interpolation along the latitude of P3. For example, Patm_RHW(−10°) indicates Patm_RHW with a phase lag of 10° to Patm_RHW of P3; thus, Patm_RHW(−10°) can be obtained by interpolation of the Rossby–Haurwitz wave signal at a location west of P3 by 10°/360° × (5 × 360°) = 50° in longitude. The above spatial interpolation method is quite advantageous because the ECMWF product we used has a spatial resolution of 0.75°; thus, it essentially provided a phase resolution of 3.75° for our sensitivity experiments. The phase resolution would be much coarser if we had obtained Patm_RHW(b) by temporal shift [e.g., Patm_RHW(−18°) is obtained as the local Patm_RHW but with a delay of 1/4 day]; a temporal resolution of one-fourth of a day resulted in a phase resolution of 360°/(4 × 5) = 18°. This process was not expected to introduce significant error because the Rossby–Haurwitz waves were of very large scale.

Fig. 8.
Fig. 8.

CEOF spatial pattern for the three leading modes. Distribution of (left) amplitude and (right) phase.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

To determine the proper coefficients a and b, a set of sensitivity experiments was conducted. We performed cross-spectral analysis between PbotaPatm_RHW(b) and the SL of Hong Kong, with the coefficient a increasing from 0 to 3 at intervals of 0.01 and Patm_RHW(b) obtained by interpolation to each grid point of ECMWF. The cross-spectral analysis results were averaged on the period band of 4 to 6 days because the period band of CTWs is broad and does not peak at 4.6 days. It was found that the coherence peaked at 0.59, when a reached 1.58 and b reached −14.8°. We have also tried the method of temporal shift to obtain Patm_RHW(b), with a similar result: a = 1.26 and b = −18° (figure not shown). Considering the estimation of a is quite sensitive to b while the phase resolution of this method is coarse, the estimated a by this method can be quite inaccurate. Thus, from the former estimation, the best coefficients a and b were determined to be 1.58° and −14.8°, respectively, and Pbot_RHW was found to be equal to 1.58 Patm_RHW(−14.8°) (Fig. 9e). Therefore, the behavior of oceanic response to the Rossby–Haurwitz wave at the near 5-day period in the northern South China Sea was nonisostatic. That is, when the atmospheric pressure caused by the Rossby–Haurwitz waves increased (decreased) by 1 mbar, the Pbot increased (decreased) by 1.58 mbar. Moreover, a phase lag of 14.8° was noted, which suggests that the oceanic response to the atmospheric Rossby–Haurwitz wave propagated from the Pacific Ocean to the South China Sea and that resonances of nonisostatic responses may exist within the South China Sea. We presume that this result is related to the depth and shape of the South China Sea Basin, which controls the behavior of the nonisostatic responses. (Ponte 1993, 1997).

Fig. 9.
Fig. 9.

Results of sensitivity and cross-spectral analysis. (a) The sensitivity of coefficients a and b to squared coherence. The black dot indicates coefficients a and b with the best coherence. Time series of (b) raw Pbot, (c) Rossby–Haurwitz waves, and (d) coastal-trapped waves in P3 at the near 5-day period band.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

Nonisostatic response can also change the cross-spectral analysis results between Pbot and the SL of Hong Kong. Specifically, the phase lag from Hong Kong increased from 30.8° to 38.7°, and the gain decreased from 3.39 to 3.98. The 95% confidence interval for phase lag estimation is ±1.87°. Therefore, ignoring the effects of Rossby–Haurwitz waves in Pbot has a nonnegligible influence on the phase speed calculation between P3 and Hong Kong (Fig. 7). The phase speed was calculated to be 9.9 m s−1 when the Rossby–Haurwitz wave-caused nonisostatic response was removed.

We also found that the improved coherences clearly increased between the Pbot and the unfiltered Patm (winds) over the China mainland, including the East China Sea, Taiwan Strait, and the northeastern South China Sea, when the nonisostatic responses forced by Rossby–Haurwitz waves were removed from Pbot and signals containing CTWs were left. This result was obvious for improved coherences with nonisostatic responses removed minus those with nonisostatic responses not removed (Fig. 10). For Patm, a region of higher coherence averaged on the period band of 4 to 6 days was identified over the China mainland around 28°N, 115°E. The best increased coherence, at 28.8°N, 115.5°E, reached 0.12. Thus, the Pbot was found to be coherent with the Patm over a larger region around the East China Sea, suggesting that the Patm-related synoptic weather systems over the China mainland are also important for CTW generation. For winds, the best improved coherence region was located on the continental shelf of the East China Sea, along the Taiwan Strait and in the northeastern South China Sea, reflecting a slight eastward shift compared with that of Patm. The best improved coherence, at 27.75°N, 123°E, reached 0.12. The regions showing the most improvement in coherence for winds were those in which CTWs were generated that eventually propagated to the PIES stations. This provides further evidence for our effective separation because CTWs in the northern South China Sea are remotely forced by synoptic weather systems (Chen and Su 1987; Li et al. 2015).

Fig. 10.
Fig. 10.

Amounts of change in the coherences for ocean bottom pressure Pbot and (a) atmospheric surface pressure Patm and (b) winds when the nonisostatic responses forced by Rossby–Haurwitz waves were removed from Pbot. Red dots indicate station P3. Gray lines indicate zero values.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0013.1

5. Conclusions

In this paper, we have for the first time observed nonisostatic responses to atmospheric pressure for the near 5-day period band in the South China Sea. The nonisostatic responses were driven globally by the atmospheric Rossby–Haurwitz wave. Specifically, it is suggested that the varied ocean bottom pressure observed propagated from the Pacific Ocean because the spatial scale of the South China Sea is quite small compared with the atmospheric forcing. It remains unclear whether there is a local response to the Rossby–Haurwitz wave in the South China Sea or whether resonances of these signals exist. The Pbot recorded by PIESs were all strongly affected by CTWs during the same period as those originating from the East China Sea. As inferred from previous studies (e.g., Chen and Su 1987; Li et al. 2015), the synoptic weather systems served as the major driving mechanism. We achieved effective separation of the CTWs and the nonisostatic responses from the Pbot records at the near 5-day period. We found that a 1-mbar increase (decrease) in atmospheric pressure caused by a Rossby–Haurwitz wave resulted in a 1.58-mbar increase (decrease) in Pbot, with a phase lag of 14.8°. On the basis of the SL of Hong Kong and the Pbot of P3, we estimated the mean CTW propagation phase speed to be 9.9 m s−1.

The most prominent highlight of this research is that we achieved separation of CTWs and the nonisostatic oceanic response to Rossby–Haurwitz waves. The methods we proposed have some applicability for the analysis of Pbot records, particularly in regions where strong CTWs are dominant and obscure the weak nonisostatic oceanic response. Successful separation of the two signals is essential in researching CTWs and the behavior of nonisostatic responses. Another important implication of this study is that the CTWs and nonisostatic response in the near 5-day period can introduce aliased peaks at longer periods for satellite altimetry observation because altimeter observation has a coarse temporal resolution (e.g., 9.9156 days for TOPEX/Poseidon, Jason-1, Jason-2, and Jason-3). Therefore, removing the aliasing effects on altimeter measurements is important for application of sea surface height products (Carrère and Lyard 2003; Park and Watts 2006; Stammer et al. 2000).

Acknowledgments

This study was supported by the Project of State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography (SOEDZZ1701), the Scientific Research Fund of SIO (JT1604 and JG1621), the National Key Research and Development Program of China (2016YFC1400104), the National Natural Science Foundation of China (41576001, 41621064, 41476020, and 41606113), Zhejiang Provincial Natural Science Foundation of China (LQ17D060003), and the National Program on Global Change and Air-Sea Interaction (GASI-IPOVAI-01-02). Jae-Hun Park was supported by “Development of satellite based ocean carbon flux model for seas around Korea” funded by the Ministry of Ocean and Fisheries, Republic of Korea.

REFERENCES

  • Barnett, T. P., 1984: Interaction of the monsoon and Pacific trade wind system at interannual time scales. II: The tropical band. Mon. Wea. Rev., 112, 23802387, doi:10.1175/1520-0493(1984)112<2380:IOTMAP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brink, K. H., 1989: Evidence for wind-driven current fluctuations in the western North Atlantic. J. Geophys. Res., 94, 20292044, doi:10.1029/JC094iC02p02029.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, W., W. Munk, F. Snodgrass, H. Mofjeld, and B. Zetler, 1975: Mode bottom experiment. J. Phys. Oceanogr., 5, 7585, doi:10.1175/1520-0485(1975)005<0075:MBE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carrère, L., and F. Lyard, 2003: Modeling the barotropic response of the global ocean to atmospheric wind and pressure forcing: Comparisons with observations. Geophys. Res. Lett., 30, 1275, doi:10.1029/2002GL016473.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, D., and J. Su, 1987: A preliminary study of the continental shelf wave along the coast of China (in Chinese). Acta Oceanol. Sin., 9, 115.

    • Search Google Scholar
    • Export Citation
  • Ding, Y., X. Bao, and M. Shi, 2012: Characteristics of coastal trapped waves along the northern coast of the South China Sea during year 1990. Ocean Dyn., 62, 12591285, doi:10.1007/s10236-012-0563-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Emery, W. J., and R. E. Thomson, 2001: Data Analysis Methods in Physical Oceanography. 2nd ed. Elsevier, 638 pp.

  • Hirose, N., I. Fukumori, and R. M. Ponte, 2001: A non-isostatic global sea level response to barometric pressure near 5 days. Geophys. Res. Lett., 28, 24412444, doi:10.1029/2001GL012907.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kennelly, M. A., K. L. Tracey, and D. R. Watts, 2007: Inverted echo sounder data processing manual. GSO Tech. Rep. 2007-02, University of Rhode Island, 87 pp.

  • Li, J., Q. Zheng, J. Hu, Z. Fan, J. Zhu, T. Chen, B. Zhu, and Y. Xu, 2015: Wavelet analysis of coastal-trapped waves along the China coast generated by winter storms in 2008. Acta Oceanol. Sin., 34, 2231, doi:10.1007/s13131-015-0701-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, L., 1993: A study of winter subtidal sea level fluctuation along the northern coast of the South China Sea (in Chinese). Trop. Oceanol., 12, 5260.

    • Search Google Scholar
    • Export Citation
  • Luther, D. S., 1982: Evidence of a 4–6 day barotropic, planetary oscillation of the Pacific Ocean. J. Phys. Oceanogr., 12, 644657, doi:10.1175/1520-0485(1982)012<0644:EOADBP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madden, R., and P. Julian, 1972: Further evidence of global-scale 5-day pressure waves. J. Atmos. Sci., 29, 14641469, doi:10.1175/1520-0469(1972)029<1464:FEOGSD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madden, R., and P. Julian, 1973: Reply. J. Atmos. Sci., 30, 935940, doi:10.1175/1520-0469(1973)030<0935:R>2.0.CO;2.

  • Mathers, E. L., and P. L. Woodworth, 2001: Departures from the local inverse barometer model observed in altimeter and tide gauge data and in a global barotropic numerical model. J. Geophys. Res., 106, 69576972, doi:10.1029/2000JC000241.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, J.-H., and D. R. Watts, 2005: Response of the southwestern Japan/ East Sea to atmospheric pressure. Deep-Sea Res. II, 52, 16711683, doi:10.1016/j.dsr2.2003.08.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, J.-H., and D. R. Watts, 2006: Near 5-day nonisostatic response of the Atlantic Ocean to atmospheric surface pressure deduced from sub-surface and bottom pressure measurements. Geophys. Res. Lett., 33, L12610, doi:10.1029/2006GL026304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pawlowicz, R., B. Beardsley, and S. Lentz, 2002: Classical tidal harmonic analysis including error estimates in MATLAB using T_TIDE. Comput. Geosci., 28, 929937, doi:10.1016/S0098-3004(02)00013-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ponte, R. M., 1993: Variability in a homogeneous global ocean forced by barometric pressure. Dyn. Atmos. Oceans, 18, 209234, doi:10.1016/0377-0265(93)90010-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ponte, R. M., 1997: Nonequilibrium response of the global ocean to the 5-day Rossby–Haurwitz wave in atmospheric surface pressure. J. Phys. Oceanogr., 27, 21582168, doi:10.1175/1520-0485(0)027<2158:NROTGO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stammer, D., C. Wunsch, and R. M. Ponte, 2000: De-aliasing of global high frequency barotropic motions in altimeter observations. Geophys. Res. Lett., 27, 11751178, doi:10.1029/1999GL011263.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woodworth, P. L., S. A. Windle, and J. M. Vassie, 1995: Departures from the local inverse barometer model at periods of 5 days in the central South Atlantic. J. Geophys. Res., 100, 18 28118 290, doi:10.1029/95JC01741.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wunsch, C., and D. Stammer, 1997: Atmospheric loading and the oceanic “inverted barometer” effect. Rev. Geophys., 35, 79107, doi:10.1029/96RG03037.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Xu, Y., D. R. Watts, M. Wimbush, and J.-H. Park, 2007: Fundamental-mode basin oscillations in the Japan/East Sea. Geophys. Res. Lett., 34, L04605, doi:10.1029/2006GL028755.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhao, R., and X.-H. Zhu, 2016: Weakest winter South China Sea western boundary current caused by the 2015–2016 El Niño event. J. Geophys. Res. Oceans, 121, 76737682, doi:10.1002/2016JC012252.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhao, R., X.-H. Zhu, and X. Guo, 2016: The impact of monsoon winds and mesoscale eddies on thermohaline structures and circulation patterns in the northern South China Sea. Cont. Shelf Res., 143, 240256, doi:10.1016/j.csr.2016.06.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, X.-H., R. Zhao, X. Guo, Y. Long, Y.-L. Ma, and X. Fan, 2015: A long-term volume transport time series estimated by combining in situ observation and satellite altimeter data in the northern South China Sea. J. Oceanogr., 71, 663673, doi:10.1007/s10872-015-0305-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save
  • Barnett, T. P., 1984: Interaction of the monsoon and Pacific trade wind system at interannual time scales. II: The tropical band. Mon. Wea. Rev., 112, 23802387, doi:10.1175/1520-0493(1984)112<2380:IOTMAP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brink, K. H., 1989: Evidence for wind-driven current fluctuations in the western North Atlantic. J. Geophys. Res., 94, 20292044, doi:10.1029/JC094iC02p02029.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, W., W. Munk, F. Snodgrass, H. Mofjeld, and B. Zetler, 1975: Mode bottom experiment. J. Phys. Oceanogr., 5, 7585, doi:10.1175/1520-0485(1975)005<0075:MBE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carrère, L., and F. Lyard, 2003: Modeling the barotropic response of the global ocean to atmospheric wind and pressure forcing: Comparisons with observations. Geophys. Res. Lett., 30, 1275, doi:10.1029/2002GL016473.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, D., and J. Su, 1987: A preliminary study of the continental shelf wave along the coast of China (in Chinese). Acta Oceanol. Sin., 9, 115.

    • Search Google Scholar
    • Export Citation
  • Ding, Y., X. Bao, and M. Shi, 2012: Characteristics of coastal trapped waves along the northern coast of the South China Sea during year 1990. Ocean Dyn., 62, 12591285, doi:10.1007/s10236-012-0563-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Emery, W. J., and R. E. Thomson, 2001: Data Analysis Methods in Physical Oceanography. 2nd ed. Elsevier, 638 pp.

  • Hirose, N., I. Fukumori, and R. M. Ponte, 2001: A non-isostatic global sea level response to barometric pressure near 5 days. Geophys. Res. Lett., 28, 24412444, doi:10.1029/2001GL012907.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kennelly, M. A., K. L. Tracey, and D. R. Watts, 2007: Inverted echo sounder data processing manual. GSO Tech. Rep. 2007-02, University of Rhode Island, 87 pp.

  • Li, J., Q. Zheng, J. Hu, Z. Fan, J. Zhu, T. Chen, B. Zhu, and Y. Xu, 2015: Wavelet analysis of coastal-trapped waves along the China coast generated by winter storms in 2008. Acta Oceanol. Sin., 34, 2231, doi:10.1007/s13131-015-0701-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, L., 1993: A study of winter subtidal sea level fluctuation along the northern coast of the South China Sea (in Chinese). Trop. Oceanol., 12, 5260.

    • Search Google Scholar
    • Export Citation
  • Luther, D. S., 1982: Evidence of a 4–6 day barotropic, planetary oscillation of the Pacific Ocean. J. Phys. Oceanogr., 12, 644657, doi:10.1175/1520-0485(1982)012<0644:EOADBP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madden, R., and P. Julian, 1972: Further evidence of global-scale 5-day pressure waves. J. Atmos. Sci., 29, 14641469, doi:10.1175/1520-0469(1972)029<1464:FEOGSD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madden, R., and P. Julian, 1973: Reply. J. Atmos. Sci., 30, 935940, doi:10.1175/1520-0469(1973)030<0935:R>2.0.CO;2.

  • Mathers, E. L., and P. L. Woodworth, 2001: Departures from the local inverse barometer model observed in altimeter and tide gauge data and in a global barotropic numerical model. J. Geophys. Res., 106, 69576972, doi:10.1029/2000JC000241.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, J.-H., and D. R. Watts, 2005: Response of the southwestern Japan/ East Sea to atmospheric pressure. Deep-Sea Res. II, 52, 16711683, doi:10.1016/j.dsr2.2003.08.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, J.-H., and D. R. Watts, 2006: Near 5-day nonisostatic response of the Atlantic Ocean to atmospheric surface pressure deduced from sub-surface and bottom pressure measurements. Geophys. Res. Lett., 33, L12610, doi:10.1029/2006GL026304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pawlowicz, R., B. Beardsley, and S. Lentz, 2002: Classical tidal harmonic analysis including error estimates in MATLAB using T_TIDE. Comput. Geosci., 28, 929937, doi:10.1016/S0098-3004(02)00013-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ponte, R. M., 1993: Variability in a homogeneous global ocean forced by barometric pressure. Dyn. Atmos. Oceans, 18, 209234, doi:10.1016/0377-0265(93)90010-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ponte, R. M., 1997: Nonequilibrium response of the global ocean to the 5-day Rossby–Haurwitz wave in atmospheric surface pressure. J. Phys. Oceanogr., 27, 21582168, doi:10.1175/1520-0485(0)027<2158:NROTGO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stammer, D., C. Wunsch, and R. M. Ponte, 2000: De-aliasing of global high frequency barotropic motions in altimeter observations. Geophys. Res. Lett., 27, 11751178, doi:10.1029/1999GL011263.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woodworth, P. L., S. A. Windle, and J. M. Vassie, 1995: Departures from the local inverse barometer model at periods of 5 days in the central South Atlantic. J. Geophys. Res., 100, 18 28118 290, doi:10.1029/95JC01741.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wunsch, C., and D. Stammer, 1997: Atmospheric loading and the oceanic “inverted barometer” effect. Rev. Geophys., 35, 79107, doi:10.1029/96RG03037.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Xu, Y., D. R. Watts, M. Wimbush, and J.-H. Park, 2007: Fundamental-mode basin oscillations in the Japan/East Sea. Geophys. Res. Lett., 34, L04605, doi:10.1029/2006GL028755.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhao, R., and X.-H. Zhu, 2016: Weakest winter South China Sea western boundary current caused by the 2015–2016 El Niño event. J. Geophys. Res. Oceans, 121, 76737682, doi:10.1002/2016JC012252.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhao, R., X.-H. Zhu, and X. Guo, 2016: The impact of monsoon winds and mesoscale eddies on thermohaline structures and circulation patterns in the northern South China Sea. Cont. Shelf Res., 143, 240256, doi:10.1016/j.csr.2016.06.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, X.-H., R. Zhao, X. Guo, Y. Long, Y.-L. Ma, and X. Fan, 2015: A long-term volume transport time series estimated by combining in situ observation and satellite altimeter data in the northern South China Sea. J. Oceanogr., 71, 663673, doi:10.1007/s10872-015-0305-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    (a) Map of the South China Sea. Black triangles indicate the PIES stations. The black box shows the region of (b). (b) Map of the northern South China Sea and the southern East China Sea. The five black triangles indicate PIES stations P1, P2, P3, P4, and P5 from north to south, respectively. The red dots represent the tide gauge stations used in this study.

  • Fig. 2.

    (a) Time series of the detided ocean bottom pressure Pbot for all PIES stations (gray lines except Pbot of station P3, which is shown by the black line). (b) Bandpass filtered Pbot for all PIES stations (red lines except Pbot of station P3, which is shown by the blue line). (c) Variance-preserving power spectrum density estimation of all PIES stations and Hong Kong. The gray line represents the 95% confidence level.

  • Fig. 3.

    Cross-spectral analysis results at the near 5-day period band for the tide gauges at (a)–(c) Kanmen, (d)–(f), Xiamen, (g)–(i) Zhapo, and (j)–(l) Hong Kong, and (m)–(o) the PIES station P3. The first two columns represent coherence and phase including SL or ocean bottom pressure Pbot lags from Patm in degrees, where the contour line interval is 30°. The last column represents coherence for the eastward and northward wind components at 10 m; however, only the larger value is plotted. The red dot for each subfigure denotes a tide gauge station or PIES station. Coherence and phase were omitted in mapping if the coherence was lower than the 95% confidence level.

  • Fig. 4.

    Cross-spectral analysis results at the near 5-day period band for the tide gauges in Hong Kong during 1988 to 1989. The black star indicates the Hong Kong tide gauge; the black triangle indicates the P3 station; and colored dots indicate the tide gauges for joint analysis with Hong Kong, with colors showing their (a) coherences or (b) phase lags. The lower panels show the coherence function between Hong Kong and (c) P3, (d) Zhapo, (e) Xiamen, and (f) Kanmen. The red lines show the 95% confidence level.

  • Fig. 5.

    As in Fig. 3, but that a 300° zonal low-pass filter was applied to the atmospheric surface pressure Patm, and the third column inserted shows the gain, defined as Patm/SL or Patm/Pbot.

  • Fig. 6.

    Bandpass filtered time series with cutoff periods of 4 and 6 days for the (a) SL of Hong Kong; (b) ocean bottom pressure Pbot of station P3; (c) atmospheric surface pressure Patm of P3; (d) Patm of P3, where the Patm was applied with a 300° zonal low-pass filter; and (e) Patm in the East China Sea, the coherence of which was the best with the Pbot of P3. All of the SL, Pbot, and Patm data are shown in blue, red, and black lines, respectively. The shaded areas represent 1 Jan to 15 Jun 2014, the period in which the Patm applied with a 300° zonal low-pass filter was weak and the coherence between the SL of Hong Kong and the Pbot of P3 was higher. It is noted that the ordinate ranges of these time series varied from ±1 to ±10.

  • Fig. 7.

    Illustration of the calculation of the CTW phase speed between the Hong Kong station and station P3; these stations are shown in red triangles. The red dashed line is assumed to have the same phase as that of Hong Kong at the 5-day period. The red solid line is assumed to be the propagation path to P3. The red arrow indicates the propagation direction. The results observed when the nonisostatic response was removed (not removed) from the ocean bottom pressure Pbot of P3 are denoted in red (blue) color, respectively.

  • Fig. 8.

    CEOF spatial pattern for the three leading modes. Distribution of (left) amplitude and (right) phase.

  • Fig. 9.

    Results of sensitivity and cross-spectral analysis. (a) The sensitivity of coefficients a and b to squared coherence. The black dot indicates coefficients a and b with the best coherence. Time series of (b) raw Pbot, (c) Rossby–Haurwitz waves, and (d) coastal-trapped waves in P3 at the near 5-day period band.

  • Fig. 10.

    Amounts of change in the coherences for ocean bottom pressure Pbot and (a) atmospheric surface pressure Patm and (b) winds when the nonisostatic responses forced by Rossby–Haurwitz waves were removed from Pbot. Red dots indicate station P3. Gray lines indicate zero values.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 346 114 1
PDF Downloads 239 66 2