Next Article in Journal
The Prediction and Dynamic Correction of Drifting Trajectory for Unmanned Maritime Equipment Based on Fully Connected Neural Network (FCNN) Embedding Model
Previous Article in Journal
Numerical Simulation of Wave-Induced Scour in Front of Vertical and Inclined Breakwaters
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JMSE
Volume 12
Issue 12
10.3390/jmse12122263
jmse-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Near-Inertial Oscillations of Thermocline in the Shelf Area off Vladivostok, the Sea of Japan, from a Set of Thermostrings

by
Olga Trusenkova
,
Igor Yaroshchuk
,
Alexandra Kosheleva
*,
Aleksandr Samchenko
,
Alexander Pivovarov
and
Vyacheslav Dubina
V.I. Il’ichev Pacific Oceanological Institute, Far Eastern Branch, Russian Academy of Sciences, 690041 Vladivostok, Russia
*
Author to whom correspondence should be addressed.
J. Mar. Sci. Eng. 2024, 12(12), 2263; https://doi.org/10.3390/jmse12122263
Submission received: 8 November 2024 / Revised: 5 December 2024 / Accepted: 6 December 2024 / Published: 9 December 2024
(This article belongs to the Section Physical Oceanography)
Browse Figures
Versions Notes

Abstract

:
The shelf area off Vladivostok in the Sea of Japan is known by the intense internal wave activity investigated for many years. The present contribution to these studies is based on data collected on 3–14 October 2022, from four moorings aligned across isobaths and equipped with thermostrings. Multivariate analysis is performed in the depth–time domain, while timescales and directions and speeds of temperature anomaly movement are estimated from wavelet transform. Approximately 50% of the variance results from vertical stratification changes, i.e., thermocline deepening or shoaling, and temperature anomalies on different timescales moved towards the shoaling seafloor. For the first time, near-inertial (NI) oscillations are detected throughout the record and turn out to be the most intense among the 6 to 70 h timescales, moving with the speeds of 0.41–0.55 m/s, although previous attention was paid to the semidiurnal internal tide. A frequency decrease, i.e., red shift, of the NI oscillations is detected towards shallower water, with the frequency eventually becoming subinertial, and is explained by anticyclonic relative vorticity at the eastern side of the mushroom-like structure detected from thermal satellite imagery. The semidiurnal and two-day oscillations were detected, moving with the speeds of 0.95–1.11 and 0.15–1.17 m/s, respectively. The two-day timescale, never reported before, is considered as a difference one caused by nonlinearity. These results are interpreted as the propagation of an internal wave generated at the steep slope offshore to the inner shelf.

1. Introduction

One of the extended shelf areas in the Sea of Japan is Peter the Great Bay, located off Vladivostok. The hydrophysical test site of the Pacific Oceanological Institute (POI) is situated in its southwestern part, Posyet Bay (Figure 1). The shelf in the Sea of Japan is very narrow, so the shelf area, with a distance of 15–20 km from the POI test site to the shelf break at 100 m of depth, can be considered rather wide. The slope at the test site is gentle, with a deepening of 0.3 m per 100 m of distance, while the continental slope off Peter the Great Bay is very steep: for every 100 m the depth changes by more than 10 m. The mean circulation in this area is traditionally represented by the cold Primorye (Liman) Current flowing southwestward along the 100 m isobath; however, numerous warm anticyclonic eddies frequently entering this area from the southwest and south are recorded from satellite imagery [1,2,3].
These conditions facilitate the generation of internal waves at the slope and their propagation onshore, so they represent an important physical mechanism of deep ocean exchange with the shelf, vertical mixing, and supply of energy to the shelf and nutrients into the euphotic layer, and affect coastal marine life in other ways [4,5]. For this reason, numerous experimental studies on the internal waves were conducted at the POI test site in Posyet Bay, considering their generation, propagation, transformation, and dissipation. A lot of attention has been paid to the intense, high-amplitude, quasi-, and nonlinear waves such as internal bores and solitons, studied from field data, modeling, and laboratory experiments, for instance, in papers [6,7,8,9,10,11,12,13,14,15,16]. An important mechanism of the internal wave generation is the interaction of tides with the continental slope, and these waves are referred to as internal tides. It is well known that the semidiurnal (SD) M2 tidal constituent is a major physical mechanism of the internal tide generation [17], and the M2 tide prevails in the Sea of Japan and, in particular, in Peter the Great Bay [18,19].
Near-inertial (NI) waves, i.e., waves near the local inertial frequency, represent an important class of internal waves. Inertial oscillations generated in the mixed layer by strong winds, for instance in cyclones, give rise to NI internal waves propagating down to the pycnocline and facilitating the diapycnic exchange in the stratified sea; they undergo transformation in interactions with oceanic fronts, mesoscale and submesoscale dynamic structures, and topography [20,21,22,23]. Wind is much stronger in an open sea compared to nearshore areas, so NI waves are likely to be generated there and to come to the shelf from offshore. It is important that the frequency of NI waves changes inside flow structures with anticyclonic or cyclonic relative vorticity, with the former resulting in the frequency decrease, i.e., red shift, and the latter to the blue shift, i.e., frequency increase [24]. These effects were revealed in current speeds measured by Doppler profiler installed at the SEAWATCH WaveScan buoy moored in Posyet Bay [25], and it is worthwhile to look for them in temperature data from POI test site.
Considering the above, the purposes of the study are
  • Multivariate analysis, performed in the same way as in [25,26], of thermocline deepening and shoaling which can be related to the internal wave activities;
  • Timescale estimation of this variability;
  • Checking contributions of NI oscillations to the thermocline variability.
For this purpose, data obtained from thermostrings installed at moorings at the POI test site in October 2022 are used. To analyze thermocline on a whole, multivariate analysis is applied, and to estimate timescales in nonstationary data, wavelet transform is used.

2. Materials and Methods

The temperature data analyzed in this study were obtained from a set of moorings installed at the shelf area in the southwestern Peter the Great Bay, the Sea of Japan, where POI FEB RAS keeps the hydrophysical test site [16,27]. The geographical coordinates of the test site are 42°33.531′ N, 131°06.357′ E. Four moorings (hereafter referred to as St-1, St-2, St-3, and St-4, respectively) equipped with thermostrings were aligned in the cross-isobath direction (Figure 1); the measurements were performed from 3 October 2022 to 14 October 2022; the time used is UTC+10 (VLAT).
Data from St-2 and St-3 are available for the entire installation period, i.e., slightly less than 11 days, while data from St-1 are available for 130 h only, and from St-4 they were retrieved from October 4 onwards due to the instrument failure (Table 1). Instrumentation was discussed in detail by [28]. However, it should be noted that the thermistors were placed below the mixed layer every 1 or 1.5 m and the deepest one was close to the sea floor (Table 1). Measurements were performed every 1 s and errors were estimated at 0.06 °C, while the subsequent smoothing and time averaging up to 1 min made it possible to reduce the errors to 0.01 °C [28]. Data characteristics are summarized in Table 1; an example of temperature time series at the characteristic depths for St-3 is shown in Figure 2, and time-averaged temperature profiles for St-1–St-4, with root mean square spreads, are shown in Figure 3.
Using data with a 1 min time step, vertical temperature gradients were computed as G = (Tk+1 − Tk)/(zk+1 − zk) and referred to the depth zk = (zk + zk+1)/2 where T is temperature, G is temperature gradient, z is depth, k = 1, 2, …, M, k is the level number, and M is number of levels (Table 1). Assuming that temperature is a fairly smooth function, data errors should be summarized when computing gradients and, therefore, they became equal to 0.02 °C/m. However, the differentiation increased noise level, which is clearly seen in the time-averaged profiles of temperature gradient (Figure 3, the second from the left panel).
Decomposition of spatiotemporal data to empirical orthogonal functions (EOFs) is the commonly used method of multivariate analysis, enabling the separation of spatial and temporal components and data representation as a set of variability modes ordered by covered fractions of variance. In this paper, the data are represented in the depth–time domain; EOFs, i.e., basis functions, are vertical profiles; and principal components (PCs) are temporal functions. This kind of analysis was performed earlier for vertical profiles of current velocity from the SEAWATCH WaveScan buoy [25] and of temperature obtained from the autonomous Aqualog profiler moored off the eastern Peter the Great Bay [26]; the multivariate technique was discussed in detail by [25,26]. Here, EOF analysis is performed both for temperature and temperature gradient, thus facilitating interpretation of the results. The gradient data are very noisy due to differentiation, so they are preliminarily low-pass filtered, with the cut-off period of 2 h, enabling the derivation of reasonable modes. For comparability, temperature data are also low-pass filtered. Low-pass-filtered temperature and its gradient are hereafter referred to as T and G, respectively.
Mode 1 derived from T accounts for about half of the data variance for St-1–St-3 and more than 60% of the variance for St-4, while Mode 1 from G covers about a quarter of the variance (Table 1); the latter can be explained by the increased noise but this is acceptable, as Modes 1 for T and G are consistent. Most of the remaining temperature variance accounts for processes in the upper and bottom layers which are not considered here, as the focus of the study is on the thermocline, and about 10% of the variance is related to noise. Therefore, the analysis hereafter covers Mode 1 only. For brevity, EOFs of Mode 1 for T and G are hereafter referred to as EOF1T and EOF1G and PCs as PC1T and PC1G, respectively. Decomposition based on the original, not-filtered temperature covers the same fraction of data variance and yields practically identical modes, while the highest frequency variabilities, with timescales less than 2 h, do not make any statistically significant contribution to spectra.
To estimate the validity of the EOFs by depth, correlations are estimated of Tk and Gk with EOF1T and EOF1G, respectively (Tk and Gk are temperature and its gradient at the k-th level, k is the level number). Modes are considered valid at the depths where correlations reach the confidence level of 95%. The Fisher test is applied for the estimation of statistical significance, considering number degrees of freedom N* (Table 1) estimated for all possible time lags [29]. N* for G are larger than those for T due to higher noise level, while both generally decrease with the depth, in accordance with variance (see the root mean square spread for T in Figure 3). Anomalies of T and G are estimated as products of EOF1T and PC1T and of EOF1G and PC1G, respectively; these anomalies averaged over time and within depths of statistical significance for Mode 1 are hereafter referred to as Ta and Ga, respectively (Table 1). Note that decomposition to EOFs is a kind of averaging [30] and averaging decreases data errors as (N*)1/2 [29].
To estimate timescales of temporal variability, wavelet analysis suitable for nonstationary time series is applied [31]. The Morlet mother wavelet of the 6th order is used for computing wavelet transform, the squaring of which yields the spectrum, providing good resolution in the frequency domain and allowing for estimation of coherences and phase shifts of joint spectra computed as products of individual wavelet transforms. The 90% confidence level is used based on the theoretical red noise spectrum, and cones (zones) of influence of edge effects (COI) are estimated, with anything inside these zones being considered dubious [31]. Areas out of COI and where the power exceeds 90% confidence level are considered as valid. Periods of oscillations for every time step are estimated based on maximum power within considered frequency ranges. The inverse wavelet transform is used for data filtering.

3. Results and Discussion

In October 2022, the pycnocline starts from 13–16 m where temperature gradients jump, on average, from less than 0.05 °C/m to more than 0.1 °C/m (Figure 3). Some negative temperature gradients, exceeding errors, are present in the original data: there are 17.5%, 10.3%, 8.4%, and 3.9% of them at St-1–St-4, respectively, meaning that temperature increases with depth. Time-averaged values are negative at the depths of 6 and 12 m and are equal to –0.17 and –0.15 °C/m, respectively, for St-1, where negative gradients are the most numerous. Considering N* as equal to 35 and 27 for these levels, gradient errors should be decreased to 0.004 and 0.003 °C/m, respectively, so the negative mean gradients exceed them (by absolute values).
In the Sea of Japan, seawater density and its vertical gradient are mostly determined by temperature [32]; however, in the shelf areas of Peter the Great Bay, salinity can be important for density variations due to the mixing of coastal and open-sea waters. Warmer and saltier water from the open sea approaching the coast can be denser than colder and fresher, due to run-off, coastal water, resulting in negative vertical temperature gradients, although density vertical stratification remains stable. This effect is the strongest for the shallowest St-1.
Spatial patterns of Mode 1 are represented by correlations of EOF1T and EOF1G with T and G, respectively (Figure 3, second from the right and right panels, respectively). It can be seen that Mode 1 for T is statistically significant at the depths of 16 to 40 m, i.e., within the thermocline and the strongest mean temperature gradients. The depth of the EOF1T maximum decreases by 5.5 m with the shoaling seafloor from St-4 to St-1, with this difference exceeding the depth step of the measurements by more than three times (1–1.5 m; Table 1). Mode 1 for G features two statistically significant extremes in depth, one in the upper layer and the other in the bottom layer above and below the maximum of mean temperature gradient, respectively (Figure 3, right panel). The depth of the lower EOF1G maximum decreases by 4.5 m from St-4 to St-1, similar to the EOF1T maximum, while the upper EOF1G maximum remains almost at the same depth at St-1–St-4 (Table 1).
Mode 1 accounts for large temperature anomalies of 7–9 °C (Table 1). Considering the decrease in errors due to averaging when computing EOFs (see Section 2) and N* = 14–28 at the depths where EOF1T is statistically significant, original errors of 0.01 °C become 0.0027–0.0019 °C, i.e., temperature anomalies substantially exceed data errors. Similarly, gradient anomalies are equal to 0.14–0.46 °C/m, i.e., they exceed data errors, considering N* = 13–32 within the depths of statistical significance and the decrease in data errors from 0.02 to 0.006–0.003 °C/m.
Mode 1 features consistent changes of temperature and its gradient, with correlations between PC1T and PC1G being equal to 0.81–0.87 for St-1–St-4; consistent temporal variability is clearly seen in the PC1T and PC1G time series (Figure 4). This means the simultaneous increase (decrease) of temperature and its upper-layer gradient and decrease (increase) of the lower-layer gradient, i.e., changes of vertical stratification demonstrating the thermocline deepening (shoaling). Internal waves accompanied by similar changes in the vertical stratification were detected at the POI hydrophysical test site for many years [16]. As for the analyzed record, the most intense fluctuations of PC1T and PC1G occurred at 12–48 h of the record (4–5 October) and at 168–192 h of the record (10–11 October), being almost in-phase at St-1–St-4 (Figure 4). These events also manifest themselves in the original temperature for St-3 at the depth of 29.5 m, where EOF1T reaches its maximum (Figure 2, left panel), and can be related to high-amplitude internal waves, such as internal bores and solitons, which were observed earlier at the test site [16]. These events were most likely induced by meteorological conditions, as a deep trough with a front approached the Sea of Japan from the west on October 4 and a cyclone arrived from the Yellow Sea on 9 October, both accompanied by the severe winds of 13–18 m/s and surface waves of 3–4 m [33].
To estimate variability timescales for Mode 1, wavelet spectra were computed. Individual spectra for PC1T or PC1G turned out uninformative; therefore, their joint spectra were analyzed. The joint spectra were similar for St-1–St-4, considering shorter records at St-1 and St-4 (Figure 5). These spectra were statistically significant for the entire range between 6 and 70 h, with few exceptions, and the coherence was high, exceeding 0.9. The phase shifts were low, meaning that oscillations of PC1T and PC1G were almost in-phase, although during the events of high-amplitude oscillations (internal bores), the gradients led the temperature by 2–4 h, which can be clearly seen from the time series charts (Figure 4).
In the analyzed records, spectral maxima hit the timescales of 9–12, 16–20, and 48–50 h, with the red noise covering no more than 2% of their power (Figure 5, Table 1). The strongest peaks in the time-averaged spectra for all four moorings correspond to near-inertial (NI) oscillations (15.9–18.3 h), with the intensity decreasing with the shoaling seafloor from St-4 to St-1 (Figure 6, left panel). No statistically significant peaks were recorded in the Fourier spectra, which can be explained by the data nonstationarity. Indeed, the NI timescale was the strongest on the first three days and from the 7th through 10th days of the record (0–71 and 121–235 h, respectively), as can be seen from the charts of periods corresponding to the maximum power shown in Figure 7, left panel (for the times when they were out of COI). In between, on the 4th and 5th days of the record (72–120 h), SD oscillations were the strongest (Figure 7, left panel), although power on the NI timescale remained statistically significant. In the earlier studies based on observation data at the POI test site, the attention was focused on the SD tidal internal waves [6,7,10,11,12,13,16,27]. However, the present results show that NI oscillations dominated in October 2022. In particular, the high-amplitude events at 12–48 and 168–192 h of the record occurred on the NI timescale after the wind strengthening, as discussed above, while wind is a well-known forcing of NI variability [20].
It is well known that internal waves cover the range between buoyancy and inertial frequencies. The inertial period is equal to 17.7 h at the latitude of the test site, while, on average, the estimated periods of NI oscillations increase with the shoaling seafloor from 15.9 h at St-4 to 16.7–17.1 h at St-3 and St-2 and to 17.9–18.3 h at St-1, with the latter exceeding the inertial period (Figure 7, Table 1). The step between discrete scales (periods) used for wavelet transform is equal to 0.38–0.42 h in the NI range, i.e., less than half of the period shift between the moorings, meaning that this shift is well resolved. The charts in Figure 7, left panel, clearly demonstrate that NI periods for St-4 shown with red lines are mostly smaller than those for the other moorings, especially during 24–48 h of the record; while periods for St-2 and St-3 are often very close (NI periods for St-1 detected only before 75 h due to the short record). It is known that frequencies of NI internal waves decrease, i.e., they are subjected to the red shift, and can even drop below the inertial frequency, i.e., cross the inertial limit in areas of anticyclonic relative vorticity [24]. The red shift was observed in anticyclonic eddies in various regions of North Pacific, in particular around the Tsushima Warm Current in the Sea of Japan [34] and in current speed in the southwestern Peter the Great Bay [25].
A thermal (infrared) image with resolution of 100 m acquired from Landsat-9 on 5 October 2022 [35] demonstrates cold water near the coast and warm water in the southern and eastern areas, with a mushroom-like structure oriented from the southeast to northwest (Figure 8). Such mesoscale or, in this case, submesoscale structures consist of a jet along the mushroom stipe ending with a vortex dipole. They are forced by wind gusts and are characterized by nonstationarity and nonlinearity with severe deformations, featuring negative (positive) relative vorticity to the right (left) side of a stipe [36]. In this area, one can expect the southwestward flowing cold Primorye Current [1]; however, anticyclonic eddies and even the eastward current carrying warm water from the southwest to the northeast can develop there in October under the forcing of anticyclonic wind stress curl, as was shown from infrared satellite imagery [2] and surface buoy data [37]. The considered mushroom-like structure features a stipe oriented from the southeast to northwest, as shown in Figure 8 by a white arrow, and anticyclonically (cyclonically) bended streamers northeastward (northwestward) of it. It is reasonable to suggest that an area of negative vorticity extended eastward towards the test site where the observations were performed, thus affecting NI motions. This occurred mostly at St-1–St-3 and also at St-4 at the times when NI period increased, such as at 72–80 and 120–140 h of the record and at 190 h and later (Figure 7).
The maximum power of the T–G spectra was at the periods of 9.4–9.6 h, between 72 and 120 h for St-1–St-3 and between 72 and 90 h for St-4 (Figure 5, Figure 6 and Figure 7), which is near the SD internal tide generated at the continental slope off Peter the Great Bay and frequently observed at the POI test site [16]. The SD internal tide generated by the M2 tidal constituent prevails in the Sea of Japan and, in particular, in Peter the Great Bay, while the diurnal tide also contributes [18,19]. It is worthwhile to compare temperature and sea level variability at the SD timescale. For this purpose, bottom pressure data from the sensor installed at St-2 are used for estimation of sea level anomalies (thereafter ST-2 SLA). As pressure at St-1–St-4 is almost identical, the St-2 data are enough for analysis. For comparison, tidal sea level anomalies from the ADMIRALTY TotalTide product, version 7.6.1.0 [38], are used (hereafter TotalTide SLA), based on the Kamaishi reference port at 39°16′ N 141°53′ E and predicted for the Furugelm Island secondary port at 42°28′ N 130°56′ E located close to the POI test site. The root mean square deviations of the St-2 and TotalTide SLA are equal to 0.11 and 0.10 m, respectively. This exceeds pressure sensor errors equal to 0.1% of depth, i.e., 0.04 m, as the depth is 41 m at St-2. Normalized St-2 and TotalTide SLA generally followed each other, especially at 72–167 h of the record; however, there was a remarkable difference at 168–192 h (Figure 9, left top panel). The SD and diurnal oscillations can be clearly seen in this chart for both time series and in the original (left bottom panel) and time-averaged (right panel) wavelet spectra for St-2 SLA. The mean spectrum has a peak at 12.36 h, corresponding to the M2 tidal constituent (12.4 h [39]); the same peak is present in the spectrum of TotalTide SLA.
The temperature oscillations near the SD timescale were the most intense at St-2, resulting in a weak but statistically significant peak at the period of 9.6 h in the time-averaged spectrum (Figure 6, left panel, Table 1). At 100–130 h of the record, there was power leakage, i.e., energy transfer from the SD to NI timescale, followed by the weakening of the former and strengthening of the latter; this effect was the strongest at St-2, weak at St-3, and absent at the deepest St-4 where the near-SD oscillations were the weakest (Figure 5). Interaction between the timescales can be explained by nonlinearity. The periods in the SD range for temperature are shorter than those of the M2 tide. This difference might be explained by properties of internal waves coming from the slope to the shelf; however, in this case, the SD oscillations should have been the strongest at the deepest St-4, where, in reality, they were the weakest. Therefore, this point remains unclear and requires further research. The period of maximum power at St-4 dropped to 7.4 h at about 90 h of the record and then to 6.6 h (Figure 7, left), which is possibly related to nonlinear generation of sum frequencies. Namely, sum frequency fm ~ 1/7.0 h−1 can be obtained as fm = fsd + fi, where fsd = 1/12.4 h−1 is SD tidal frequency and fi ~ 1/15.9 h−1 is NI frequency. The NI oscillations were weak during this time at St-4, so temperature anomalies on the sum timescale should have come from another area, probably from the outer shelf or slope.
In the temperature spectra, two-day oscillations are detected, with periods of 48–50 h, which are out of COI from 66 through 192 h of the record for St-2 and St-3 and from 86 through 192 h for St-4 (Figure 5); they can be clearly seen in the time-averaged spectra (Figure 6, left panel; Table 1). The two-day oscillations are the most intense at St-2, where the semidiurnal oscillations are also the most intense. As can be seen from temperature time series taken at the depths of the EOF1T maximum and band-pass filtered in the 6–20 h range, strong and weak oscillations alternate bi-daily in concert at all moorings (Figure 2, right panel). This can be also seen in the PC1T and PC1G time series (Figure 4). Although a two-day maximum is absent from the St-1 spectrum due to the insufficient record length, this variability is clearly seen in Figure 2 (right panel) and Figure 4. The two-day frequency (f2d) can be considered as a difference between the SD and NI ones, arising because of nonlinearity, namely, f2d = fsd − fi. Taking fsd ~ 1/12 h−1 and fi ~ 1/16 h−1, we obtain the difference period P2d = 1/f2d ~ 48 h. At the same time, the periods of SD and NI temperature oscillations somewhat differ between the moorings and so should P2d; therefore, this suggestion requires further checking. The two-day timescale is also present in the St-2 SLA variability, ranging from 48.3 to 50.6 h and with mean period of 50.6 h (Figure 9); however, there is no variability on this timescale in the TotalTide SLA, implying its close relationship with the internal waves. At the same time, the NI timescale is a minimum rather than maximum in the St-2 SLA spectrum, as discussed above; the causes of this effect are unclear.
Joint wavelet spectra between PC1T are computed for pairs of the adjacent moorings, enabling the estimation of directions and speeds of temperature anomaly movement on different timescales. These spectra resemble each other and the joint spectra between PC1T and PC1G for individual moorings: they feature maxima at the SD, NI, and two-day timescales, with very low red noise (compare Figure 5 and Figure 10, Figure 6 left and right panels, and Table 1 and Table 2). The joint spectra between the moorings are statistically significant for periods of more than 3–6 h, and coherency exceeds 0.9. As is the case for the individual moorings, the maximum power was at the NI timescale, with the exception of the time between 72 and 120 h, when the SD timescale dominated (Figure 10). The NI oscillations were the most intense for the pair of the two deepest moorings, St-3 and St-4, and the SD ones for the pair of two shallowest moorings, St-1 and St-2, with the latter yielding a weak but statistically significant peak in the mean spectrum (Figure 6, right panel). The frequency decrease in the NI oscillations and intensification of the SD and two-day oscillations along the shoaling seafloor also occurred in the same manner as found for the individual moorings (Table 2, Figure 6 and Figure 7, right panels).
Phase shifts are of the same sign for three mooring pairs, indicating the anomaly movement from St-4 towards St-1, i.e., across the isobaths in the direction of the shoaling seafloor. The speed of anomaly movement on different timescales (U) can be estimated as U = L/(P·Δϕ/2π), where L is the distance between the moorings, P is the period, Δϕ is the phase shift in radians, and π = 3.14. Mean speeds for the SD and NI timescales are similar in all cases, amounting to 0.95–1.11 and 0.41–0.55 m/s, respectively (Table 2). Speed estimates for the SD timescale computed for the period of its predominance in the spectra (72–120 h) are very close to those for the entire record out of COI (Table 2). Unexpectedly, the speeds on the two-day timescale from St-4 to St-3 and from St-3 to St-2 differ by an order of magnitude, amounting to 1.17 and 0.15 m/s, respectively, while this timescale is the most intense at the shallowest St-2 and St-1 (Figure 5 and Figure 10). The cause of such blocking is unclear (the short record at St-1 prevents the estimations of two-day timescale between St-2 and St-1). Anomaly movement towards the shoaling seafloor can be explained by the propagation of internal waves from the continental slope, as discussed in [16], where the speeds of high-amplitude events were estimated at 0.1–0.6 m/s [16]. Climatological speeds of internal wave propagation off Peter the Great Bay in summer were estimated at 0.2–0.4 m/s, based on the WOA18 Atlas [40]. These values are in good agreement with our estimates of the most energetic NI timescale for the entire record.

4. Conclusions

Short-term temperature variability was studied in the shelf area off Vladivostok, the Sea of Japan, on 3–14 October 2022, using data collected from four moorings installed at the POI test site, aligned across isobaths at the depths of 39–47 m and equipped with thermostrings. Multivariate analysis was performed in the depth–time domain, and about 50% of the total variance accounts for variability of vertical stratification, i.e., the thermocline deepening or shoaling, manifesting itself in the concerted changes of temperature and its vertical gradient. In previous studies, attention was focused on high-amplitude events associated with internal bores [16], while the multivariate approach enables investigation of variability throughout the record, with timescales of nonstationary processes being successfully estimated based on wavelet transform.
It was found, unexpectedly, that throughout most of the record, NI oscillations were the most intense among the 6 to 70 h timescales, although in previous studies, such as [6,7,10,11,12,13,16,27], attention was focused on the SD internal tide, while the NI internal waves were not reported before in the studied area. Two high-amplitude NI events occurred on 4–5 and 10–11 October, preceded by severe winds and surface waves caused by meteorological conditions—a deep trough approached the Sea of Japan from the west and a cyclone from the Yellow Sea in the first and second cases, respectively. On the NI timescale, speeds of temperature anomaly movement between the moorings were estimated at 0.41–0.55 m/s throughout the record, based on joint wavelet spectra, which is in line with the earlier estimates for high-amplitude events (internal bores) [16,40]. It is reasonable to consider this movement as being associated with the NI internal waves abundant in the world ocean [20] and documented for the Sea of Japan, in particular by [34]. The red shift, i.e., frequency decrease and period increase, of the NI oscillations was detected along the shoaling seafloor and the period eventually exceeded the inertial period, 17.7 h, at the latitude of the test site. The frequency decrease in NI internal waves occurred in areas of strong currents with anticyclonic relative vorticity [24], for instance, near the Tsushima Warm Current in the Sea of Japan [34], in the continental slope of the East China Sea [41], and in the South China Sea [42]. Examination of high-resolution thermal imagery from the Landsat-9 satellite [35] revealed a mushroom-like structure, with anticyclonic vorticity at its eastern edge near the test site, resulting in the red shift.
The SD temperature oscillations were the most intense at the two shallowest stations; they dominated in the spectra on 6–8 October (from 72 to 120 h of the record), while the NI oscillations weakened at that time. The detected periods of 9.4–9.6 h were less than those characteristic of SD tidal constituents; in particular the M2 (12.4 h) tide, which prevails in the Sea of Japan [18,19], is present in the bottom pressure and is predicted by the ADMIRALTY TotalTide product [38]. The intense high-amplitude events, probably internal bores coming from the open sea, affected tidal SLA oscillations, resulting in the energy exchange between the diurnal and SD timescales or in a sharp drop in SLA; these effects were absent from the predicted TotalTide SLA. On the SD timescale, speeds of movement of temperature anomalies were estimated at 0.95–1.11 m/s, which is about twice more than those for the NI timescale. At the deepest mooring, the generation of sum (SD plus NI) frequencies was detected on 8 October, resulting in oscillations with the periods of 6.6–7.4 h; this effect can be explained by nonlinearity.
The two-day oscillations of temperature and bottom pressure, never reported before, were detected, being the most intense at two shallower moorings where the semidiurnal oscillations were also the most intense. The speeds on these timescales differ by an order of magnitude between the deeper and shallower moorings, amounting to 1.17 and 0.15 m/s, respectively; the cause of such a blocking is unclear. The two-day frequency can be considered as a difference between the SD and NI ones, arising from nonlinearity. The interaction of SD and NI internal waves was discussed, for instance, by [23]. However, the detected periods somewhat differ from those computed as differences, so this suggestion requires further checking.
Some issues remain unclear in this study, such as the deviation of the near-SD timescale for temperature anomalies from the M2 tide, forcing of the two-day oscillations, and causes of their blocking in the middle of the mooring line. Some of this could be explained by the properties of internal waves coming from the slope, but this assumption requires checking in future. For further research, simultaneous moorings are needed in the inner shelf and at the shelf break and slope off Vladivostok.

Author Contributions

Conceptualization, O.T. and I.Y.; methodology, O.T. and I.Y.; software, O.T. and A.K.; formal analysis, O.T. and I.Y.; investigation and experimental studies, I.Y., A.S., and A.P.; data curation, A.K. and V.D.; writing—original draft preparation, O.T.; writing—review and editing, O.T., I.Y., and A.K.; visualization, O.T., V.D., and A.K.; supervision, I.Y.; project administration, I.Y. project funding acquisition, I.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the State Tasks Nos. 124022100079-4 and 124022100074-9 for POI FEB RAS and by the project of the Ministry of Science and Education of Russia No. 075–15-2022–1127.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Temperature and bottom pressure data are archived at the Laboratory of Statistical Hydroacoustics (POI FEB RAS) and will be made available upon request with some restrictions applied to the availability of these data. Satellite images are available on [35]. The predicted tides were produced by ADMIRALTY TotalTide [38].

Acknowledgments

The authors acknowledge C. Torrence and G.P. Compo for their free software for wavelet analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Danchenkov, M.A.; Lobanov, V.B.; Riser, S.C.; Kim, K.; Takematsu, M.; Yoon, J.-H. A history of physical oceanographic research in the Japan/East Sea. Oceanography 2006, 19, 18–31. [Google Scholar] [CrossRef]
  2. Ladychenko, S.Y.; Lobanov, V.B. Mesoscale eddies in the area of Peter the Great Bay according to satellite data. Izv. Atmos. Ocean. Phys. 2013, 49, 939–951. [Google Scholar] [CrossRef]
  3. Budyansky, M.V.; Ladychenko, S.Y.; Lobanov, V.B.; Prants, S.V.; Udalov, A.A. Evolution and structure of a mesoscale anticyclonic eddy in the northwestern Japan Sea and its exchange with surrounding waters: In situ observations and Lagrangian analysis. Ocean Dyn. 2024, 74, 901–917. [Google Scholar] [CrossRef]
  4. Johnson, J.; Chapman, P. Preface “Deep Ocean Exchange with the Shelf (DOES)”. Ocean Sci. 2011, 7, 101–109. [Google Scholar] [CrossRef]
  5. Garwood, J.C.; Musgrave, R.C.; Lucas, A.J. Life in internal waves. Oceanography 2020, 33, 38–49. [Google Scholar] [CrossRef]
  6. Serebryany, A.N. Internal waves on a shelf and in vicinity of continental slope as registered by towed line temperature sensor. Oceanology 1987, 27, 225–226. (In Russian) [Google Scholar]
  7. Serebryany, A.; Khimchenko, E.; Popov, O.; Denisov, D.; Kenigsberger, G. Internal Waves Study on a Narrow Steep Shelf of the Black Sea Using the Spatial Antenna of Line Temperature Sensors. J. Mar. Sci. Eng. 2020, 8, 833. [Google Scholar] [CrossRef]
  8. Grimshaw, R.; Pelinovsky, E.; Talipova, T. Solitary wave transformation in a medium with sign-variable quadratic nonlinearity and cubic nonlinearity. Phys. D Nonlinear Phenom. 1999, 132, 40–62. [Google Scholar] [CrossRef]
  9. Helfrich, K.R. Decay and return of internal solitary waves with rotation. Phys. Fluids 2007, 19, 026601. [Google Scholar] [CrossRef]
  10. Navrotsky, V.V.; Lozovatsky, I.D.; Pavlova, E.P.; Fernando, H.J.S. Observations of internal waves and thermocline splitting near a shelf break of the Sea of Japan (East Sea). Cont. Shelf Res. 2004, 24, 1375–1395. [Google Scholar] [CrossRef]
  11. Navrotsky, V.V.; Liapidevskii, V.Y.; Pavlova, E.P.; Khrapchenkov, F.F. Transformations and effects of internal waves in the near-shore region of sea. J. Oceanol. Res. 2019, 47, 230–245. (In Russian) [Google Scholar] [CrossRef]
  12. Novotryasov, V.V.; Karnaukhov, A.A. Nonlinear interaction of internal waves in the coastal zone of the Sea of Japan. Izv. Atmos. Ocean. Phys. 2009, 45, 262–270. [Google Scholar] [CrossRef]
  13. Novotryasov, V.V.; Zakharkov, S.P.; Stepanov, D.V. Internal tides in the coastal zone of the Sea of Japan in autumn. Russ. Meteorol. Hydrol. 2016, 41, 564–568. [Google Scholar] [CrossRef]
  14. Kamchatnov, A.M.; Kuo, Y.-H.; Lin, T.-C.; Horng, T.-L.; Gou, S.-C.; Clift, R.; El, G.A.; Grimshaw, R.H.J. Undular bore theory for the Gardner equation. Phys. Rev. E 2012, 86, 036605. [Google Scholar] [CrossRef]
  15. Kirillov, V.V.; Liapidevskii, V.Y.; Sutorikhin, I.A.; Khrapchenkov, F.F. Special Aspects of Transformation of Non-Linear Internal Waves on the Shelf and in a Deep Lake. Fluid Dyn. 2023, 6, 1087–1097. [Google Scholar] [CrossRef]
  16. Yaroshchuk, I.; Liapidevskii, V.; Kosheleva, A.; Dolgikh, G.; Pivovarov, A.; Samchenko, A.; Shvyrev, A.; Gulin, O.; Korotchenko, R.; Khrapchenkov, F. Observation and Modeling of Nonlinear Internal Waves on the Sea of Japan Shelf. J. Mar. Sci. Eng. 2024, 12, 1301. [Google Scholar] [CrossRef]
  17. Morozov, E.G. Semidiurnal internal wave global field. Deep Sea Res. I Oceanogr. Res. Pap. 1995, 42, 135–148. [Google Scholar] [CrossRef]
  18. Rabinovich, A.B.; Shevchenko, G.V.; Sokolova, S.E. An estimation of extreme sea levels in the northern part of the Sea of Japan. La Mer 1992, 30, 179–190. [Google Scholar]
  19. Supranovich, T.I.; Bogdanov, K.T. Tides. In Hydrometeorology and Hydrochemistry of Seas. The Sea of Japan. Hydrometeorological Conditions; Vasil’ev, A.S., Terziev, F.S., Kosarev, A.N., Eds.; Gidrometeoizdat: St. Petersburg, Russia, 2003; Volume 8, pp. 257–269. (In Russian) [Google Scholar]
  20. Alford, M.H.; MacKinnon, J.A.; Simmons, H.L.; Nash, J.D. Near-Inertial Internal Gravity Waves in the Ocean. Annu. Rev. Mar. Sci. 2016, 8, 95–123. [Google Scholar] [CrossRef]
  21. Shcherbina, A.Y.; Talley, L.D.; Firing, E.; Hacker, P. Near-Surface Frontal Zone Trapping and Deep Upward Propagation of Internal Wave Energy in the Japan/East Sea. J. Phys. Oceanogr. 2003, 33, 900–912. [Google Scholar] [CrossRef]
  22. Nam, S. Near-Inertial Current Variability off the East Coast of Korea. Ph.D. Thesis, Seoul National University, Seoul, Republic of Korea, 2006. [Google Scholar]
  23. Noh, S.; Nam, S. Observations of enhanced internal waves in an area of strong mesoscale variability in the southwestern East Sea (Japan Sea). Sci. Rep. 2020, 10, 9068. [Google Scholar] [CrossRef] [PubMed]
  24. Kunze, E. Near-Inertial Wave Propagation in Geostrophic Shear. J. Phys. Oceanogr. 1985, 15, 544–565. [Google Scholar] [CrossRef]
  25. Trusenkova, O.O.; Lobanov, V.B.; Lazaryuk, A.Y. Currents in the Southwestern Peter the Great Bay, the Sea of Japan, from the stationary Wavescan Buoy Data in 2016. Oceanology 2022, 62, 310–323. [Google Scholar] [CrossRef]
  26. Trusenkova, O.O.; Ostrovskii, A.G.; Lazaryuk, A.Y.; Lobanov, V.B. Evolution of the Thermohaline Stratification in the Northwestern Sea of Japan: Mesoscale Variability and Intra-annual Fluctuations. Oceanology 2021, 61, 319–328. [Google Scholar] [CrossRef]
  27. Yaroshchuk, I.O.; Kosheleva, A.V.; Lazaryuk, A.Y.; Dolgikh, G.I.; Pivovarov, A.A.; Samchenko, A.N.; Gulin, O.E.; Korotchenko, R.A. Estimation of Seawater Hydrophysical Characteristics from Thermistor Strings and CTD Data in the Sea of Japan Shelf Zone. J. Mar. Sci. Eng. 2023, 11, 1204. [Google Scholar] [CrossRef]
  28. Leontyev, A.P.; Yaroshchuk, I.O.; Smirnov, S.S.; Kosheleva, A.V.; Pivovarov, A.A.; Samchenko, A.N.; Shvyrev, A.N. A spatially distributed measuring complex for monitoring hydrophysical processes on the ocean shelf. Instrum. Exp. Tech. 2017, 60, 130–136. [Google Scholar] [CrossRef]
  29. Thomson, R.E.; Emery, W.J. Data Analysis Methods in Physical Oceanography, 3rd ed.; Elsevier Science: New York, NY, USA, 2014; 728p. [Google Scholar] [CrossRef]
  30. Preisendorfer, R.W. Principal Component Analyses in Meteorology and Oceanography; Elsevier: Amsterdam, The Netherlands, 1988; 425p. [Google Scholar]
  31. Torrence, C.; Compo, G.P. A practical guide to wavelet analysis. Bull. Amer. Meteorol. Soc. 1998, 79, 61–78. [Google Scholar] [CrossRef]
  32. Talley, L.D.; Min, D.-H.; Lobanov, V.B.; Luchin, V.A.; Ponomarev, V.I.; Salyuk, A.N.; Shcherbina, A.Y.; Tishchenko, P.Y.; Zhabin, I. Japan/East Sea water masses and their relation to the sea’s circulation. Oceanography 2006, 19, 32–49. [Google Scholar] [CrossRef]
  33. Monthly Hydrometeorological Bulletin. FERHRI. 2022. Available online: http://www.ferhri.ru/images/stories/FERHRI/Bulletins/Bul_2022/10/2022.10_ch1_meteo.pdf (accessed on 1 November 2024). (In Russian).
  34. Kawaguchi, Y.; Wagawa, T.; Igeta, Y. Near-inertial internal waves and multiple-inertial oscillations trapped by negative vorticity anomaly in the central Sea of Japan. Prog. Oceanogr. 2020, 181, 102240. [Google Scholar] [CrossRef]
  35. EarthExplorer. Available online: https://earthexplorer.usgs.gov (accessed on 15 October 2024).
  36. Fedorov, K.N.; Ginsburg, A.I. Mushroom-Like Currents (Vortex Dipoles): One of the Most Widespread Forms of Non-Stationary Coherent Motions in the Ocean. Elsevier Oceanogr. Ser. 1989, 50, 1–14. [Google Scholar] [CrossRef]
  37. Trusenkova, O.O.; Lobanov, V.B.; Ladychenko, S.Y.; Kaplunenko, D.D. Drift of Surface Lagrangian Buoys in the Central Sea of Japan in October–November 2011. Izv. Atmos. Ocean. Phys. 2021, 57, 969–979. [Google Scholar] [CrossRef]
  38. ADMIRALTY TotalTide. Available online: https://www.admiralty.co.uk (accessed on 20 October 2024).
  39. Kowalik, Z.; Luick, J. The Oceanography of Tides; University of Alaska: Fairbanks, USA, 2013; pp. 1–155. [Google Scholar]
  40. Kokoulina, M.V.; Kurkina, O.E.; Talipova, T.G.; Kurkin, A.A.; Pelinovsky, E.N. Average Climatic Characteristics of Internal Waves in the Sea of Japan Based on the WOA18 Atlas. Phys. Oceanogr. 2023, 30, 563–580. [Google Scholar]
  41. Le Boyer, A.; Alford, M.H.; Pinkel, R.; Hennon, T.D.; Yang, Y.J.; Ko, D.; Nash, J. Frequency Shift of Near-Inertial Waves in the South China Sea. J. Phys. Oceanogr. 2020, 50, 1121–1135. [Google Scholar] [CrossRef]
  42. Yang, B.; Hu, P.; Hou, Y. Variation and Episodes of Near-Inertial Internal Waves on the Continental Slope of the Southeastern East China Sea. J. Mar. Sci. Eng. 2021, 9, 916. [Google Scholar] [CrossRef]
Jmse 12 02263 g001
Figure 1. Schematic of St-1–St-4 moorings at the POI test site. Inset: the Sea of Japan, with the test site marked by the red rectangle. Bathymetry (m) is shown.
Figure 1. Schematic of St-1–St-4 moorings at the POI test site. Inset: the Sea of Japan, with the test site marked by the red rectangle. Bathymetry (m) is shown.
Jmse 12 02263 g001
Jmse 12 02263 g002
Figure 2. Time series of (left) original temperature (T) at St-3 for characteristic depths and (right) T band-pass filtered in the 8 to 20 h range at St-1–St-4 for the depths of the EOF1T maximum (see Table 1); units are °C.
Figure 2. Time series of (left) original temperature (T) at St-3 for characteristic depths and (right) T band-pass filtered in the 8 to 20 h range at St-1–St-4 for the depths of the EOF1T maximum (see Table 1); units are °C.
Jmse 12 02263 g002
Jmse 12 02263 g003
Figure 3. Mean vertical profiles of temperature (T, °C), temperature gradient (G, °C/m) and correlations (R) between PC1T and T and PC1G and G. PC1T and PC1G stand for the principal components of Mode 1 for T and G, respectively. Root mean square spread for mean T and 95% confidence levels for R are shown with dashed lines.
Figure 3. Mean vertical profiles of temperature (T, °C), temperature gradient (G, °C/m) and correlations (R) between PC1T and T and PC1G and G. PC1T and PC1G stand for the principal components of Mode 1 for T and G, respectively. Root mean square spread for mean T and 95% confidence levels for R are shown with dashed lines.
Jmse 12 02263 g003
Jmse 12 02263 g004
Figure 4. Time series of PC1T (°C) and PC1G (°C/m) for St-1–St4; PC1T and PC1G are shown with yellow and blue colors and refer to the left-hand and right-hand y-axes, respectively.
Figure 4. Time series of PC1T (°C) and PC1G (°C/m) for St-1–St4; PC1T and PC1G are shown with yellow and blue colors and refer to the left-hand and right-hand y-axes, respectively.
Jmse 12 02263 g004
Jmse 12 02263 g005
Figure 5. Joint WT-spectra ((°C)2/m) of PC1T and PC1G for St-1–St-4. Cones of influence of edge effects (COI) are shown with dashed lines, 90% confidence levels are shown with yellow contours. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Figure 5. Joint WT-spectra ((°C)2/m) of PC1T and PC1G for St-1–St-4. Cones of influence of edge effects (COI) are shown with dashed lines, 90% confidence levels are shown with yellow contours. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Jmse 12 02263 g005
Jmse 12 02263 g006
Figure 6. Time-averaged joint spectra (left) between PC1T and PC1G ((°C)2/m) for St-1–St-4 and (right) of PC1T (°C)2) between pairs of the adjacent moorings.
Figure 6. Time-averaged joint spectra (left) between PC1T and PC1G ((°C)2/m) for St-1–St-4 and (right) of PC1T (°C)2) between pairs of the adjacent moorings.
Jmse 12 02263 g006
Jmse 12 02263 g007
Figure 7. Periods of maximum power in the interval between 6 and 20 h for the joint spectra of (left) PC1T and PC1G for St-1–St-4 and (right) for PC1T between the pairs of the adjacent moorings. The periods are shown out of COI and for times when the power exceeded the red noise.
Figure 7. Periods of maximum power in the interval between 6 and 20 h for the joint spectra of (left) PC1T and PC1G for St-1–St-4 and (right) for PC1T between the pairs of the adjacent moorings. The periods are shown out of COI and for times when the power exceeded the red noise.
Jmse 12 02263 g007
Jmse 12 02263 g008
Figure 8. Thermal (infrared) image acquired from Landsat-9 on 5 October 2022 at 01:57 UTC [35]; warm and cold waters are shown with dark and light shades, respectively. The stipe of the mushroom-like structure is shown with the white arrow.
Figure 8. Thermal (infrared) image acquired from Landsat-9 on 5 October 2022 at 01:57 UTC [35]; warm and cold waters are shown with dark and light shades, respectively. The stipe of the mushroom-like structure is shown with the white arrow.
Jmse 12 02263 g008
Jmse 12 02263 g009
Figure 9. (Top left) Normalized sea level anomalies (SLA, dimensionless) at St-2 (red curve) and from ADMIRALTY TotalTide (blue curve). (Bottom left) Its wavelet spectrum and (right) time-averaged wavelet spectrum. Cones of influence are shown with dashed lines; 90% confidence level is shown with yellow contour. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Figure 9. (Top left) Normalized sea level anomalies (SLA, dimensionless) at St-2 (red curve) and from ADMIRALTY TotalTide (blue curve). (Bottom left) Its wavelet spectrum and (right) time-averaged wavelet spectrum. Cones of influence are shown with dashed lines; 90% confidence level is shown with yellow contour. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Jmse 12 02263 g009
Jmse 12 02263 g010
Figure 10. Joint WT-spectra ((°C)2) of PC1T between the pairs of the adjacent moorings. Cones of influence are shown with dashed lines; 90% confidence levels are shown with yellow contours. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Figure 10. Joint WT-spectra ((°C)2) of PC1T between the pairs of the adjacent moorings. Cones of influence are shown with dashed lines; 90% confidence levels are shown with yellow contours. Validated are the areas out of COI and where the power exceeds the 90% confidence level.
Jmse 12 02263 g010
Table 1. Data characteristics.
Table 1. Data characteristics.
ParameterSt-1St-2St-3St-4
Start of the record10/3 19:0010/3 19:0010/3 19:0010/4 12:17
End of the record10/9 5:5810/14 12:1510/14 12:1510/14 12:15
Depth, m4.5–39.55.5–39.57–41.512.5–47
Depth step, m111.51.5
M for T36352424
Time step, min1111
N7859 15,43615,43614,399
N* for T7–31 31–929–1828–6
λ1T (%)51.948.451.565.8
Dm for EOF1T, m25.527.029.530.5
N* for PC1T15141615
Ta, °C7.27.29.48.0
N* for G58–1156–1857–2528–6
λ1G(%)28.824.127.725.5
Dm for EOF1G, m18.0/31.020.5/33.519.75/33.2520.75/35.5
N* for PC1G11162720
Ga, °C 0.14/0.140.40 0.380.44/0.460.17/0.16
Pm, h9.6, 17.9–18.39.6–9.8, 16.7–17.1, 48.3–49.09.8, 16.7–17.1, 48.3–49.515.9, 48.3, 48.3–49.5
Designations: M is number of levels in depth; N is number of time counts; N* is number of degrees of freedom for different levels; T and G are low-pass-filtered temperature and its gradient; λ1T and λ1G are eigenvalues expressed as percentage of the total variance; EOF1T and EOF1G are empirical orthogonal functions of Mode 1; PC1T and PC1G are principal components of Mode 1; Ta and Ga are mean anomalies at Dm, for T and G, respectively; Dm is the depth where EOF has its maximum value; Pm is the period of maximum power of time-averaged joint spectrum between T and G. Ga and Dm for G are shown for the upper/lower maximums. Time is UTC+10 (VLAT).
Table 2. Time-averaged characteristics of the joint T spectra.
Table 2. Time-averaged characteristics of the joint T spectra.
ParameterSt-1–St-2St-2–St-3St-3–St-4
L, m52613001971
Pm, h9.8–10.0;
18.7–19.2		16.3–16.7;
49.5–50.6		15.9; 48.3
Usd, m/s1.111.060.95
UNI, m/s0.440.410.55
U2d, m/s–0.151.17
Designations: L is the distance between the moorings; Pm is the period; Usd, UNI, and U2d stand for propagation speed of temperature anomalies on the SD, NI, and two-day timescales, respectively.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Trusenkova, O.; Yaroshchuk, I.; Kosheleva, A.; Samchenko, A.; Pivovarov, A.; Dubina, V. Near-Inertial Oscillations of Thermocline in the Shelf Area off Vladivostok, the Sea of Japan, from a Set of Thermostrings. J. Mar. Sci. Eng. 2024, 12, 2263. https://doi.org/10.3390/jmse12122263

AMA Style

Trusenkova O, Yaroshchuk I, Kosheleva A, Samchenko A, Pivovarov A, Dubina V. Near-Inertial Oscillations of Thermocline in the Shelf Area off Vladivostok, the Sea of Japan, from a Set of Thermostrings. Journal of Marine Science and Engineering. 2024; 12(12):2263. https://doi.org/10.3390/jmse12122263

Chicago/Turabian Style

Trusenkova, Olga, Igor Yaroshchuk, Alexandra Kosheleva, Aleksandr Samchenko, Alexander Pivovarov, and Vyacheslav Dubina. 2024. "Near-Inertial Oscillations of Thermocline in the Shelf Area off Vladivostok, the Sea of Japan, from a Set of Thermostrings" Journal of Marine Science and Engineering 12, no. 12: 2263. https://doi.org/10.3390/jmse12122263

APA Style

Trusenkova, O., Yaroshchuk, I., Kosheleva, A., Samchenko, A., Pivovarov, A., & Dubina, V. (2024). Near-Inertial Oscillations of Thermocline in the Shelf Area off Vladivostok, the Sea of Japan, from a Set of Thermostrings. Journal of Marine Science and Engineering, 12(12), 2263. https://doi.org/10.3390/jmse12122263

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop