## 1. Introduction

The East Sea (Sea of Japan) is a semienclosed marginal sea connected to the Pacific Ocean through four straits shallower than 200 m, as shown in Fig. 1. The horizontal length scale of the East Sea is about 1000 km and its mean depth is about 1700 m with a maximum depth of 4000 m. The Tsushima Current, which flows through the Korea Strait (also known as the Korea/ Tsushima Strait), is a major contributor to the circulation of the East Sea and to the transport of heat and salt from the Pacific Ocean to the East Sea (Moriyasu 1972). Numerical experiments show that the variability of the circulation in the East Sea depends largely on the transport variation through the Korea Strait (Holloway et al. 1995; Kim and Yoon 1999). Therefore, continuous monitoring of the transport in the Korea Strait is essential to understand the circulation of the East Sea.

Estimates of the transport through the Korea Strait vary widely depending upon methods. Ranges for the annual mean are 0.5 ∼ 4.2 × 10^{6} m^{3} s^{−1} with a seasonal variation of 0.7 ∼ 4.6 × 10^{6} m^{3} s^{−1} (Table 1). Yi (1966) estimated the transport through the Korea Strait using a dynamical calculation with a reference level at the bottom, which cannot give a good estimate because of the shallow bottom. The transport was also calculated from the sea level difference across the strait (Mizuno et al. 1989) based on an assumption that the current is vertically barotropic. However, it was reported later that the baroclinic effect on the sea level difference is significant in the Korea Strait (Isobe 1994; Lyu and Kim 2003). Miita and Ogawa (1984) used 431 current observations lasting at least one day. Isobe et al. (1994) and Katoh et al. (1996) estimated the transport using the towed ADCP. As the methods provide instantaneous current fields, it is difficult to remove large short-term variations, such as tidal currents. Long-term moorings of current meters are very few because of high fishing activities in the Korea Strait.

Since February 1997 currents in the Korea Strait have been measured by a vessel-mounted ADCP along a track between Pusan, Korea, and Hakata, Japan (Fig. 1), six times per week (Takikawa et al. 1999). The U.S. Naval Research Laboratory (NRL) also deployed 12 trawl- resistant bottom-mounted (TRBM) ADCPs along two lines in the Korea Strait (Fig. 1) for the time period of May 1999 to March 2000 (Perkins et al. 2000). The average spacing between ADCPs is 25 km and their sampling interval is generally 30 min. However, it is difficult to maintain these measurements continuously because of economic considerations. Thus the characteristics of temporal variations of the transport in the Korea Strait have been poorly understood until now.

According to Faraday's law, when seawater possessing electric conductivity flows through a strait under the geomagnetic field, an electric potential difference is induced that is related to the volume transport through the strait (Sanford and Flick 1975). Longuet-Higgins (1949) originally introduced this possibility and Bowden (1956) applied it to the Dover Strait. Larsen (1992) and Baringer and Larsen (2001) have successfully monitored the volume transport through the Florida Strait using this technique.

The same technique was applied to the Korea Strait using an in-service submarine telephone cable between Pusan, Korea, and Hamada, Japan (Fig. 1). Kawatate et al. (1991) found the energy peaks at tidal frequencies and high coherencies between the voltage and current meter data for the period from 1987 to 1996. Choi et al. (1992) and Choi et al. (1997) showed also high spectral energies at low and tidal frequencies, and estimated the conversion factor for voltage to transport as 30 ∼ 35 × 10^{6} m^{3} s^{−1} V^{−1} by comparing the amplitude of the *M*_{2} signal in the measured voltage at Pusan with that of the *M*_{2} tidal transport from a tidal model. Recently Lyu et al. (2002b) analyzed the cable voltage measured simultaneously at Hamada, Japan, and Pusan, Korea, using the in-service telephone cable in 1990. Comparing the voltage difference in 1990 with that measured using the now abandoned cable in 1998, they showed that both datasets have approximately the same amplitude and phase for *M*_{2} and *O*_{1}. They concluded that the relationship between the voltage and the volume transport through the Korea Strait can be considered robust and stable over time.

As a new optical fiber cable began service for communication between Korea and Japan in July 1997, the abandoned telephone cable was given to the Research Institute of Oceanography, Seoul National University, and to the Ocean Research Institute, the University of Tokyo from Korea Telecom and Kokusai Denshin Denwa for a cooperative research program of NorthEast Asian Region Global Ocean Observing System. Since 1 March 1998 the voltage has been measured at the Pusan Submarine Cable Relay Station, which is in fact the potential difference across the Korea Strait as the cable was grounded at Hamada and Pusan using the copper electrodes (Fig. 2).

The voltage measurement system and data are described in section 2. Geomagnetic correction and error estimates of voltage are discussed in section 3 and the conversion of voltage to transport using observed transports is described in section 4. The voltage-derived transports are discussed for different timescales and compared with the observed transport in the Korea Strait in section 5. A summary is provided in section 6.

## 2. Voltage measurement system and data

The voltage measurement system consists of a digital volt meter (Keithley DMM 2000), a personal computer as a data logger, a GPS (Global Positioning System), and a modem (Fig. 2). The computer clock is controlled by the GPS time. The sampling interval is 5 min and the data are automatically transferred once per day to the OCEAN Laboratory at Seoul National University, Seoul, Korea via modem.

Observed voltage is averaged hourly for analyses here for the period from March 1998 to August 2001. The US NRL deployed 12 TRBM ADCPs (Fig. 1) from May 1999 through March 2000 along two lines in the Korea Strait (Perkins et al. 2000). The deployment was divided into two periods: the first half from May to October 1999 and the second half from October 1999 to March 2000. Transports are estimated from these observed current data (Jacobs et al. 2001; Teague et al. 2002). First, eight tidal constituents are removed from each velocity component by harmonic analysis. Each velocity component is low-pass filtered with a cutoff period of 40 h. Transport values are then obtained by integrating the currents normal to the mooring section after optimal spatial interpolation of observed currents at 3-h intervals. Monthly mean transports from April to November 1998, calculated using data from a vessel-mounted ADCP (Takikawa et al. 1999) acquired along a track between Pusan, Korea, and Hakata, Japan (Fig. 1) are also compared with the voltage measurements.

The hourly raw voltage has short-period variations with a range of about 1.5 V and a mean of 0.793 V. Ten-day segments of the time series of the voltage (dark line in Fig. 3a) and the depth-averaged along-strait current (Fig. 3b) measured at N2 located on the northern ADCP mooring line by the US NRL show prominent semidiurnal and diurnal variations in both datasets. A high coherence between datasets is visible with little phase difference. However, there are occasionally abrupt changes in voltage that are not associated with the along- strait current. These changes are not related to transport variations and will be discussed in the next section.

The observed voltage (Fig. 4a) and the along-strait current (Fig. 4b) have prominent peaks in their power spectra at diurnal and semidiurnal tidal frequencies. At low frequencies spectral energy density is high, especially at subinertial and monthly timescales in the voltage. One can notice in the voltage spectra, however, that there are significant peaks at frequencies of 3, 4, and 5 cpd, which are not significant in the spectra of the along- strait current. These variations may be related to the solar diurnal variations of geomagnetic fields at 1, 2, 3, 4, and 5 cpd as reported by Larsen (1992). Since these geomagnetic effects on the observed voltage are significant at timescales shorter than 5 days according to Larsen (1992), they must be removed from the observed voltage to investigate oceanographic tidal and subinertial variations.

## 3. Geomagnetic correction and error estimates of voltage

### a. Geomagnetic correction

The method of Larsen et al. (1996) is applied for the geomagnetic correction. This method involves construction of a transfer function, *Z,* between voltage and geomagnetic fields, and removal of the voltage variations that are coherent with geomagnetic ones. The horizontal geomagnetic data at Kanoya and Memambetsu, Japan (Fig. 1), are used to remove the geomagnetic effect from the observed voltage. The dominant geomagnetic component is located 125° clockwise from the north, that is, about 35° from the cable line. The magnitude of the transfer function, *Z,* increases with frequency and its phase has values from 90° to 180° (Fig. 5). At low frequencies the magnitude is very small, which suggests that the geomagnetic effect is negligible for long periods.

The geomagnetic-induced voltage, *V*_{G}, is derived from the transfer function and the corrected voltage, *V*_{cor}, is obtained by subtracting *V*_{G} from the raw voltage, *V*_{raw} (Fig. 6). Tide-induced voltage, *V*_{T}, is calculated by harmonic analyses of *V*_{cor}. The voltage related to the current, *V*_{res}, becomes available by subtracting *V*_{T} from *V*_{cor}. Most of the abrupt changes in *V*_{raw} are clearly eliminated by this procedure as shown in Fig. 3a, where *V*_{cor} is designated by a light line.

Spectral analysis of *V*_{G} (Fig. 4c) shows that *V*_{G} contains peaks at frequencies of 1, 2, 3, 4, and 5 cpd. Its energy density is very low for periods longer than 2 days. In the spectrum of *V*_{cor} (Fig. 4d) geomagnetic signals are reduced at 3, 4, and 5 cpd. The relatively high spectrum at 3 and 4 cpd seems to be partly related to nonlinear effects of tidal currents in the shallow strait, as peaks also occur in the along-strait current (Fig. 4b).

### b. Error estimates of voltage in relation with the volume transport

_{V}) is given by

_{V}

*t*

_{L}

*t*

_{I}

*t*

_{L}is the local motion-induced voltage near the cable and ΔΦ

_{I}is caused by the horizontal electric current (Sanford 1971). According to Sanford (1971) the local motion-induced voltage is

*σ*is the electrical conductivity of the ocean,

*τ*′ the conductance of the sediment and conducting crust under the strait,

*H*the water depth,

*L*the width of the strait,

*F*

_{Z}the vertical geomagnetic component,

*υ*the downstream velocity. The local motion-induced voltage, ΔΦ

_{L}, is proportional to the conductivity-weighted, vertically averaged velocity. Therefore, ΔΦ

_{L}may vary even with a fixed volume transport if there are changes in the water conductivity and the spatial structure of the flow.

_{L}from the linear relation between ΔΦ

_{L}and the volume transport by splitting

*σ*and

*υ*in Eq. (2) into two parts, average and deviation, as follows:

_{L}can be written as

*T*the volume transport through the strait, and ɛ the deviation from the linear relation between ΔΦ

_{L}and transport. Thus, ΔΦ

_{L}is decomposed into two parts in Eq. (5c); the first term is linearly related to the volume transport and the second term is the deviation that is caused by change of the flow structure over the varying bottom depth and water conductivity.

It is possible to estimate this deviation, ɛ, by using current data measured along the northern ADCP line in the Korea Strait by the U.S. NRL (Jacobs et al. 2001) and water conductivity data calculated from hydrographic data of line 208 of the Korean Oceanographic Data Center (KODC) in 1998, which is nearly parallel to the northern ADCP line (Fig. 1). The current data are low-pass filtered with a half-power period of 36 h and subsampled every 3 h. Figure 7 shows the time- averaged structure of the along-strait current in the northern section for the second half of the ADCP records. Two main northeastward flows appear in the western and eastern side of the Korea Strait, while the flow is very weak and variable at the center due to the Tsushima Island-induced wake (Perkins et al. 1999). Hydrographic data are measured at standard depths every two months. The magnitude of ɛ is then estimated as 0.011 V from these current and conductivity data.

_{V}) can be changed from the local-induced voltage (ΔΦ

_{L}) by ΔΦ

_{I}, which is caused by the electric current through the horizontal section when there are downstream changes of velocity and conductivity (Sanford 1971). If the characteristic downstream length scale of the flow is

*Y,*the horizontal electric current effect can be estimated as follows (Larsen 1992):

*τ** is the conductance of the land. The width L of the Korea Strait, the water conductivity

*H*

_{0}are about 200 km, 4 S m

^{−1}, and 100 m, respectively. Here

*τ*

^{′}

_{0}

*τ** are assumed to be 1000 S and 400 S using the values reported by Utada et al. (1986), and

*κ*can be estimated for various characteristic downstream length scales (Table 2). The cross-stream voltage (ΔΦ

_{V}) generated by a fixed transport will change if

*Y*varies. If the length scale

*Y*becomes larger, the horizontal electric current effect becomes smaller from Eq. (6) and

*κ*changes less with

*Y.*If downstream flows are assumed to have characteristic length scales between 200 and 600 km along the Korea Strait,

*κ*ranges from 0.11 to 0.27 and the standard error in

*κ*is estimated to be 0.14. If the transport through the Korea Strait is 3.0 × 10

^{6}m

^{3}s

^{−1}and the conversion factor from voltage to transport is 8 × 10

^{6}m

^{3}s

^{−1}V

^{−1}, this error corresponds to a voltage error of 0.053 V. Since the horizontal electric current effect (ΔΦ

_{I}) increases with the width of the strait as described in Eq. (6), the error of ΔΦ

_{I}is much larger than that of ΔΦ

_{L}in the Korea Strait with the width of 200 km. This result is different from that in the Florida Strait with the width of 60 km in Larsen (1992), where the horizontal electric current effect is not dominant. The total rms error of voltage in estimating transport in the Korea Strait comes to 0.054 V.

## 4. Conversion from voltage to transport

### a. Conversion at subtidal frequencies

The residual voltage, *V*_{res}, is compared with the observed transport from current measurements by the U.S. NRL (Jacobs et al. 2001) after low-pass filtering with a half-power period of 36 h and subsampling every 3 h. The cross spectra between transport and voltage are shown in Fig. 8. Dark and light lines denote the cross spectra of the transport of the southern section (*T*_{S}) and the northern section (*T*_{N}) with voltage.

Over the first half of the ADCP transport observation (spring and summer), *T*_{S} is highly correlated with voltage at all periods (Fig. 8a), while coherency between *T*_{N} and voltage is lower by 0.2. It is suspected that the lower coherency results from the missing current data at N1 station (Fig. 1) in the first half of the deployment period. The transport for the northern section could not resolve adequately the current intensification by the East Korea Warm Current (EKWC) close to the Korean coast (Isobe et al. 1994). Additionally, there may also be more surface intensification of current in the northern section than in the southern section, which could not be measured by the bottom-mounted ADCP. This effect is more pronounced during the first half of the record, since currents are more baroclinic in spring and summer than in fall and winter (Cho and Kim 1998; Lyu and Kim 2003). Therefore *T*_{N} may be underestimated in the first half of the deployment period and this underestimation is clearly shown from May to July in 1999 (Jacobs et al. 2001). Phase differences are near 0° in both cases with the variation range less than 10° (Fig. 8b). This implies that there is little time lag between transport and voltage and that the current is almost horizontally nondivergent in the Korea Strait at timescales longer than 2 days.

In the second half of the ADCP records (fall and winter) coherencies for transports from both ADCP sections with voltage are high at all periods (Fig. 8c), reflecting a weak baroclinicity and the complete data return from the northern ADCP section for this period. Phase differences are also near 0° in both cases with the variation range less than 10° (Fig. 8d). These results imply that voltage is closely related to the real transport through the Korea Strait. Here *T*_{S} is used for calculating the relationship between transport and voltage since *T*_{N} data coverage is incomplete near Korea.

The maximum likelihood method (Macdonald and Thompson 1992) is used to obtain the linear relation between *T*_{S} and voltage. To satisfy the basic assumption in linear regression that each sample is not correlated, both time series are subsampled every two days, which corresponds to the critical time determined from autocorrelations. Standard errors of each coefficient of the linear relation are estimated on the assumption that both datasets have known errors. The transport standard error was estimated to be 0.5 × 10^{6} m^{3} s^{−1} by Jacobs et al. (2001). The voltage standard error is estimated to be 0.054 Volt in section 3.

For the above errors, the conversion factor from voltage to the transport is given as Λ_{0} = (8.06 ± 0.63) × 10^{6} m^{3} s^{−1} V^{−1} and the voltage bias corresponding to zero net volume transport in the vertical section including the cable ground points is *V*_{0} = 0.48 ± 0.07 V (Fig. 9). The correlation coefficient is 0.85 and the standard error in Λ_{0} is ±7.8%, which amounts to a voltage- derived transport error of ±0.23 × 10^{6} m^{3} s^{−1} for a transport of 3 × 10^{6} m^{3} s^{−1}. Considering the error in the estimated voltage bias, the total standard error becomes ±0.6 × 10^{6} m^{3} s^{−1} for a transport of 3 × 10^{6} m^{3} s^{−1}.

As the cable is grounded at Pusan and Hamada using the same electrode of copper, the bias voltage *V*_{0} of 0.48 V in the observed voltage across the Korea Strait may be caused by the differences in the temperature, salinity, and electrochemical state of the ground sites or DC current in the vicinity of the cable (Larsen 1992). However, it is still not clear what causes this bias voltage and how large its temporal changes may be. Since a good correlation between this voltage-derived transport and the sea level difference (SLD) across the Korea Strait was reported by Lyu and Kim (2003), this SLD may be used as an independent data to continuously inspect steadiness of the conversion factor from cable voltage to transport and any changes of the bias voltage in order to investigate long-term transport variations from cable voltage.

Takikawa et al. (1999) has measured the velocity field between Pusan, Korea, and Hakata, Japan, using a vessel-mounted ADCP since February 1997 (Fig. 1). A linear relationship is also calculated using monthly mean transport from these measurements and voltage from April to November in 1998 (Fig. 10), after removing the mean transport for August 1998 that deviates two times more than the rms deviation from the regression line. The correlation coefficient is as high as 0.98 with a bias of *V*_{1} = 0.45 V and conversion factor of Λ_{1} = 8.0 × 10^{6} m^{3} s^{−1} V^{−1}, which are within the error bounds of *V*_{0} and Λ_{0} estimated above, in spite of the difference in measurement methods of transport and sampling intervals. This consistency indicates clearly that the linear relation between voltage and transport is reliable at these subtidal frequencies.

### b. Conversion at tidal frequencies

Tidal transport in the Korea Strait has been estimated from tide models by Kang et al. (1991) and Choi et al. (1994) and from limited current meter data by Odamaki (1989). They reported that the amplitude of *M*_{2} tidal transport is about 5.0 ∼ 5.6 × 10^{6} m^{3} s^{−1} across the northeastern part of the Korea Strait, which is larger than the known mean transport (2.6 ∼ 2.7 × 10^{6} m^{3} s^{−1}) in the Korea Strait (Takikawa et al. 1999; Teague et al. 2002).

Recently tidal currents have been assimilated in a barotropic model for the Korea Strait using coastal sea level data, TOPEX/Poseidon altimetry data and current data (Book et al. 2001). Amplitudes of tidal transports across the cable section are calculated as 4.45, 1.94, 2.26, and 1.78 × 10^{6} m^{3} s^{−1} for *M*_{2}, *S*_{2}, *K*_{1}, and *O*_{1}, respectively, from this assimilation (Table 3). Conversion factors derived by comparing tidal amplitudes of model-derived transport and voltage are 11.8, 11.8, 16.4, and 14.8 × 10^{6} m^{3} s^{−1} V^{−1} for *M*_{2}, *S*_{2}, *K*_{1}, and *O*_{1}, respectively, as shown in Table 3.

The frequency of *K*_{1} is close to 1 cpd, where the geomagnetic effect is dominant and this effect cannot readily be removed if the geomagnetic data are not measured near the cable ground site (Larsen 1992). Since the geomagnetic data used in this study are measured in the region farther than 400 km from the cable ground site (Fig. 1), geomagnetic effects may remain in the voltage-derived transport at the frequency of *K*_{1}. However derived conversion factors from other tidal constituents are still larger than Λ_{0} derived from subtidal transport variations (Table 3).

The difference in conversion factors can be explained by the effect of horizontal electric current on the observed voltage in Eq. (6). The characteristic length scale of tidal currents in the Korea Strait can be estimated to be about 200 km (Book et al. 2001; Odamaki 1989). In this case the cross-stream electric potential difference, ΔΦ_{V} may be reduced by about 27% at most from the local motion-induced potential difference, ΔΦ_{L} (Table 2). Since the amplitude of ΔΦ_{V} is 0.376 V for *M*_{2}, ΔΦ_{L} estimated from Eqs. (1) and (6) is 0.52 V. Comparing this ΔΦ_{L} and the tidal transport for *M*_{2}, the conversion factor changes to 8.6 × 10^{6} m^{3} s^{−1} V^{−1}. For *O*_{1}, ΔΦ_{V} is 0.120 V and ΔΦ_{L} is estimated to be 0.17 V, which gives the conversion factor of 10 × 10^{6} m^{3} s^{−1} V^{−1}.

The difference in conversion factors for *M*_{2} and *O*_{1} could be explained by considering three factors. First, the amplitude and the phase of *O*_{1} signal in cable voltage have relatively larger variances than those of *M*_{2} signal (Lyu et al. 2002b), that is, the former is less stable than the latter. Second, the amphidromic point of *O*_{1} tide is reproduced near Pusan from the tidal model (Book et al. 2001). This implies that *O*_{1} tidal currents are more complex than *M*_{2} currents near the cable section. Third, tidal flow in the deep ocean will have length scales as large as 1000 km and this will give rise to horizontal electric currents that will combine with the more locally induced horizontal electric currents, especially in the direction of the strait. This will further confuse the tidal interpretation since the cable is not perpendicular to the coastline in the case of the Korea Strait (J. C. Larsen 2001, personal communication).

The horizontal electric current effects must be considered to analyze tidal transports from the observed voltage across the Korea Strait. However, these effects are less than about 10% of the local motion-induced voltage for subtidal downstream currents, assuming characteristic length scales longer than 500 km along the strait.

## 5. Transport variations in the Korea Strait

The voltage-derived transport, *T*_{V} (dark line in Fig. 11), from the conversion factor Λ_{0} and the bias *V*_{0} estimated in section 4 and observed transport *T*_{S} (light line in Fig. 11) from the bottom-mounted ADCP (Teague et al. 2002) have similar mean values and standard deviations: (2.69 ± 0.83) × 10^{6} m^{3} s^{−1} and (2.64 ± 0.82) × 10^{6} m^{3} s^{−1} for *T*_{V} and *T*_{S}, respectively. Their correlation coefficient is as high as 0.85. The differences in their time series are mostly less than the estimated standard error 0.6 × 10^{6} m^{3} s^{−1} in voltage-derived transport.

There are transport variations as large as 2 ∼ 3 (× 10^{6} m^{3} s^{−1}) on a timescale of 3 ∼ 7 days and about a month in *T*_{V} as shown in Fig. 11. It should be pointed out that these subinertial variations are larger than the known seasonal variations in the Korea Strait (Takikawa et al. 1999; Teague et al. 2002). These transport fluctuations in the Korea Strait have not been known previously. Although there are some differences in amplitude and phase between *T*_{S} and *T*_{V}, volume transport variations clearly occur with timescales of 3 ∼ 7 days and variation ranges of 2 ∼ 3 (× 10^{6} m^{3} s^{−1}) in the Korea Strait. Lyu et al. (2002a) reported that these variations are related to nonisostatic responses of the East Sea to the atmospheric pressure forcing. These variations as well as tidal variations must be separated from the instantaneous transport observations to investigate long-term variations in the Korea Strait.

To investigate seasonal variations, a low-pass filter with a half-power period of 90 days is applied to *T*_{V}. The result shown in Fig. 12 has a mean of 2.5 × 10^{6} m^{3} s^{−1} with a maximum of 3.4 × 10^{6} m^{3} s^{−1} in October 1999 and a minimum of 1.6 × 10^{6} m^{3} s^{−1} in January 2000. The monthly mean transports from the vessel- mounted ADCP (Takikawa et al. 1999) and the bottom- mounted ADCP (Teague et al. 2002), indicated in Fig. 12, agree well with *T*_{V} within a 95% confidence interval. There seems to be a tendency that the transport becomes large in summer and autumn and small in winter and spring. However, interannual variations are prominent. The seasonal variation is very weak from spring 2000 to spring 2001 and annual mean transport reduces by 0.4 × 10^{6} m^{3} s^{−1} from 1999 to 2000.

## 6. Summary

Cable voltage measured from the abandoned submarine telephone cable between Pusan, Korea, and Hamada, Japan, since March 1998 is used to monitor the volume transport through the Korea Strait. The geomagnetic-induced voltage, which is substantially large at periods shorter than two days, is removed from the measured voltage by the method of Larsen et al. (1996). The observed voltage in the Korea Strait is found to be contaminated by the horizontal electric current due to a relatively large width of the strait. This horizontal electric current effect can cause time/length scale dependence of the relationship between voltage and transport in the Korea Strait. It is found that there is a very good linear relationship between voltage and the observed transports from the bottom-mounted ADCP (Teague et al. 2002) and the vessel-mounted ADCP (Takikawa et al. 1999). The conversion factor from voltage to transport is estimated to be Λ_{0} = (8.06 ± 0.63) × 10^{6} m^{3} s^{−1} V^{−1} with the bias voltage of *V*_{0} = 0.48 ± 0.07 V.

In the voltage-derived transport, tidal variations are dominant with amplitudes larger than the mean transport known in the Korea Strait. However, the horizontal electric current effects must be considered to estimate tidal transports from the observed voltage across the Korea Strait because the characteristic downstream length scale of tidal currents is comparable to the width of the strait. It is fortunate, however, that these effects are small for subtidal downstream currents, which have long characteristic length scales.

There are transport variations as large as 2 ∼ 3 (× 10^{6} m^{3} s^{−1}) on timescales of 3 ∼ 7 days. There are also large variations in the transport with a range of 2 × 10^{6} m^{3} s^{−1} on timescales of about a month. Three-year data show a relatively larger transport in summer and autumn than in winter and spring but there is a large interannual variation. The voltage measured between Pusan and Hamada provides a good means to monitor the transport through the Korea Strait continuously and enables us to investigate various temporal transport variations and their causes, which affect the circulation and the hydrography in the East Sea decisively.

## Acknowledgments

This work was supported by OCEAN Laboratory and the BK21 project of the Korean Government in 1999–2003. H. T. Perkins, W. J. Teague, and J. W. Book were supported by the U.S. Office of Naval Research as part of the Basic Research Project “Linkages of Asian Marginal Seas” and “Japan East Sea DRI” under Program Element 0601153N.

## REFERENCES

Baringer, M., and Larsen J. C. , 2001: Sixteen years of Florida current transport at 27°N.

,*Geophys. Res. Lett***28****,**3179–3182.Book, J. W., and Coauthors, 2001: Tides in the Korea/Tsushima Strait: Observation and model predictions.

*Proc. 11th PAMS/JECSS Workshop,*Cheju, Korea, Korean Science and Engineering Foundation and the Korean Foundation of Science and Techonology Societies, 207–210.Bowden, K. F., 1956: The flow of water through the Strait of Dover related to wind and differences in sea level.

, A248. 517–551.*Philos. Trans. Roy. Soc. London*Byun, S. K., and Seung Y. H. , 1984: Description of current structure and coastal upwelling in the southwest Japan Sea—Summer 1981 and spring 1982.

*Ocean Hydrodynamics of the Japan and East China Seas,*T. Ichiye, Ed., Elsevier, 83–93.Cho, Y. K., and Kim K. , 1998: Structure of the Korea Strait Bottom Cold Water and its seasonal variation in 1991.

,*Cont. Shelf Res***18****,**791–804.Choi, B-H., Kim K. , Kim Y-G. , Bahk K. S. , Choi J. O. , and Kawatate K. , 1992: Submarine cable measurements between Pusan and Hamada.

,*La Mer***30****,**157–167.Choi, B-H., Bang I. K. , and Kim K. H. , 1994: Vertical distribution of tidal current in the Korea Strait.

,*J. Korean Soc. Coastal Ocean Eng***6****,**421–438.Choi, B-H., Kim K. , Kim Y-G. , and Kawatate K. , 1997: Submarine cable voltage measurement between Pusan and Hamada for the years 1987–1996.

,*Acta Oceanogr. Taiwan***36****,**33–46.Holloway, G., Sou T. , and Eby M. , 1995: Dynamics of circulation of the Japan Sea.

,*J. Mar. Res***53****,**539–569.Isobe, A., 1994: Seasonal variability of the barotropic and baroclinic motion in the Tsushima–Korea Strait.

,*J. Oceanogr***50****,**223–238.Isobe, A., Tawara S. , Kaneko A. , and Kawano M. , 1994: Seasonal variability in the Tsushima Warm Current, Tsushima–Korea Strait.

,*Cont. Shelf Res***14****,**23–35.Jacobs, G. A., Perkins H. T. , Teague W. J. , and Hogan P. J. , 2001: Summer transport through the Korea–Tsushima Strait.

,*J. Geophys. Res***106****,**6917–6929.Kang, S. K., Lee S-R. , and Yum K-D. , 1991: Tidal computation of the East China Sea, the Yellow Sea and the East Sea.

*Oceanography of Asian Marginal Seas,*K. Takano, Ed., Elsevier Oceanography Series, Vol. 54, Elsevier, 24–48.Katoh, O., Teshima K. , Kubota K. , and Tsukiyama K. , 1996: Downstream transition of the Tsushima Current west of Kyushu in summer.

,*J. Oceanogr***52****,**93–108.Kawatate, K., and Coauthors, 1991: A cross-spectral analysis of small voltage variation in a submarine cable between Hamada and Pusan with speed variation of the Tsushima Warm Current.

*Oceanography of Asian Marginal Seas,*K. Takano, Ed., Elsevier Oceanography Series, Vol. 54, Elsevier, 207–222.Kim, C-H., and Yoon J-H. , 1999: A numerical modeling of the upper and the intermediate layer circulation in the East Sea.

,*J. Oceanogr***55****,**327–345.Larsen, J. C., 1992: Transport and heat flux of the Florida Current at 27°N derived from cross-stream voltages and profiling data: Theory and observations.

, A338. 169–236.*Philos. Trans. Roy. Soc. London*Larsen, J. C., Mackie R. L. , Manzella A. , Fiordelisi A. , and Rieven S. , 1996: Robust smooth magnetotelluric transfer functions.

,*Geophys. J. Int***124****,**801–819.Longuet-Higgins, M. S., 1949: The electrical and magnetic effect of tidal streams.

,*Mon. Not. Roy. Astron. Soc***16****,**285–307.Lyu, S. J., and Kim K. , 2003: Absolute transport from the sea level difference across the Korea Strait.

*Geophys. Res. Lett.,***30**, 1285, doi:10.1029/2002GL016233.Lyu, S. J., Kim K. , and Perkins H. T. , 2002a: Atmospheric pressure-forced subinertial variations in the transport through the Korea Strait.

*Geophys. Res. Lett.,***29,**1294, doi:10.1029/2001GL014366.Lyu, S. J., Kim Y-G. , Kim K. , Book J. W. , and Choi B. H. , 2002b: Tidal variations in the cable voltage across the Korea Strait.

,*J. Korean Soc. Oceanogr***37****,**1–9.Macdonald, J. R., and Thompson W. J. , 1992: Least-squares fitting when both variables contain errors: Pitfalls and possibilities.

,*Amer. J. Phys***60****,**66–73.Miita, T., and Ogawa Y. , 1984: Tsushima currents measured with current meters and drifters.

*Ocean Hydrodynamics of the Japan and East China Seas,*T. Ichiye, Ed., Elsevier Oceanography Series, Vol. 39, Elsevier, 67–76.Mizuno, S., Kawatate K. , Nagahama T. , and Miita T. , 1989: Measurements of east Tsushima current in winter and estimation of its seasonal variability.

,*J. Oceanogr. Soc. Japan***45****,**375–384.Moriyasu, S., 1972: The Tsushima Current.

*Kuroshio,*H. Stommel and K. Yoshida, Eds., University of Tokyo Press, 353–369.Odamaki, M., 1989: Tides and tidal currents in the Tsushima Strait.

,*J. Oceanogr. Soc. Japan***45****,**65–82.Perkins, H. T., Teague W. J. , Jacobs G. A. , Chang K-I. , and Suk M-S. , 2000: Currents in Korea–Tsushima Strait during summer 1999.

,*Geophys. Res. Lett***27****,**3033–3036.Sanford, T. B., 1971: Motionally induced electric and magnetic fields in the sea.

,*J. Geophys. Res***76****,**3476–3492.Sanford, T. B., and Flick R. E. , 1975: On the relationship between transport and motional electric potentials in broad, shallow currents.

,*J. Mar. Res***33****,**123–139.Takikawa, T., Yoon J-H. , Hase H. , and Cho K-D. , 1999: Monitoring of the Tsushima Current at the Tsushima/Korea Straits.

*Proc. 3rd CREAMS Int. Symp.,*Fukuoka, Japan, Japanese Ministry of Education, Science, Sports and Culture, 15–18.Tawara, S., Miita T. , and Fujiwara T. , 1984: The hydrography and variability in the Tsushima Straits (in Japanese).

,*Bull. Coastal Oceanogr***22****,**50–58.Teague, W. J., Jacobs G. A. , Perkins H. T. , Book J. W. , Chang K-I. , and Suk M-S. , 2002: Low-frequency current observations in the Korea/Tsushima Strait.

,*J. Phys. Oceanogr***32****,**1621–1641.Uda, M., 1934: The results of simultaneous oceanographical investigations in the Japan Sea and its adjacent waters in May and June, 1932 (in Japanese).

,*J. Imp. Fish. Exp. Sta***5****,**57–190.Utada, H., Hamano Y. , and Yukutake T. , 1986: A two-dimensional conductivity model across central Japan.

,*J. Geomagn. Geoelectr***38****,**447–473.Yi, S-U., 1966: Seasonal and secular variations of the water volume transport across the Korea Strait.

,*J. Oceanol. Soc. Korea***1****,**7–13.

Cable voltage measurement system at the Pusan Submarine Cable Relay Station and the vertical section across the cable ground sites, Pusan and Hamada

Citation: Journal of Atmospheric and Oceanic Technology 21, 4; 10.1175/1520-0426(2004)021<0671:MVTTMO>2.0.CO;2

Cable voltage measurement system at the Pusan Submarine Cable Relay Station and the vertical section across the cable ground sites, Pusan and Hamada

Citation: Journal of Atmospheric and Oceanic Technology 21, 4; 10.1175/1520-0426(2004)021<0671:MVTTMO>2.0.CO;2

Cable voltage measurement system at the Pusan Submarine Cable Relay Station and the vertical section across the cable ground sites, Pusan and Hamada

Citation: Journal of Atmospheric and Oceanic Technology 21, 4; 10.1175/1520-0426(2004)021<0671:MVTTMO>2.0.CO;2

Ten-day time series of (a) raw voltage (dark line) and voltage corrected for geomagnetic induced voltage (light line) and (b) depth-averaged, along-strait current measured at N2 station of the northern ADCP mooring line (Fig. 1). Here and elsewhere time is in days with day 0 being 1 Jan 1998 in UTC

Ten-day time series of (a) raw voltage (dark line) and voltage corrected for geomagnetic induced voltage (light line) and (b) depth-averaged, along-strait current measured at N2 station of the northern ADCP mooring line (Fig. 1). Here and elsewhere time is in days with day 0 being 1 Jan 1998 in UTC

Ten-day time series of (a) raw voltage (dark line) and voltage corrected for geomagnetic induced voltage (light line) and (b) depth-averaged, along-strait current measured at N2 station of the northern ADCP mooring line (Fig. 1). Here and elsewhere time is in days with day 0 being 1 Jan 1998 in UTC

Power spectrum with 95% confidence interval of (a) raw voltage, *V*_{raw}; (b) depth-averaged, along-strait current measured at N2 station (Fig. 1) for the first (dark line) and second (light line) half of the ADCP records by the US NRL; (c) geomagnetic-induced voltage, *V*_{G}; and (d) corrected voltage, *V*_{cor}.

Power spectrum with 95% confidence interval of (a) raw voltage, *V*_{raw}; (b) depth-averaged, along-strait current measured at N2 station (Fig. 1) for the first (dark line) and second (light line) half of the ADCP records by the US NRL; (c) geomagnetic-induced voltage, *V*_{G}; and (d) corrected voltage, *V*_{cor}.

Power spectrum with 95% confidence interval of (a) raw voltage, *V*_{raw}; (b) depth-averaged, along-strait current measured at N2 station (Fig. 1) for the first (dark line) and second (light line) half of the ADCP records by the US NRL; (c) geomagnetic-induced voltage, *V*_{G}; and (d) corrected voltage, *V*_{cor}.

(a) Magnitude and (b) phase of smoothed (solid line) and band-averaged (dots with vertical bars for 95% confidence limits) transfer function, *Z,* calculated by the geomagnetic correction method of Larsen et al. (1996)

(a) Magnitude and (b) phase of smoothed (solid line) and band-averaged (dots with vertical bars for 95% confidence limits) transfer function, *Z,* calculated by the geomagnetic correction method of Larsen et al. (1996)

(a) Magnitude and (b) phase of smoothed (solid line) and band-averaged (dots with vertical bars for 95% confidence limits) transfer function, *Z,* calculated by the geomagnetic correction method of Larsen et al. (1996)

Time series of (a) raw voltage, *V*_{raw}; (b) geomagnetic-induced voltage, *V*_{G}; (c) corrected voltage, *V*_{cor}; (d) tide-induced voltage, *V*_{T}; and (e) residual voltage, *V*_{res}

Time series of (a) raw voltage, *V*_{raw}; (b) geomagnetic-induced voltage, *V*_{G}; (c) corrected voltage, *V*_{cor}; (d) tide-induced voltage, *V*_{T}; and (e) residual voltage, *V*_{res}

Time series of (a) raw voltage, *V*_{raw}; (b) geomagnetic-induced voltage, *V*_{G}; (c) corrected voltage, *V*_{cor}; (d) tide-induced voltage, *V*_{T}; and (e) residual voltage, *V*_{res}

Time-averaged along-strait current (cm s^{−1}) in the northern vertical section for the second half of the ADCP records by the NRL, where northeastward currents are positive. The *x* axis is the distance in km from the Korean coast and *y*-axis depth in m

Time-averaged along-strait current (cm s^{−1}) in the northern vertical section for the second half of the ADCP records by the NRL, where northeastward currents are positive. The *x* axis is the distance in km from the Korean coast and *y*-axis depth in m

Time-averaged along-strait current (cm s^{−1}) in the northern vertical section for the second half of the ADCP records by the NRL, where northeastward currents are positive. The *x* axis is the distance in km from the Korean coast and *y*-axis depth in m

Cross-spectra between transport and voltage after low-pass filtering with a half-power period of 36 h, where dark solid lines designate the cross-spectra between the transport of the southern section (*T*_{S}) and voltage and light solid lines designate the cross-spectra between the transport of the northern section (*T*_{N}) and voltage. (left) Coherency with 95% significance level (dashed line). (right) Cross- phase spectra with 95% confidence intervals (dark and light dashed lines for *T*_{S} and *T*_{N}, respectively). Upper panels are for the first transport measurement period and lower panels are for the second period

Cross-spectra between transport and voltage after low-pass filtering with a half-power period of 36 h, where dark solid lines designate the cross-spectra between the transport of the southern section (*T*_{S}) and voltage and light solid lines designate the cross-spectra between the transport of the northern section (*T*_{N}) and voltage. (left) Coherency with 95% significance level (dashed line). (right) Cross- phase spectra with 95% confidence intervals (dark and light dashed lines for *T*_{S} and *T*_{N}, respectively). Upper panels are for the first transport measurement period and lower panels are for the second period

Cross-spectra between transport and voltage after low-pass filtering with a half-power period of 36 h, where dark solid lines designate the cross-spectra between the transport of the southern section (*T*_{S}) and voltage and light solid lines designate the cross-spectra between the transport of the northern section (*T*_{N}) and voltage. (left) Coherency with 95% significance level (dashed line). (right) Cross- phase spectra with 95% confidence intervals (dark and light dashed lines for *T*_{S} and *T*_{N}, respectively). Upper panels are for the first transport measurement period and lower panels are for the second period

Linear regression between *T*_{S} and voltage using 144 data after low-pass filtering with a half-power period of 36 h and subsampling every 2 days. The thick solid line is the linear regression line and dashed lines are standard error bounds

Linear regression between *T*_{S} and voltage using 144 data after low-pass filtering with a half-power period of 36 h and subsampling every 2 days. The thick solid line is the linear regression line and dashed lines are standard error bounds

Linear regression between *T*_{S} and voltage using 144 data after low-pass filtering with a half-power period of 36 h and subsampling every 2 days. The thick solid line is the linear regression line and dashed lines are standard error bounds

Linear regression between monthly mean voltage and transport measured by vessel-mounted ADCP between Pusan, Korea, and Hakata, Japan (Takikawa et al. 1999)

Linear regression between monthly mean voltage and transport measured by vessel-mounted ADCP between Pusan, Korea, and Hakata, Japan (Takikawa et al. 1999)

Linear regression between monthly mean voltage and transport measured by vessel-mounted ADCP between Pusan, Korea, and Hakata, Japan (Takikawa et al. 1999)

Time series of voltage-derived transport *T*_{V} (dark line) and *T*_{S} (light line), which is observed by the bottom-mounted ADCP, after low-pass filtering with a half-power period of 36 h.

Time series of voltage-derived transport *T*_{V} (dark line) and *T*_{S} (light line), which is observed by the bottom-mounted ADCP, after low-pass filtering with a half-power period of 36 h.

Time series of voltage-derived transport *T*_{V} (dark line) and *T*_{S} (light line), which is observed by the bottom-mounted ADCP, after low-pass filtering with a half-power period of 36 h.

Seasonal variations of *T*_{V} (solid dark line) after low-pass filtering with a half-power period of 90 days, where dashed lines are 95% confidence intervals. Open and closed circles designate monthly mean transport observed by the vessel-mounted ADCP and the bottom-mounted ADCP, respectively. The solid light line is the mean transport of *T*_{V}

Seasonal variations of *T*_{V} (solid dark line) after low-pass filtering with a half-power period of 90 days, where dashed lines are 95% confidence intervals. Open and closed circles designate monthly mean transport observed by the vessel-mounted ADCP and the bottom-mounted ADCP, respectively. The solid light line is the mean transport of *T*_{V}

Seasonal variations of *T*_{V} (solid dark line) after low-pass filtering with a half-power period of 90 days, where dashed lines are 95% confidence intervals. Open and closed circles designate monthly mean transport observed by the vessel-mounted ADCP and the bottom-mounted ADCP, respectively. The solid light line is the mean transport of *T*_{V}

Previous estimates of the volume transport through the Korea Strait. (10^{6} m^{3} s^{−1} )

The values of κ, which denotes the amount of the hor izontal electric current effect on the local motion-induced voltage ΔΦ_{L} in Eq. (6), for various downstream length scales (*Y* )

Amplitudes of tidal transport variations from the tide model of Book et al. (2001) and tidal voltage variations and con version factors from voltage to transport for four major tidal con stituents

^{*}

OCEAN Laboratory Contribution Number 17.