Change in the Recent Warming Trend of Sea Surface Temperature in the East Sea (Sea of Japan) Over Decades (1982–2018)

remote sensing

Article Change in the Recent Warming Trend of Sea Surface Temperature in the East Sea (Sea of Japan) over Decades (1982–2018)

Eun-Young Lee 1 and Kyung-Ae Park 2,* 1 Department of Science Education, Seoul National University, Seoul 08826, Korea; [email protected] 2 Department of Earth Science Education/Research Institute of Oceanography, Seoul National University, Seoul 08826, Korea * Correspondence: [email protected]; Tel.: +82-2-880-7780

Received: 11 October 2019; Accepted: 6 November 2019; Published: 7 November 2019

Abstract: Long-term trends of sea surface temperature (SST) of the East Sea (Sea of Japan, EJS) were estimated by using 37-year-long satellite data, for the observation period from 1982 to 2018. Overall, the SST tended to increase with time, for all analyzed regions. However, the warming trend was steeper in the earlier decades since the 1980s and slowed down during the recent two decades. Based on the analysis of the occurrence of events with extreme SST (high in the summertime and low in the wintertime), a shift toward the more frequent occurrence of events with extremely high SST and the less frequent occurrence of events with extremely low SST has been observed. This supports the observations of the consistent warming of the EJS. However, seasonal trends revealed continuous SST warming in the summertime, but frequent extreme SST cooling in the wintertime, in recent decades. The observed reduction in the warming rates occurred more frequently in speciﬁc regions of the EJS, where the occurrence frequency of events with extremely low SST was unusually high in the recent decade. The recent tendency toward the SST cooling was distinctively connected with variations in the Arctic Oscillation index. This suggests that changes in the Arctic Ocean environment likely aﬀect the recently observed SST changes in the EJS, as one of the marginal seas in the mid-latitude region far from the polar region.

Keywords: sea surface temperature (SST); warming trend; SST cooling; East Sea (Sea of Japan); Arctic Oscillation

1. Introduction Sea water temperature of the global ocean has been increasing signiﬁcantly over the past decades. The rate of the temperature increase in the upper ocean layer (within 75 m of the ocean surface) was 1 reported as 0.11 ◦C decade− , for the observation period from 1971 to 2010 [1]. Since the 1970s, as the ocean observations were recorded regularly, data on the mean temperature change in the global ocean have been available [2,3]. In addition to in-situ temperature measurements, satellite observations of the ocean allow to monitor and understand long-term changes in the sea surface temperature (SST) of the global ocean. Frequent observations, performed by various satellites, provide synoptic views of the SST increase, as well as its temporal variations (e.g., [4,5]). Due to global warming, much attention has been given to the obvious warming of the Earth, including the ocean, owing to the increase in the rate and amount of emitted greenhouse gases [6–8]. The SST warming rate seems to be obviously detectable, as evidenced by the analyses of data from numerous SST databases worldwide (e.g., [1,9–11]). Recently, some studies reported a slowdown in the global ocean’s SST warming rate during recent decades [1,12,13]. The occurrence of events with extremely high temperature has been increasing, yielding surface warming [14].

Remote Sens. 2019, 11, 2613; doi:10.3390/rs11222613 www.mdpi.com/journal/remotesensing Remote Sens. 2019, 11, x FOR PEER REVIEW 2 of 18

Remote Sens. 2019, 11, 2613 2 of 17 to changes in the atmospheric circulation, which might slow down the SST increase rate [15]. In local sea regions, the effect of extremely hot and cold days on the SST increase rate was also investigated forConversely, the East China the occurrence Sea [16]. Therefore, of extremely it is low-temperatureimportant to elucidate events the in impact the wintertime of climate has change also been on localincreasing, seas, especially owing to changesthe marginal in the atmosphericseas of the global circulation, ocean; which this mightunderstanding slow down can the help SST increasein the developmentrate [15]. In local of feasible sea regions, strategies the e fftoect address of extremely climate hot change. and cold days on the SST increase rate was also investigatedThe East forSea the (Sea East of ChinaJapan, SeaEJS) [16 is]. one Therefore, of the marginal it is important seas of to the elucidate Northwest the impact Pacific of Ocean, climate locatedchange at on the local far-eastern seas, especially end of the the Asian marginal continen seast of with the globalthe cold ocean; and warm this understanding current systems can (Figure help in 1)the [17]. development As it is located of feasible in the strategiesmid-latitude to addressregion affected climate change.by the Siberian High in winter and by the NorthThe Pacific East High Sea (Sea in summer, of Japan, EJS)it provides is one of a thegood marginal opportunity seas of for the investigating Northwest Pacific the SST Ocean, trends located in associationat the far-eastern with changes end of thein the Asian atmospheric continent conditio with thens cold over and decades. warm currentTime series systems of the (Figure SST 1in)[ the17 ]. EJSAs reveal it is located an increasing in the mid-latitude trend, depending region on aff theected peri byod the of Siberianthe data, High that is in much winter stronger and by than the North that ofPacific the global High inocean summer, for the it provides observation a good period opportunity from 1982 for investigatingto 2006 [11]. the Only SST a trends few studies in association have investigatedwith changes the in spatiotemporal the atmospheric pattern conditions of the SST over warming decades. trend Time in series the EJS of [1 the8,19], SST despite in the EJSits being reveal anan ideal increasing model trend, system depending for studying on the the period impact of of the recent data, thatregional is much changes stronger due than to change that of in the global global oceanocean warming. for the observation Moreover, periodnone of from the existing 1982 to 2006studies [11 ].have Only considered a few studies the potential have investigated effect of the the observedspatiotemporal increasing pattern tendency of the toward SST warming extremely trend cold in weather the EJS [in18 ,the19], wintertime despite its using being multi-decadal an ideal model satellitesystem data. for studying the impact of recent regional changes due to change in global ocean warming. Moreover,Therefore, none in of this the study, existing we studiesexamined have long-term considered SST the trends potential and recent effect oftrend the observedchanges in increasing the EJS, whichtendency are very toward important extremely to coldthe climate weather of in the the Korean wintertime peninsula. using The multi-decadal objectives satelliteof this study data. are 1) to estimateTherefore, the fundamental in this study, variations we examined and derive long-term the trend SST trends in the and long-term recent trendSST in changes the EJS inover the the EJS, 37-year-longwhich are very observation important period, to the from climate 1982 of to the 2018; Korean 2) to peninsula. investigate The any objectives tendency ofof thisthe SST study trends are 1) byto applying estimate thea 20-year-wide fundamental variationssliding window and derive to the the total trend SST in thedatabase long-term to illustrate SST in the a EJSsignificant over the increase37-year-long in the observation occurrence period, of SST from cooling 1982 toevents 2018; in 2) torecent investigate decades, any and tendency reveal of specific the SST seasonal trends by variationsapplying ain 20-year-wide the SST trends sliding focused window on the to SST the cool totaling; SST 3) database to uncover to illustrate any relation a significant of the SST increase trends in tothe extremely occurrence hot ofor SSTcold cooling events; events and 4) into recentexamine decades, the relationship and reveal between specific the seasonal slowdown variations of the in SST the increaseSST trends rate focused and the on Arctic the SST Oscillation cooling; (AO), 3) to uncover using the any wavelet relation coherency of the SST analysis. trends to extremely hot or cold events; and 4) to examine the relationship between the slowdown of the SST increase rate and the Arctic Oscillation (AO), using the wavelet coherency analysis.

Figure 1. (a) Bathymetry of the seas around Korean peninsula; and (b) a schematic current map with Figurecold (blue) 1. (a) Bathymetry and warm (red) of the currents seas arou (EKWC:nd Korean East peninsula; Korea Warm and Current, (b) a schematic NKCC: current North Koreamap with Cold coldCurrent, (blue) PCC: and Primoryewarm (red) Cold currents Current) (EKWC: (adapted East from Korea [17 ]),Warm where Current, the line NKCC: width represents North Korea the overallCold Current,strength PCC: of the Primorye current and Cold the Curre greennt) box (adapted shows the from study [17]), area. where the line width represents the overall strength of the current and the green box shows the study area.

Remote Sens. 2019, 11, 2613 3 of 17

2. Data and Methods

2.1. Satellite Data Several satellite databases include data on the EJS (e.g., optimum interpolation SST (OISST), multi-scale ultra-high resolution SST (MUR SST), operational SST and sea ice analysis (OSTIA), extended reconstructed SST (ERSST), and Hadley Centre sea ice and SST (HadISST)). These datasets are characterized by various spatial-temporal resolutions and data periods. MUR SST and OSTIA contain data in high temporal resolution (one day) and spatial resolutions of 1 km and 0.05◦, respectively. Despite such high spatiotemporal resolutions, these datasets cover relatively short time periods (since May 2002 for the MUR SST database and since April 2006 for the OSTIA database). The other SST databases, such as ERSST and HadISST, cover longer time periods, since 1854 and 1870, respectively. However, their spatial resolutions are relatively low, at 2◦ and 1◦, respectively, and their temporal resolutions are low (one month), which makes it difficult to determine the details of the SST changes in the EJS. Therefore, in this study, the daily OISST V2 high-resolution data provided by National Oceanic and Atmospheric Administration (NOAA)/Earth System Laboratory (ESRL)/Physical Science Division (PSD), were used to investigate the long-term SST variations in the EJS. This is a global dataset, produced by the optimal interpolation of the advanced very high-resolution radiometer (AVHRR) satellite data with ship and buoy data [20]. It has a spatial resolution of 0.25 0.25 , a time resolution ◦ × ◦ of one day, and has been producing SST data since September 1981. Thus, the present study used the OISST database for the 37-year-long observation time, from 1982 to 2018.

2.2. In-Situ Data For validating trends in the satellite SST data, in-situ temperature measurements performed by the Korea Oceanographic Data Center (KODC), with a temporal resolution of 2 months, were additionally used when analyzing the southwestern part of the EJS. These data have been acquired by regular stations in an ongoing study that has been initiated by the National Fisheries Research and Development Institute (NFRDI) in 1961. We selected only the 0-m temperature data out of the KODC database, at the standard depths from 0 m to 500 m or deeper depths (e.g., 0, 10, 20, 30, 50, 75, 100, 125, 150, 200, 250, 300, 400, 500 m). The regular measurement stations of the KODC are concentrated in the coastal regions with relatively small numbers of stations in the oﬀshore regions. By contrast, the satellite SST database has a coarse spatial resolution of 0.25 0.25 . Owing to the low resolution ◦ × ◦ of the satellite data, we dropped the data from a few stations nearest to the coastal line, for comparison of the SST trends in the oﬀshore regions, by considering the resolution of the satellite data.

2.3. Arctic Oscillation (AO) Index The AO index data from NOAA/National Centers for Environmental Prediction (NCEP)/Climate Prediction Center (CPC) were used to delineate the potential causes of recent changes attributed to extreme events in the EJS. The AO index pattern was defined as the first leading mode based on the EOF (Empirical Orthogonal Function) analysis of monthly mean height anomalies at 1000 hPa [21]. As the AO index exhibits high variability during the wintertime, the index mainly reflects the AO characteristics during the wintertime. Although the EJS is located in the mid-latitude region, far from the Arctic region, the atmospheric conditions in the wintertime are quite similar, with very low air temperature and high wind speed over the sea surface, especially in the northern part of the EJS. The AO index is strongly correlated to surface air temperature variations over the Eurasian continent [21–23]. Thus, a negative (positive) AO index reflects a higher (lower) Siberian High and a very low (high) surface air temperature over eastern China [24,25]. The variability of Aleutian Low also shows a strong relation to the AO index [26]. Therefore, the relationship between the AO index and SST variability was investigated to explain recent changes in the SST variations. Remote Sens. 2019, 11, 2613 4 of 17

2.4. Empirical Orthogonal Function (EOF) Analysis Principal components of variations in time series data can be decomposed using the EOF analysis. The EOF analysis was performed to extract the dominant variability components in the SST data. This method aims at decomposing a data set into a product of a set of spatial patterns and time series [27–29], as follows: XM d xi, tj = φk(xi)ak tj (1) k=1 where d is the data set, xi is the position, tj is the time, φk is the spatial pattern of the k-th mode, ak is the amplitude time series of the k-th mode, and M is the number of spatial elements. The functions φk, which are the basis functions, are chosen to be orthogonal to the other basis functions, to account for as much variance as possible, as follows:

XM φi(xk)φj(xk) = δij (2) k=1 where δij is the Kronecker delta. Using this method, it is possible to study spatiotemporal patterns of climate variability. Therefore, since its introduction to geophysics by [30], this approach has been widely used in meteorology and oceanography (e.g., [31–34]).

2.5. Wavelet Coherency Analysis We applied wavelet coherence analysis to detect variations in coherence and phase difference between two data sets. Wavelet analysis was first applied to each data. Wavelet analysis is one of the tools for analysis of localized power variations within a time series, and has been used in numerous geophysics studies [35–37]. By decomposing a time series into the time-frequency space, this analysis allows to determine both the dominant modes of variability and how these modes vary in time. One particular wavelet is defined as [37]

1/4 2 ψ (η) = π− exp(iω η)exp η /2 (3) 0 0 − where ω0 is dimensionless frequency and η is dimensionless time. Wavelet coherence captures the correlation between two signals in both the time and frequency domains. It is deﬁned as [38]

2 1 XY S s− Wn (s) R2 (s) = (4) n 2 2 S(s 1 WX(s) ) S(s 1 WY(s) ) − n · − n XY X Y where S is a smoothing operator, s is a scale, W is a continuous wavelet transform, and W = W W ∗, where denotes complex conjugation. Wavelet coherence takes values between 0 (no correlation) and 1 ∗ (full correlation).

3. Results

3.1. Spatial and Temporal Variability of SST Figure2 shows the average characteristics of the SST variations in the study area. Going from the southern part of the EJS to its northern part, the mean SST variation decreases from ~20 ◦C in the Korea Strait to ~5 ◦C in the Tatar Strait, with relatively concentrated SST diﬀerences along the subpolar front (SPF) along the zonal line of about 40◦N[39]. The spatial distribution of the standard deviation of SST variations reveals a striking contrast between the continental side (with high values, above 6 ◦C) Remote Sens. 2019, 11, 2613 5 of 17

and the oﬀshore Japanese side (with relatively low values, under 6 ◦C) (Figure2b). Especially near

VladivostokRemote Sens. (near 2019, 11 132, x FOR◦Eand PEER42 REVIEW◦N), large standard deviations (>8 ◦C) are observed, which5 of implies 18 extreme SST cooling owing to the Siberian outbreak in the wintertime [40]. In contrastIn contrast to the to the time-averaged time-averaged pattern patternof of thethe SST, the the time time series series of ofspatially spatially averaged averaged monthly monthly SST anomaliesSST anomalies (SSTAs), (SSTAs), from from 1982 1982 to to 2018, 2018, over over thethe entireentire EJS, EJS, exhibits exhibits high high variability variability with with frequent frequent and random-likeand random-like peaks, peaks, as as shown shown in in Figure Figure2 c.2c. ItIt isis notednoted that that the the SSTA SSTA variation variation tends tends to increase to increase duringduring the initialthe initial stage stage of theof the observation observation period, period, from 1982 1982 to to mid-1990s. mid-1990s. The The overall overall trend trend in the in the SSTA amounted to 0.26 °C decade–1, and was statistically significant within the 95% confidence level, SSTA amounted to 0.26 C decade 1, and was statistically significant within the 95% confidence as indicated by the dashed◦ line in Figure− 2c. This warming trend is considered to be relatively strong, level, as indicated by the dashed line in Figure2c. This warming trend is considered to be relatively compared with that of the global ocean, which was estimated at 0.11 °C decade–1 based on the 1 strong,observations compared from with 1971 that to of 2010 the [1]. global ocean, which was estimated at 0.11 ◦C decade− based on the observationsTo understand from 1971 the to 2010principal [1]. mode of the spatial and temporal variability of the SST, the EOF Toanalysis understand was performed the principal using mode the 37-year-long of the spatial SSTA and data. temporal In the variability first mode of the(39.02%), SST, thethe EOF analysiseigenvectors was performed of the SSTA using were the 37-year-longall positive, implying SSTA data. a characteristic In the first modeSST warming (39.02%), (Figure the eigenvectors 3). The of thehighest SSTA values were allwere positive, observed implying to the southwest a characteristic of the Tsugaru SST warming Strait (136–139°E (Figure and3). 39–40°N), The highest which values was reported from the analysis of in-situ temperature measurements in [19]. This core corresponds were observed to the southwest of the Tsugaru Strait (136–139◦E and 39–40◦N), which was reported fromto the the analysis region with of in-situthe highest temperature warming trend, measurements as will be explained in [19]. later. This A core slowdown corresponds in the increasing to the region tendency of the EOF amplitudes appeared over the recent two decades, from 2000 to 2018, which with the highest warming trend, as will be explained later. A slowdown in the increasing tendency of corresponded to the temporal variations in the SSTA, as shown in Figure 2c. the EOF amplitudes appeared over the recent two decades, from 2000 to 2018, which corresponded to the temporal variations in the SSTA, as shown in Figure2c.

Figure 2. Spatial distribution of (a) mean and (b) standard deviation of the SST data for the observation Figure 2. Spatial distribution of (a) mean and (b) standard deviation of the SST data for the period from 1982 to 2018 in the EJS, and (c) time series of spatial mean anomaly, where the blue dashed observation period from 1982 to 2018 in the EJS, and (c) time series of spatial mean anomaly, where line is a least squares linear ﬁt. the blue dashed line is a least squares linear fit.

Remote Sens. 2019, 11, x FOR PEER REVIEW 6 of 18 Remote Sens. 2019, 11, 2613 6 of 17

Figure 3. (a) Spatial distribution of eigenvectors from the first EOF mode using OISST anomaly data, Figure 3. (a) Spatial distribution of eigenvectors from the first EOF mode using OISST anomaly data, for the observation period from 1982 to 2018, for the EJS; (b) temporal variability of the amplitude of (a), for the observation period from 1982 to 2018, for the EJS; (b) temporal variability of the amplitude of where the red dashed line is a least squares linear fit (y = 1.17x 2340.80); and (c) month-year plot of (b). (a), where the red dashed line is a least squares linear fit (y =− 1.17x − 2340.80); and (c) month-year 3.2. Long-Termplot of (b). Trends of the SST The slowdown-like feature of the SST warming, suggested by the SSTA variations in Figure2c 3.2. Long-Term Trends of the SST and by the time-varying amplitudes for the first EOF of the SST in Figure3b, revealed a steeper trend in theThe initial slowdown-like observation feature stage ofof thethe observationSST warming, period suggested and more by the moderate SSTA variations warming in during Figure the 2c andrecent by decades.the time-varying This raises amplitudes the question for the of whetherfirst EOF theof the sea SST surface in Figure has been 3b, revealed experiencing a steeper warming trend indespite the initial the slowdown observation with stage frequently of the observatio occurringn coolingperiod events.and more To moderate explore the warming SST warming during ratethe recentat different decades. spatial This locations, raises the the question linear tendencyof whether in the the sea SST surface warming has wasbeen estimated experiencing for thewarming entire despiteobservation the slowdown period, from with 1982 frequently to 2018 occurring (Figure4). co Alloling the events. SST variations To explore revealed the SST positivewarming trends, rate at 1 1 differentranging up spatial to 0.60 locations,◦C decade the− , withlinear an tendency average ofin 0.27 the ◦SSTC decade warming− . This was indicates estimated that for the the EJS entire is still observationexposed to a period, warming from environment 1982 to 2018 with (Figure positive 4). trends. All the The SST trend variations cores, withrevealed larger positive values, trends, appear –1 –1 rangingto be in theup centralto 0.60 °C and decade eastern,EJS. with The an warmingaverage of trend 0.27 was°C decade the highest. This in indicates the 136–139 that◦E the and EJS 39–40 is still◦N 1 exposedregions of to the a warming middle eastern environment part of with the EJS. positive Cores tren withds. trends The trend above cores, 0.50 ◦ withC decade larger− values,were observed appear 1 toin thebe in west the of central Tsugaru and Strait eastern (~41 EJS.◦N) andThe coreswarm withing trend trends was above the 0.40highest◦C decade in the− 136–139°Ewere observed and 39– in 1 –1 40°Nthe west regions of Soya of the Strait middle (~46◦ N).eastern Other part cores of the with EJS. trends Cores above with 0.40 trends◦C decadeabove −0.50appeared °C decade even were near –1 Wonsanobserved Bay in the (~39 west◦N). of Although Tsugaru there Strait are (~41°N) some diandfferences cores with between trends regions above in 0.40 terms °C of decade the warming were observedrate, all the in analyzedthe west of EJS Soya regions Strait revealed (~46°N). positive Other cores trends with (Figure trends4). above 0.40 °C decade–1 appeared 1 even Regardingnear Wonsan the Bay regions (~39°N). with Although low SST warmingthere are trends,some differences below 0.10 between◦C decade regions− , there in terms were of some the warmingspecific regions rate, all with the veryanalyzed weak EJS warming; regions for revealed example, positive the region trends around (Figure 40 4).◦N and 133–135◦E, along –1 the RussianRegarding coast the o ffregionsthe Primorye, with low or SST the warming southwestern trends, portion below of 0.10 the EJS°C decade (36–37◦N, there and 131–132were some◦E). Asspecific evidenced regions in with Figure very4, mostweak ofwarming; the EJS havefor ex beenample, experiencing the region around severe SST40°N warming and 133–135°E, over almost along thefour Russian decades. coast In contrast off the toPrimorye, the general or the trends southwes in mosttern of theportion EJS, why of the do EJS some (36–37°N regions and have 131–132°E). such weak warmingAs evidenced trends? in Figure We posited 4, most that of dithefferent EJS have regions been must experiencing have been severe exposed SST to warming different atmosphericover almost fourenvironments decades. orIn dicontrastfferent oceanicto the general environments trends in over most decades. of the ToEJS, be why able do to understandsome regions the have processes such weakresponsible warming for thetrends? appearance We posited of such that weakly-warming different regions sites, must it have is necessary been exposed to validate to different the SST warmingatmospheric rates. environments Thus, we estimated or different the oceanic warming enviro ratesnments using over long-term decades. in-situ To be measurements able to understand from the processes coastal region responsible around for the the Korean appearance peninsula; of such the weakly-warming results of this estimation sites, it is necessary are presented to validate in the thefollowing SST warming section. rates. Thus, we estimated the warming rates using long-term in-situ measurements from the coastal region around the Korean peninsula; the results of this estimation are presented in the following section.

Remote Sens. 2019, 11, x FOR PEER REVIEW 7 of 18

Remote Sens. 2019, 11, 2613 7 of 17

Figure 4. SST trends ( C decade 1) for the observation period from 1982 to 2018. ◦ − 3.3. Comparison of the SST Trends to In-Situ Measurements Figure 4. SST trends (°C decade–1) for the observation period from 1982 to 2018. Using long-term temperature measurements from the KODC database, the trends of the surface temperatures3.3. Comparison at eachof the stationSST Trends were to estimated In-Situ Measurements as shown in Figure5a. The circles in Figure5a denote the locations of the KODC regular observation stations, based on bimonthly observations at 79 stations in Using long-term temperature measurements from the KODC database, the trends of the surface the EJS, to measure the water temperature and salinity in the southwestern portion of the EJS. The in-situ temperatures at each station were estimated as shown in Figure 5a. The circles in Figure 5a denote measured temperatures exhibited positive correlation with surface warming, with 95%-confidence the locations of the KODC regular observation stations, based on bimonthly observations at 79 level, regardless of coastal or offshore regions, which implied a general tendency of surface warming. stations in the EJS, to measure the water temperature and salinity in the southwestern portion of the One interesting observation is that the stations north of 37 N reported much higher warning rates EJS. The in-situ measured temperatures exhibited positive◦ correlation with surface warming, with (>0.40 C decade 1) than the stations located to the south (<0.20 C decade 1). 95%-confidence◦ −level, regardless of coastal or offshore regions, which◦ implied− a general tendency of The satellite SST database used in this study has a spatial resolution of 0.25 0.25 , which is surface warming. One interesting observation is that the stations north of 37°N reported◦ × much◦ higher different from the point measurements of the KODC data. Due to the differences of data samplings, warning rates (>0.40 °C decade–1) than the stations located to the south (<0.20 °C decade–1). such as spatial differences and observational times, it is expected that the satellite SSTs might produce × non-negligibleThe satellite di ffSSTerences database of the used trends in this with study in-situ has measurements.a spatial resolution In spiteof 0.25° of such 0.25°, diff whicherences, is satellite-baseddifferent from SSTthe warmingpoint measurements rates in Figure of 5theb were KODC in a data. good Due agreement to the differ withences the in-situ of data SST samplings, trends in termssuch as of spatial the overall differences distribution and observational of the trends, times, with it relatively is expected higher that warmingthe satellite and SSTs weaker might warming produce innon-negligible the southern region.difference Whens of comparingthe trends allwith data in-situ between measurem the trendsents. of In in-situ spite SST of andsuch satellite differences, SST, thesatellite-based obtained slope SST was warming 0.49 (R 2rates= 0.36) in Figure and the 5b correlation were in a coegoodfficient agreement was 0.60 with (p =the6.13E–7) in-situ (FigureSST trends5c). in terms of the overall distribution of the trends, with relatively higher warming and weaker warming The satellite data used in this study have a relatively large spatial grid of ~0.25◦, and those data includedin the southern microwave region. SSTs When owing comparing to a higher all probability data between of clouds the trends in the of coastal in-situ regions, SST and with satellite frequent SST, 2 variationsthe obtained of airslope temperature was 0.49 (R and= moisture0.36) and conditions.the correlation When coefficien all of thet was points, 0.60 including(p = 6.13E–7) both (Figure close (bright5c). The gray) satellite and data far (dark used gray)in this points study fromhave thea relatively coastline, large were spatial considered, grid of the ~0.25°, satellite-based and those anddata in-situ based SST trends were in general concordance, but had a relatively weak correlation and high scatter in the weak-trend range of in-situ temperatures. By contrast, the KODC trends without the Remote Sens. 2019, 11, x FOR PEER REVIEW 8 of 18 included microwave SSTs owing to a higher probability of clouds in the coastal regions, with frequent variations of air temperature and moisture conditions. When all of the points, including both close (bright gray) and far (dark gray) points from the coastline, were considered, the satellite-based and in-situ based SST trends were in general concordance, but had a relatively weak correlation and high scatter in the weak-trend range of in-situ temperatures. By contrast, the KODC trends without the Remote Sens. 2019, 11, 2613 8 of 17 points near the coastal regions within 0.25° from the coastline revealed a good correlation between the two trends, as illustrated by the red dashed line in Figure 5c. An improved comparison result was obtainedpoints near for the offshore coastal regions regions; within the obtained 0.25◦ from slope the was coastline 0.80 (R revealed2 = 0.67) and a good the correlationcorrelation betweencoefficient the wastwo 0.82 trends, (p = as1.22E–9). illustrated All byof thethese red results dashed support line in the Figure hypothesis5c. An improvedthat the retrieved comparison SST resultdata from was theobtained satellite for data, the oonceffshore those regions; too close the to obtained the coast slope are wasexcluded, 0.80 (R can2 = be0.67) used and for the the correlation analysis of coe oceanicfficient surfacewas 0.82 warming. (p = 1.22E–9). All of these results support the hypothesis that the retrieved SST data from the satellite data, once those too close to the coast are excluded, can be used for the analysis of oceanic surface warming.

1 Figure 5. (a) Trends (◦C decade− ) from in-situ near-surface water temperatures measurements, where the black circles represent the trends at 95%-conﬁdence level; (b) same but for satellite SSTs; Figure 5. (a) Trends (°C decade–1) from in-situ near-surface water temperatures measurements, where and (c) comparison between trends from in-situ and satellite SST, where the light (dark) grey dots the black circles represent the trends at 95%-confidence level; (b) same but for satellite SSTs; and (c) represent the points close to (far from) the coast and the blue (red) dashed line represents its least comparison between trends from in-situ and satellite SST, where the light (dark) grey dots represent squares linear ﬁt (blue line: y = 0.49x + 0.01 (R2 = 0.36), red line: y = 0.80x + 0.00 (R2 = 0.67)). the points close to (far from) the coast and the blue (red) dashed line represents its least squares linear 3.4.fit Moving (blue line: Trends y = in 0.49x the SST + 0.01 (R2 = 0.36), red line: y = 0.80x + 0.00 (R2 = 0.67)).

3.4. MovingAs demonstrated Trends in the bySST the analysis of long-term variations of the SSTA and the time-varying amplitude of the first EOF mode of the SSTA (Figures2 and3b), the slopes for those features that showedAs demonstrated increasing trends by the exhibited analysis slow-time-scale of long-term variationsvariations with of the relatively SSTA and rapid the warming time-varying during amplitudethe initial of observation the first EOF period mode decades, of the whileSSTA milder(Figures warming 2 and 3b), was the observed slopes for during those recent features decades. that showedTo analyze increasing the temporal trends variability exhibited inslow-time-scale the warming trends, variations we calculatedwith relatively the multi-year rapid warming moving during trends thein theinitial SST observation by applying period a 20-year-wide decades, while temporal milder domain warming for was the estimationobserved during of linear recent trends, decades. for all Tospatial analyze grids. the The temporal 20-year variability period was in determined the warmin byg considering trends, we relatively calculated high the interannual multi-year variability moving trendsof SSTs inin the the SST EJS by at applying dominant a 20-year-wide frequency amounting temporal todomain 7 years for according the estimation to the of previous linear trends, study [for41 ]. allFigure spatial6a showsgrids. theThe time 20-year series period of the spatial was determin distributioned by for considering the 20-year-long relatively moving high trends interannual in the SST variabilityvariations, of during SSTs in the the observation EJS at dominant period frequency from 1982 amounting to 2018. The to first 7 years image according shows the to the SST previous warming studyrates over[41]. theFigure entire 6a region shows for the the time first series 20 years of the of observation, spatial distribution from 1982 for to the 2001. 20-year-long The warming moving trends trendsincreased in the continuously, SST variations, until during the period the observatio from 1986n toperiod 2005; from however, 1982 theyto 2018. began The to first reduce image gradually shows theand SST even warming switched rates to over cooling the entire from theregion SPF for region, the first for 20 the years observation of observation, period from from 1982 1987 to to 2001. 2006. TheEspecially, warming they trends seemed increased to have continuously, been strongly until reducing the period along from the 1986 Russian to 2005; coast, however, for the they observation began toperiod reduce from gradually 1988 to 2007.and even One switched notable feature to coolin of thatg from trend the is SPF that region, a negative for trend,the observation that is SST period cooling, fromappeared 1987 to along 2006. the Especially, SPF in the they central seemed portion to have of the been EJS, strongly as can bereducing clearly along seen forthe theRussian observation coast, forperiod the observation from 1996 toperiod 2015, from as well 1988 as to for 2007. the followingOne notable observation feature of period.that trend is that a negative trend, that isTo SST examine cooling, such appeared slowdown along features the SPF in the ce SSTntral warming portion trends of the inEJS, more as can detail, be clearly the time seen series for theof theobservation areal percentages period from of the1996 positive to 2015, (red) as well and as negative for the following (blue) trends observation for each period. period are shown in Figure6b. For the initial observation periods, from 1982 to 2001 and from 1986 to 2005, all of the EJS regions exhibited a positive warming tendency by occupying 100% of the positive trend without any regions with negative trends. However, for the observation period from 1987 to 2006, negative mean trends began to appear at a rate of 1.80% of the total number of grids. The number of the grids Remote Sens. 2019, 11, 2613 9 of 17 corresponding to the SST cooling, as marked by blue bars in Figure6b, increased exponentially with a wide spreading of the surface cooling with time. This tendency began to appear for the observation period from 1987 to 2006 and have been intensifying until the recent two decades. The averaged trends in the SST warming and cooling exhibited gradual variations over time 1 (Figure6c). The trend reached a maximum of ~0.55 ◦C decade− for the period of 1986–2005 and 1 continuously decreased to 0.15 ◦C decade− during the recent two decades. As can be seen in the last three bars for the observation period from 1997 to 2016, the mean rate of the SST cooling (–0.19, 1 –0.17, and –0.17 ◦C decade− , respectively) amounted to a similar trend of the SST warming (0.20, 0.15, 1 and 0.15 ◦C decade− , respectively), in spite of the widespread region for an overall observed SST warming. In other words, the number of the grids with the SST cooling tendency gradually increased while that with the SST warming tendency reversely decreased with time. This implies that the SST cooling events have been more intensively increasing in relatively small regions concentrated in the Remote Sens. 2019, 11, x FOR PEER REVIEW 10 of 18 frontal zones in the central part of the EJS.

Figure 6. (a) Temporal trend for the observation period of 20 years; (b) time series of the percentage of Figure 6. (a) Temporal trend for the observation period of 20 years; (b) time series of the percentage the number of the grids; and (c) regional averaged trend of (a) only for positive (red) and negative (blue)of the values.number of the grids; and (c) regional averaged trend of (a) only for positive (red) and negative (blue) values.

Remote Sens. 2019, 11, 2613 10 of 17

3.5. Monthly Variations in the SST Trends As mentioned earlier, the degree of the SST warming of the EJS has been mitigated over recent decades. It is important to investigate any potential seasonal impact on the observed reduction in the increase of the warming rate. Thus, we calculated the monthly SST trends, their areal fractions, and the average trend values, for both warming and cooling separately (Figure7). The monthly trends for the summer season, from June to August, appeared to show positive trends over the entire EJS (Figure7a). However, the other months, except for the summer months, revealed the existence of weak negative trends along the Russian coast, the SPFs, and the southwestern part of the EJS near the Korea Strait. Especially in the springtime (during the March–May period), negative trends were concentrated along the SPF region in the central part of the EJS. By contrast, strongest warmings appeared from June to November, and extended to the eastern part of the EJS. In the wintertime (during the December–February period), the pattern of monthly SST trends was similar to that for the total trend in Figure4. In other words, the warming trend was large for all seasons except for spring, and especially so from August to November. This observation was quite diﬀerent from previously reported warming that has been attributed to the increasing role of wintertime SST variations [17]. The areal coverage of cooling constituted a relatively high fraction, from 7% in December to the maximum of 18% in March (Figure7b) in winter and spring. The strongest warming trend of 1 1 ~0.40 ◦C decade− appeared in August, while the weakest warming trend, ~0.18 ◦C decade− , appeared 1 in March (Figure7c). Concerning the cooling, the strongest cooling trend was about –0.09 ◦C decade − , and was observed for February and March.

3.6. Occurrence of Extremely High/Low SSTs We analyzed the occurrence of extreme SST events for determining the potential causes of the observed weakening of the warming trend and the seasonal shift of the warming trend leading from winter to summer. Before defining extreme events, daily SSTAs were calculated by subtracting the climatological mean from the SST values for the same date. Extremely high SST days (EHDs) were defined as days within the top 10% (90th percentile) of anomalies in the summertime (June–September). In a similar way, extremely low SST days (ELDs) were defined as days in the bottom 10% of anomalies, but for the wintertime (December–March). Figure8a shows the spatial distribution of the trends of the EHD occurrence frequency for the entire observation period of 37 years. It is notable that all regions exhibit positive trends, implying continuous warming of the EJS over the past decades. The spatially averaged value of the EHD occurrence 1 frequency tended to increase with time at a rate of 5.82 days decade− (Figure8b). Regarding the cooling events, the spatial distribution of the trends of the ELD occurrence frequency exhibited negative trends for most of the regions in the EJS except for a few regions associated with SPFs, with weakly 1 positive trends of ~0.25 days decade− (Figure8d). The time series of the ELDs exhibit the decreasing 1 tendency (–4.31 days decade− ) with high peaks of ~57 days in the 1980s and minor days of less than 20 days during other periods (Figure8e). This overall characteristic is consistent with the long-term warming trend in the EJS. To understand the difference between cooling and warming regions in the SST variations, we selected two representative regions, marked by blue (weak warming) and red (strong warming) boxes in Figure8a,d. As the trend in the strong-warming sites corresponds to the overall warming trend, we focused on the trend in the weak-warming sites. Figure8c shows the time series of the EHDs in small regions that exhibited warming trends, revealing the tendency of the EHDs to increase temperature with time, with the highest peak occurring in 2010. However, the trend of the occurrence frequency of the ELDs differed from that of the entire EJS. The high frequency (more than 45 days) of the ELDs seemed to occur somewhat periodically every decade (Figure8f). Except for the extremely cold days, most of the ELDs were relatively small, lasting less than 20 days. The occurrence of the ELDs was the highest in 2015 instead of the extremely cold days prior to substantial global warming Remote Sens. 2019, 11, 2613 11 of 17

decades ago. This feature exempliﬁes the signiﬁcant role of extremely cold days on the slowdown of

Remotethe recent Sens. 2019 SST, 11 increase., x FOR PEER REVIEW 11 of 18

FigureFigure 7. (a) MonthlyMonthly variationsvariations of of the the SSTA SSTA trends trends from from the OISSTthe OISST database database for the for observation the observation period b c periodfrom 1982 from to 1982 2018; to ( 2018;) monthly (b) monthly percentage percentage of the number of the number of grids; of and grids; ( ) monthly and (c) regionalmonthly averagedregional trend of (a) only for positive (red) and negative (blue) values. averaged trend of (a) only for positive (red) and negative (blue) values.

RemoteRemote Sens. Sens. 20192019, 11, 11, x, 2613FOR PEER REVIEW 13 12 of of 18 17

Figure 8. (a) Spatial distribution of trend of EHD occurrence frequencies, where blue (red) box represents Figure 8. (a) Spatial distribution of trend of EHD occurrence frequencies, where blue (red) box the weak (strong) warming area; (b) time series of regional averaged EHD occurrence frequencies; represents(c) time series the weak of regional (strong) averaged warming EHD area; occurrence (b) time frequencies series of regional in the weak averaged warming EHD area occurrence (blue box); frequencies;and (d, e, f )( samec) time as series (a, b, cof) butregional for ELD, averaged where theEHD dashed occurrence lines offrequencies (b) and (e) in are they =weak0.58 xwarming1151.10 − areaand (blue y = box);0.43x and+ 875.41, (d, e, f respectively.) same as (a, b, c) but for ELD, where the dashed lines of (b) and (e) are y = 0.58x − 1151.10 and y = −0.43x + 875.41, respectively. 3.7. Relation to the AO 3.7. Relation to the AO What causes the observed periodic appearance of extremely low SST events? One representatively influentialWhat causes climate the index observed that captures periodic SST variationsappearance in theof EJSextremely is the AOlow index. SST Weevents? studied One the representativelyvariability of AO, influential which, climate as we posited, index that could captur affectes SST the variations SST in winter in the in EJS the is EJS, the especiallyAO index. withWe studiedrespect the to extremevariability events. of AO, Thus, which, we as sought we posited, to determine could affect if there the was SST significant in winter in change the EJS, in especially the energy withof AO respect variations to extreme with events. time by Th performingus, we sought wavelet to determine analysis if ofthere available was significant data; the change results in of the this energyanalysis of areAO shownvariations in Figure with 9timea. When by performing considering wavelet time periods analysis longer of available than 3 years,data; the the results AO index of thisexhibited analysis relatively are shown low in variability Figure 9a. in When the earlier consid decadesering time of the periods observation longer period. than 3 However, years, the higher AO indexvariability exhibited with relatively strong energy low variability variations in appeared the earlier in decades the recent of the decades observation of the observationperiod. However, period, higheras shown variability in Figure with9a. strong The results energy of variations the wavelet appeared analysis in of the the recent AO for decades wintertime of the data observation revealed period,a large as variability shown in with Figure 2–2.5 9a. year-long The results cycles of aroundthe wavelet the year analysis 2000, andof the with AO 2.5–3.5 for wintertime year-long cyclesdata revealedafter 2010 a large (Figure variability9a). This with suggests 2–2.5 thatyear-long the higher cycles variability around the of year changes 2000, in and the with atmospheric 2.5–3.5 year- and longoceanic cycles features after in2010 the Arctic(Figure Ocean 9a). mayThis besuggests related tothat the the recent higher fast warmingvariability and of meltingchanges of in the the sea atmosphericice [42–44]. and Next, oceanic we asked features whether in the this Arctic AO variabilityOcean may can be arelatedffect the to SST the trendsrecent infast the warming EJS, which and is meltingone of theof the marginal sea ice seas[42–44]. of the Next, Northwest we asked Pacific whether in the this mid-latitude AO variability region. can affect To answer the SST this trends question, in thewe EJS, performed which is wavelet one of analysisthe marginal of SST seas variations of the Northwest in the ELDs, Pacific and in the the results mid-latitude of this analysisregion. To are answer this question, we performed wavelet analysis of SST variations in the ELDs, and the results of this analysis are shown in Figure 9b. The highest variability in the SSTA was observed in 2010, with a temporal period of ~3.5 years.

Remote Sens. 2019, 11, 2613 13 of 17

Remote Sens. 2019, 11, x FOR PEER REVIEW 14 of 18 shown in Figure9b. The highest variability in the SSTA was observed in 2010, with a temporal period of ~3.5Wavelet years. coherence was computed for examining a possible relationship between the AO and SSTAWavelet in the ELDs. coherence As wavelet was computed coherence forquantifies examining correlation, a possible the relationship coherence takes between on the the value AO of and 1 ifSSTA the two in the analyzed ELDs. signals As wavelet are hi coherenceghly correlated. quantifies Figure correlation, 9c shows that the coherencethe coherence takes between on the valuethe two of signals1 if the is two above analyzed 0.7 (corresponding signals are highly 95% of correlated. confidence Figure level),9c for shows the observation that the coherence period from between 2010 the to thetwo present signals time. is above This 0.7 suggests (corresponding that the AO 95% index of confidence captures the level), SST for changes the observation and the ELDs period during from recent2010 to decades, the present and in time. particular This suggests after 2005. that One the in AOteresting index observation captures the is that SST the changes coherence and thebetween ELDs theduring AO recentand the decades, SSTA in and the in ELDs particular in the after EJS 2005.exhibi Oneted insignificant interesting observation correlation is (below that the 0.1) coherence for the observationbetween the period AO and from the SSTA1982 to in the2005, ELDs over in most the EJS of exhibitedthe temporal insignificant periods. correlationArctic variations (below are 0.1) not for expectedthe observation to affect period the SST from warming 1982 to or 2005, cooling over in most the ofEJS. the This temporal implies periods. that the Arctic EJS has variations long been are protectednot expected from to the aff ectextreme the SST cold warming condition or of cooling the Arctic in the region. EJS. This Thus, implies changes that in the the EJS SST has warming long been of theprotected EJS, as from shown the in extreme Figure cold6, especially condition the of theobserved Arctic slowdown region. Thus, of warming changes inwith the SSTfrequent warming and randomof the EJS, extremely as shown cold in days Figure in6 winter,, especially may thehave observed been induced slowdown by changes of warming in the withAO that frequent occur andas a partrandom of the extremely global warming cold days trend. in winter, may have been induced by changes in the AO that occur as a part of the global warming trend.

Figure 9. Wavelet analysis of (a) AO index in winter; and (b) SSTA in ELDs; and (c) wavelet coherence Figureof (a) and 9. Wavelet (b). analysis of (a) AO index in winter; and (b) SSTA in ELDs; and (c) wavelet coherence of (a) and (b). 4. Discussion 4. DiscussionAccording to [45], who analyzed seasonal trends in the global temperature for the period from 1979According to 2010, the to SST [45], in who the Northern analyzed Hemisphere seasonal trends exhibited in the a significantglobal temperature warming for trend the for period all seasons from 1979in the to past 2010, decades. the SST At in that the time, Northern the warming Hemisphe ratere was exhibited reported a significant to be larger warming in autumn trend and for winter all seasonsthan in springin the past and summer.decades. At A similarthat time, result the for warming the SST rate warming was reported of the EJS to wasbe larger reported, in autumn stating and that winterthe warming than in was spring more and significant summer. in A the similar wintertime result than for the in the SST summertime warming of for the the EJS period was reported, from 1891 statingto 2005 that [18]. the As warming one of the was effects more of significant global climate in the change wintertime on changes than inin thethe SSTsummertime variation infor local the periodregions, from [11] 1891 showed to 2005 that the[18]. SST As inone the of EJS the exhibited effects of an global overall climate increase change of more on thanchanges 0.9 ◦inC overthe SST the variationtime period in local of 25 regions, years, from [11] 1982 showed to 2006, that including the SST in a rapidthe EJS increase exhibited (more an thanoverall 12-fold) increase in the of globalmore 1 thanwarming 0.9 °C rate over at the 1.67 time◦C decadeperiod− of during25 years, the from observation 1982 to 2006, period including from 1986 a rapid to 1998. increase (more than 12-fold)In contrastin the global to the warming previous rate studies at 1.67 of the °C warmingdecade–1 ofduring the EJS, thethis observation study demonstrated period from a 1986 reverse to 1998.tendency that the higher SST warming have been led by change in summertime SSTs rather thanIn that contrast in winter to the by previous accompanying studies of with the warming the reduction of the of EJS, warming this study rates demonstrated regardless ofa reverse season. tendencyNevertheless, that the the higher overall SST warming warming rates have were been still led positive, by change not in only summertime over the entire SSTs rather period, than but that also inin winter the recent by accompanying decades. In particular, with the SSTreduction cooling of atwarming specific rates regions regardless of the EJS, of season. such as Nevertheless, in the central theregion overall of the warming SPFs and rates along were the still Russian positive, Primorye not only coast, over has the begun entire to period, appear but for thealso recent in the decades recent decades.after 2003. In Suchparticular, a cooling SST trendcooling appeared at specific in region all seasonss of the except EJS, forsuch the as summer. in the central This region suggests of thatthe SPFswinter, and which along led the to Russian a warming Primorye trend previously,coast, has begun may contribute to appear tofor the the frequent recent decades increase after of ELDs 2003. as Sucha measure a cooling of surface trend appeared cooling in in the all lastseasons years. except for the summer. This suggests that winter, which led toWhat a warming are the trend main previously, causes of themay monthly contribute distinction to the frequent of SST increase cooling of rates? ELDs What as a measure causesthe of surfaceslowdown cooling of general in the last SST years. warming in the EJS? What is the role of change in SST in winter on the What are the main causes of the monthly distinction of SST cooling rates? What causes the slowdown of general SST warming in the EJS? What is the role of change in SST in winter on the weakening of SST warming? It is known that the Arctic warming is much stronger than changes in

Remote Sens. 2019, 11, 2613 14 of 17 weakening of SST warming? It is known that the Arctic warming is much stronger than changes in lower latitudes [42,44]. It was reported that such Arctic warming can influence the occurrence of extreme atmospheric weather in the mid-latitudes through dynamical pathways such as changes in the storm tracks, the jet stream and the planetary waves [46]. In addition, [47] argued that severe winters may occur in the mid-latitude regions due to the northerly flow of cold air in the atmosphere, resulting from the Arctic warming accompanied by the local development of an anomalous anticyclone and the downstream development of the mid-latitude trough. Such atmospheric cooling of mid-latitude regions induced by the Arctic warming is expected to produce SST cooling during winter outbreaks over the sea surface of the EJS. This argument supports the results of this study concerning the less warming and re-initiation of surface cooling in the recent decades. This corresponds to recent finding on the re-initiation of bottom water formation in the EJS sensitive to changing surface conditions and atmospheric forcing [48]. The findings of the present study could be attributed to the regional-scale imprints of the large-scale global warming hiatus (GWH) that is referred to as the global warming slowdown, on a period of relatively small warming in the surface temperature of the Earth [12,13]. Additionally, it is suggested that the Antarctic Oscillation (AAO), also known as the Southern Annular Mode (SAM), which is a low-frequency mode of atmospheric variability of the Southern Hemisphere, can affect not only the climate change in the Southern Hemisphere but also in the Northern Hemisphere [49–51]. The AAO is capable of influencing the AO through remotely connected atmospheric variability. This study demonstrates one of the examples of the impact of Artic environmental changes and GWH on the regional SST variations, and discusses the characteristics and causes of the regional SST variations in the EJS. Therefore, it is expected that this study could be further explored for a deep understanding of the linkage between regional responses and other climate indices.

5. Conclusions The present study showed long-term SST warming trends in almost all of the EJS regions, during the 37-year-long observation period, from 1982 to 2018. The trend was related to the slowdown of the SST warming, especially in the SPF area. Analyzing the changes in the extremely high SST events in the summertime and extremely low SST events in the wintertime, regional averages for the EJS exhibited a reduction in the ELD and an increase in the EHD, consistent with the overall warming trend. However, in the areas where warming was relatively weak, the ELD exhibited large values frequently, with the maximal value observed within the last 5 years. As a result of examining the relationship between the AO and the SSTA corresponding to the ELD, the consistency between these two variables increased, especially for the recent decade. This suggests that changes that take place in the Arctic Ocean can explain some recent SST changes in the EJS through changes in atmospheric eﬀects. The SST warming and cooling, presented in this study, corroborate an oceanic response of the EJS through the atmospheric connection between the extremely cold Arctic environment and mid-latitude regions. This implies that the EJS, as a small-scale replica of the global ocean, is exposed to the overall global warming trend as well as to the extreme surface cooling that originates in the Arctic Ocean. This suggests that the EJS, as one of the marginal mid-latitude seas, might be strongly linked with the polar region in winter, in spite of the distance. Thus, further analysis is needed for an in-depth understanding of long-term SST changes in the EJS in the warming climate.

Author Contributions: Conceptualization, K.-A.P.; data curation, E.-Y.L.; methodology, K.-A.P.and E.-Y.L.; writing –original draft preparation, E.-Y.L.; writing – review and editing; K.-A.P. and E.-Y.L. Funding: This work was funded by the Korea Meteorological Administration Research and Development Program under Grant KMI2018-05110. Acknowledgments: We thank anonymous reviewers for their insightful comments, which helped to make an improvement to this manuscript. NOAA OISST V2 data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their Web site at https://www.esrl.noaa.gov/psd. Conﬂicts of Interest: The authors declare no conﬂicts of interest. Remote Sens. 2019, 11, 2613 15 of 17

Abbreviations The following abbreviations are used in this manuscript:

AO Arctic Oscillation AAO Antarctic Oscillation AVHRR advanced very high resolution radiometer EJS East Sea (Sea of Japan) EKWC East Korea Warm Current ELD extremely low SST day EHD extremely high SST day EOF empirical orthogonal function ERSST extended reconstructed SST GWH global warming hiatus HadISST Hadley Centre sea ice and SST KODC Korea Oceanographic Data Center MUR SST multi-scale ultra-high resolution SST NFRDI National Fisheries Research and Development Institute NKCC North Korea Cold Current NOAA National Oceanic and Atmospheric Administration OISST optimum interpolation SST OSTIA operational SST and sea ice analysis PCC Primorye Cold Current SPF subpolar front SST sea surface temperature SSTA SST anomaly

References

1. IPCC. Climate Change 2014: Synthesis Report. Contribution of Working Group I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change; Core Writing Team, Pachauri, R.K., Meyer, L.A., Eds.; IPCC: Geneva, Switzerland, 2014; p. 151. 2. Domingues, C.M.; Church, J.A.; White, N.J.; Gleckler, P.J.; Wijﬀels, S.E.; Barker, P.M.; Dunn, J.R. Improved estimates of upper-ocean warming and multi-decadal sea-level rise. Nature 2008, 453, 1090–1093. [CrossRef] 3. Lyman, J.M.; Johnson, G.C. Estimating annual global upper-ocean heat content anomalies despite irregular in-situ ocean sampling. J. Clim. 2008, 21, 5629–5641. [CrossRef] 4. Liu, L.; Lozano, C.; Iredell, D. Time–Space SST Variability in the Atlantic during 2013: Seasonal Cycle. J. Atmos. Oceanic Technol. 2015, 32, 1689–1705. [CrossRef] 5. Gentemann, C.L.; Fewings, M.R.; García-Reyes, M. Satellite sea surface temperatures along the West Coast of the United States during the 2014–2016 northeast Paciﬁc marine heat wave. Geophys. Res. Lett. 2017, 44, 312–319. [CrossRef] 6. Levitus, S.; Antonov, J.I.; Wang, J.; Delworth, T.L.; Dixon, K.W.; Broccoli, A.J. Anthropogenic warming of Earth’s climate system. Science 2001, 92, 267–270. [CrossRef][PubMed] 7. Barnett, T.P.; Pierce, D.W.; Schnur, R. Detection of anthropogenic climate change in the world’s oceans. Science 2001, 292, 270–274. [CrossRef][PubMed] 8. Pierce, D.W.; Barnett, T.P.; AchutaRao, K.M.; Gleckler, P.J.; Gregory, J.M.; Washington, W.M. Anthropogenic warming of the oceans: Observations and model results. J. Clim. 2006, 19, 1873–1900. [CrossRef] 9. Levitus, S.; Antonov, J.I.; Boyer, T.P.; Stephens, C. Warming of the world ocean. Science 2000, 287, 2225–2229. [CrossRef] 10. Barnett, T.P.; Pierce, D.W.; AchutaRao, K.M.; Gleckler, P.J.; Santer, B.D.; Gregory, J.M.; Washington, W.M. Penetration of human-induced warming into the world’s oceans. Science 2005, 309, 284–287. [CrossRef] 11. Belkin, I.M. Rapid warming of large marine ecosystems. Prog. Oceanogr. 2009, 81, 207–213. [CrossRef] Remote Sens. 2019, 11, 2613 16 of 17

12. Easterling, D.R.; Wehner, M.F. Is the climate warming or cooling? Geophys. Res. Lett. 2009, 36, L08706. [CrossRef] 13. Meehl, G.A.; Teng, H.; Arblaster, J.M. Climate model simulations of the observed early-2000s hiatus of global warming. Nat. Clim. Change 2014, 4, 898. [CrossRef] 14. Seneviratne, S.I.; Donat, M.G.; Mueller, B.; Alexander, L.V. No pause in the increase of hot temperature extremes. Nat. Clim. Change 2014, 4, 161. [CrossRef] 15. Johnson, N.C.; Xie, S.P.; Kosaka, Y.; Li, X. Increasing occurrence of cold and warm extremes during the recent global warming slowdown. Nat. Commun. 2018, 9, 1724. [CrossRef][PubMed] 16. Wang, Q.; Li, Y.; Li, Q.; Liu, Y.; Wang, Y.N. Changes in means and extreme events of sea surface temperature in the east china seas based on satellite data from 1982 to 2017. Atmosphere 2019, 10, 140. [CrossRef] 17. Park, K.A.; Park, J.E.; Choi, B.J.; Byun, D.S.; Lee, E.I. An oceanic current map of the East Sea for science textbooks based on scientific knowledge acquired from oceanic measurements. Sea 2013, 18, 234–265. [CrossRef] 18. Yeh, S.W.; Park, Y.G.; Min, H.; Kim, C.H.; Lee, J.H. Analysis of characteristics in the sea surface temperature variability in the East/Japan Sea. Prog. Oceanogr. 2010, 85, 213–223. [CrossRef] 19. Na, H.; Kim, K.Y.; Chang, K.I.; Park, J.J.; Kim, K.; Minobe, S. Decadal variability of the upper ocean heat content in the East/Japan Sea and its possible relationship to northwestern Pacific variability. J. Geophys. Res. Oceans 2012, 117, C02017. [CrossRef] 20. Reynolds, R.W.; Smith, T.M.; Liu, C.; Chelton, D.B.; Casey, K.S.; Schlax, M.G. Daily high-resolution-blended analyses for sea surface temperature. J. Clim. 2007, 20, 5473–5496. [CrossRef] 21. Thompson, D.W.J.; Wallace, J.M. The Arctic Oscillation signature in the winter geopotential height and temperature fields. Geophys. Res. Lett. 1998, 25, 1297–1300. [CrossRef] 22. Thompson, D.W.J.; Wallace, J.M. Annular modes in the extratropical circulation. Part I: Month-to-month variability. J. Clim. 2000, 13, 1000–1016. [CrossRef] 23. Kerr, R.A. New force in high-latitude climate. Science 1999, 284, 241–242. [CrossRef] 24. Gong, D.Y.; Wang, S.W.; Zhu, J.H. East Asian winter monsoon and Arctic oscillation. Geophys. Res. Lett. 2001, 28, 2073–2076. [CrossRef] 25. Isobe, A.; Beardsley, R.C. Atmosphere and marginal-sea interaction leading to an interannual variation in cold-air outbreak activity over the Japan Sea. J. Clim. 2007, 20, 5707–5714. [CrossRef] 26. Overland, J.E.; Adams, J.M.; Bond, N.A. Decadal variability of the Aleutian low and its relation to high-latitude circulation. J. Clim. 1999, 12, 1542–1548. [CrossRef] 27. Lagerloaf, G.S.E.; Bernstein, R.L. Empirical orthogonal function analysis of advanced very high resolution radiometer surface temperature patterns in Santa Barbara Channel. J. Geophys. Res. 1988, 93, 6863–6873. [CrossRef] 28. Kelly, K.A. Comment on ‘EOF analysis of advanced very high resolution radiometer surface temperature patterns in Santa Barbara Channel’ by Gary, S.E. Lagerloef and Robert, L. Bernstein. J. Geophys. Res. 1988, 93, 15753–15754. [CrossRef] 29. Emery, W.J.; Thomson, R.E. Data Analysis Methods in Physical Oceanography, 1st ed.; Pergamon Press: Oxford, UK, 1998; p. 634. 30. Lorenz, E.N. Empirical Orthogonal Functions and Statistical Weather Prediction; Massachusetts Institute of Technology: Cambridge, MA, USA, 1956; p. 49. 31. Craddock, J.M.; Flood, C.R. Eigenvectors for representing the 500 mb geopotential surface over the Northern Hemisphere. Q. J. R. Meteorolog. Soc. 1969, 95, 576–593. [CrossRef] 32. Kidson, J.W. Eigenvector analysis of monthly mean surface data. Mon. Weather Rev. 1975, 103, 177–186. [CrossRef] 33. Trenberth, K.E. A quasi-biennial standing wave in the Southern Hemisphere and interrelations with sea surface temperature. Q. J. R. Meteorolog. Soc. 1975, 101, 55–74. [CrossRef] 34. Weare, B.C.; Navato, A.R.; Newell, R.E. Empirical orthogonal analysis of Pacific sea surface temperatures. J. Phys. Oceanogr. 1976, 6, 671–678. [CrossRef] 35. Daubechies, I. Ten Lectures on Wavelets; Society for Industrial and Applied Mathematics: Philadelphia, PA, USA, 1992; p. 357. 36. Foufoula-Georgiou, E.; Kumar, P. Wavelets in Geophysics; Academic Press: Cambridge, MA, USA, 1995; p. 373. Remote Sens. 2019, 11, 2613 17 of 17

37. Torrence, C.; Compo, G.P. A practical guide to wavelet analysis. Bull. Am. Meteorol. Soc. 1998, 79, 61–78. [CrossRef] 38. Torrence, C.; Webster, P. Interdecadal changes in the ENSO-Monsoon system. J. Clim. 1999, 12, 2679–2690. [CrossRef] 39. Park, K.A.; Ullman, D.S.; Kim, K.; Chung, J.Y.; Kim, K.R. Spatial and temporal variability of satellite-observed subpolar front in the East/Japan Sea. Deep Sea Res. Part. I 2007, 54, 453–470. [CrossRef] 40. Park, K.A.; Chung, J.Y.; Kim, K.; Cornillon, P.C. Wind and bathymetric forcing of the annual sea surface temperature signal in the East (Japan) Sea. Geophys. Res. Lett. 2005, 32, L05610. [CrossRef] 41. Park, S.; Chu, P.C. Interannual SST variability in the Japan/East Sea and relationship with environmental variables. J. Oceanogr. 2006, 62, 115–132. [CrossRef] 42. Serreze, M.C.; Barrett, A.P.; Stroeve, J.C.; Kindig, D.N.; Holland, M.M. The emergence of surface-based Arctic amplification. Cryosphere 2009, 3, 11–19. [CrossRef] 43. Serreze, M.C.; Barry, R.G. Processes and impacts of Arctic amplification: A research synthesis. Global Planet. Change 2011, 77, 85–96. [CrossRef] 44. Screen, J.A.; Simmonds, I. The central role of diminishing sea ice in recent arctic temperature amplification. Nature 2010, 464, 1334. [CrossRef] 45. Cohen, J.L.; Furtado, J.C.; Barlow, M.; Alexeev, V.A.; Cherry, J.E. Asymmetric seasonal temperature trends. Geophys. Res. Lett. 2012, 39, L04705. [CrossRef] 46. Cohen, J.; Screen, J.A.; Furtado, J.C.; Barlow, M.; Whittleston, D.; Coumou, D.; Jones, J. Recent Arctic amplification and extreme mid-latitude weather. Nat. Geosci. 2014, 7, 627. [CrossRef] 47. Kug, J.S.; Jeong, J.H.; Jang, Y.S.; Kim, B.M.; Folland, C.K.; Min, S.K.; Son, S.W. Two distinct influences of arctic warming on cold winters over North America and East Asia. Nat. Geosci. 2015, 8, 759. [CrossRef] 48. Yoon, S.T.; Chang, K.I.; Nam, S.; Rho, T.; Kang, D.J.; Lee, T.; Kim, K.R. Re-initiation of bottom water formation in the East Sea (Japan Sea) in a warming world. Sci. Rep. 2018, 8, 1576. [CrossRef][PubMed] 49. Zheng, F.; Li, J.; Liu, T. Some advances in studies of the climatic impacts of the Southern Hemisphere annular mode. J. Meteorolog. Res. 2014, 28, 820–835. [CrossRef] 50. Davis, W.J.; Taylor, P.J. The Antarctic centennial oscillation: A natural paleoclimate cycle in the Southern Hemisphere that influences global temperature. Climate 2018, 6, 3. [CrossRef] 51. Davis, W.J.; Taylor, P.J.; Davis, W.B. The origin and propagation of the Antarctic centennial oscillation. Climate 2019, 7, 112. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).