Article Responses of Summer Upwelling to Recent Climate Changes in the Taiwan Strait
Caiyun Zhang
State Key Laboratory of Marine Environmental Science, Key Laboratory of Underwater Acoustic Communication and Marine Information Technology Ministry of Education, College of Ocean and Earth Sciences, Xiamen University, Xiamen 361102, China; [email protected]
Abstract: The response of a summer upwelling system to recent climate change in the Taiwan Strait has been investigated using a time series of sea surface temperature and wind data over the period 1982–2019. Our results revealed that summer upwelling intensities of the Taiwan Strait decreased with a nonlinear fluctuation over the past four decades. The average upwelling intensity after 2000 was 35% lower than that before 2000. The long-term changes in upwelling intensities show strong correlations with offshore Ekman transport, which experienced a decreasing trend after 2000. Unlike the delay effect of canonical ENSO events on changes in summer upwelling, ENSO Modoki events had a significant negative influence on upwelling intensity. Strong El Niño Modoki events were not favorable for the development of upwelling. This study also suggested that decreased upwelling could not slow down the warming rate of the sea surface temperature and would probably cause the decline of chlorophyll a in the coastal upwelling system of the Taiwan Strait. These results will contribute to a better understanding of the dynamic process of summer upwelling in the Taiwan Strait, and provide a sound scientific basis for evaluating future trends in coastal upwelling and their potential ecological effects. Keywords: upwelling; southwest wind; ENSO Modoki; climate change; Taiwan Strait Citation: Zhang, C. Responses of Summer Upwelling to Recent Climate Changes in the Taiwan Strait. Remote Sens. 2021, 13, 1386. https://doi.org/ 1. Introduction 10.3390/rs13071386 Coastal upwelling is an important component of circulation in the shelf marginal sea. It develops when prevailing alongshore winds force offshore Ekman transport of Academic Editor: Gang Zheng surface waters, which brings cold, nutrient-rich water up to the euphotic layer, leading to high concentrations of chlorophyll [1,2]. Understanding how coastal upwelling systems Received: 5 March 2021 respond to climate change has become an area of intensive research in recent years because Accepted: 1 April 2021 Published: 3 April 2021 of the importance of upwelling to marine fisheries, biodiversity, and the carbon cycle. Many studies on eastern boundary upwelling systems [3–5] demonstrate an enhancement
Publisher’s Note: MDPI stays neutral trend in upwelling intensity under modern climate change. Meanwhile, weakening up- with regard to jurisdictional claims in welling trends are observed in other regions, such as the NW Iberian Peninsula coastal published maps and institutional affil- upwelling [6], which is contrary to the hypothesis of Bakun [3]. These studies have re- iations. vealed that upwelling trends depend heavily on a time series of different seasons, lengths, and sources, as well as the location analyzed [7–10]. However, little research has been conducted on the effect of global warming on the western boundary upwelling during the Asian summer monsoon region [11]. This topic is especially important due to the already stressed condition of the upwelling ecosystem in this region due to human activities such Copyright: © 2021 by the author. Licensee MDPI, Basel, Switzerland. as overfishing and pollution. This article is an open access article The Taiwan Strait (TWS) is an important channel connecting the South China Sea distributed under the terms and and the East China Sea. It is characterized by rough bottom topography (Figure1) and conditions of the Creative Commons a complicated current system. The circulation of the TWS is controlled mainly by the Attribution (CC BY) license (https:// East Asia monsoon [12]. Strong northeasterly winds prevail in the winter, and weak creativecommons.org/licenses/by/ southwesterly winds dominate in the summer [13]. During the summer monsoon, the 4.0/). circulation of TWS is controlled by the warm SCS water. At this time, upwelling events are
Remote Sens. 2021, 13, 1386. https://doi.org/10.3390/rs13071386 https://www.mdpi.com/journal/remotesensing Remote Sens. 2021, 13, 1386 2 of 16
Remote Sens. 2021, 13, 1386 Asia monsoon [12]. Strong northeasterly winds prevail in the winter, and weak 2south- of 16 westerly winds dominate in the summer [13]. During the summer monsoon, the circula- tion of TWS is controlled by the warm SCS water. At this time, upwelling events are gen- generallyerally observed observed along along the the coast coast of of the the wester westernn TWS TWS and and around around the Taiwan BankBank [14[14,15].,15]. BeginningBeginning in in the the 1980s, 1980s, extensive extensive studies studies have have been been conducted conducted to to examine examine the the location, location, variability,variability, control control mechanisms, mechanisms, and and ecological ecological response response of of upwelling upwelling in in the the TWS TWS based based onon inin situsitu measurements, satellite satellite observations, observations, or or numerical numerical modelling modelling [16–22]. [16–22 The]. char- The characteristicsacteristics of several of several dominant dominant upwelling upwelling subregions subregions such such as the as Pingtan the Pingtan upwelling upwelling zone zone(PTU), (PTU), the Dongshan the Dongshan Upwelling Upwelling zone zone(DSU), (DSU), and the and Taiwan the Taiwan Bank Bankupwelling upwelling zone (TBU) zone (TBU)have havealso been also beeninvestigated. investigated. These These studies studies have have demonstrated demonstrated that that these these upwellings upwellings are areprincipally principally induced induced by by the the southwesterly southwesterly mo monsoonnsoon and and the ascending movementmovement ofof the the northward,northward, near-bottom near-bottom current current along along the the bottom bottom topography topography [ 14[14].]. Moreover, Moreover, the the size size or or intensityintensity of of upwelling-related upwelling-related low low temperature temperature zones zones have have substantial substantial variability, variability, and that and variationthat variation over short-term over short-term or interannual or interannual time scales time is significantly scales is significantly controlled by controlled alongshore by windalongshore stress [wind23–26 stress]. Furthermore, [23–26]. Furthermore, long-term changeslong-term in changes sea surface in sea temperature surface tempera- have beenture investigatedhave been investigated using long-term using time long-term series of time satellite-derived series of satellite-derived sea surface temperature sea surface datasetstemperature [27,28 datasets]. However, [27,28]. long-term However, changes long-term in summer changes upwelling in summer in the upwelling TWS have in not the beenTWS fully have studied not been and fully understood. studied and Detailed understood. information Detailed on the information influence ofon upwelling the influence on seaof upwelling surface temperature on sea surface and chlorophylltemperaturea andis particularly chlorophyll lacking. a is particularly lacking.
China ECS
SCS
FigureFigure 1. 1.Map Map ofof thethe studystudy area, overlaid with the the ba bathymetrythymetry contours contours in in meters. meters. TWS: TWS: Taiwan Taiwan Strait; ECS: East China Sea; SCS: South China Sea; Is.: Island. Strait; ECS: East China Sea; SCS: South China Sea; Is.: Island.
AsAs the the dominant dominant modes modes of of earth earth climate climate indices, indices, ENSO ENSO (El Niño(El Niño Southern Southern Oscillation) Oscilla- eventstion) events play an play important an important role in the role hydrodynamics in the hydrodynamics and ecosystems and ecosystems of many upwelling of many systemsupwelling [29 ,systems30]. Studies [29,30]. have Studies shown have that show theren that are twothere types are two of ENSOtypes of events. ENSO Oneevents. is theOne canonical is the canonical El Niño El event, Niño in event, which in anomalouswhich anomalous warming warming occurs inoccurs the easternin the eastern equa- torialequatorial Pacific Pacific Ocean. Ocean. Its variation Its variation can becan represented be represented by the by normalizedthe normalized Multivariate Multivariate El Niño/SouthernEl Niño/Southern Oscillation Oscillation Index Index (MEI) (MEI) [31 [31].]. Significantly Significantly positive positive (negative) (negative) values values of of thethe MEI MEI indicateindicate El Niño Niño (La (La Niña) Niña) conditions conditions.. Another Another variation variation is El is Niño El Niño Modoki, Modoki, char- characterizedacterized by a by warm a warm SST SSTanomaly anomaly in the in central the central equatorial equatorial Pacific. Pacific. The Thenormalized normalized El Niño El NiñoModoki Modoki index index (EMI) (EMI) is used is used to distinguish to distinguish the changes the changes of El of Niño El Niño Modoki Modoki events events [32]. [32 The]. Themore more positive positive the EMI, the EMI, the stronger the stronger the El the Ni Elño Niño Modoki. Modoki. Thus far, Thus research far, research on the on hydro- the hydrodynamicsdynamics and ecosystems and ecosystems of the of theTaiwan Taiwan Strait Strait byby canonical canonical ENSO ENSO events events has has received received a alot lot of of attention attention [28,33,34]. [28,33,34]. These These studies suggestsuggest that thethe delayeddelayed ENSOENSO effecteffect is is a a major major mechanismmechanism for for the the interannual interannual variability variability of hydrodynamicsof hydrodynamics in the in Taiwanthe Taiwan Strait. Strait. However, How- littleever, research little research has been has conducted been conducted on the impacton the impact of ENSO of ModokiENSO Modoki events onevents changes on changes in the ecologicalin the ecological environment environment of the TWS of the upwelling TWS upwelling system. system. In this study, we utilize available time series of sea surface temperature (SST), wind, and chlorophyll a to investigate the long-term changes in upwelling intensity in the Taiwan Strait. The questions addressed in this research include: How has the upwelling system of
Remote Sens. 2021, 13, 1386 3 of 16
In this study, we utilize available time series of sea surface temperature (SST), wind, and chlorophyll a to investigate the long-term changes in upwelling intensity in the Tai- wan Strait. The questions addressed in this research include: How has the upwelling sys- tem of the TWS responded to climate change over the past decades? Are there any differ- ences in the trends of upwelling intensity for the main three upwelling subregions includ- ing PTU, DSU, and TBU? Do ENSO Modoki events have any effects on upwelling? What are the influences of long-term changes in upwelling intensity on sea surface temperature and chlorophyll a? The answers to these questions are very important to our understand- ing of the dynamic process of upwelling as well as to the management and protection of
Remote Sens.fishery2021, 13 ,resources 1386 in the TWS in the context of global change. 3 of 16
2. Materials and Methods 2.1. SST Data the TWS responded to climate change over the past decades? Are there any differences in the trends of upwelling intensity for the main three upwelling subregions including PTU, The AVHRR DSU,Pathfinder and TBU? Version Do ENSO 5.3 Modoki (PFV5.3) events sea have surface any effects temperature on upwelling? (SST) What data are the from 1982 to 2019 wereinfluences obtained of long-term from changesthe US in National upwelling intensityOceanographic on sea surface Data temperature Center and and GHRSST (http://pathfinder.nodc.noaa.gov,chlorophyll a? The answers to these ac questionscessed areon very05 March important 2021). to our understandingThe PFV5.3 ofSST the dynamic process of upwelling as well as to the management and protection of fishery data are an updatedresources version in the of TWS the inPathfinder the context of Versions global change. 5.0 and 5.1 collection described in Saha et al. [35]. The PFV5.3 SST dataset is a collection of global, twice daily (day and night) 2. Materials and Methods 4-km sea surface temperature2.1. SST Data data. The validation of PFSST with a drifting buoy indicated that it agreed well withThe in AVHRR situ observations Pathfinder Version in 5.3the (PFV5.3) study sea region, surface with temperature a mean (SST) bias data ± fromstand- ard deviation (STD)1982 of to − 20190.33 were °C obtained± 0.79 °C from [36]. the USThe National PFSST Oceanographic dataset has Data been Center widely and GHRSST used for global or regional( http://pathfinder.nodc.noaa.govocean and climate studies ,[37]. accessed on 5 March 2021). The PFV5.3 SST data are an updated version of the Pathfinder Versions 5.0 and 5.1 collection described in Saha In this study,et the al. [night35]. The daily PFV5.3 SST SST data dataset were is a collection used to of calculate global, twice the daily monthly (day and SST night) through 4-km arithmetic averaging.sea surface Next temperaturethe summer data. average The validation SST was of PFSST calculated with a drifting from buoy the indicatedmonthly that data for June, July, andit August. agreed well Anomalies with in situ observationsin SST (SSTA) in the were study calculated region, with aas mean the biasdeviation± standard from deviation (STD) of −0.33 ◦C ± 0.79 ◦C[36]. The PFSST dataset has been widely used for the summer climatologicalglobal or regional means ocean between and climate 1982 studies and [ 372019.]. In this study, the night daily SST data were used to calculate the monthly SST through 2.2. Characterizationarithmetic of Upwe averaging.lling Based Next on the SST summer and average Wind SST was calculated from the monthly data for June, July, and August. Anomalies in SST (SSTA) were calculated as the deviation from In order betterthe to summer characterize climatological changes means in between the upwelling 1982 and 2019. zones of TWS, the difference in SST (ΔSST) between the TWS and non-upwelling zone were calculated. The ΔSST cal- 2.2. Characterization of Upwelling Based on SST and Wind culation was performed by subtracting the average SST of the non-upwelling zone from In order better to characterize changes in the upwelling zones of TWS, the difference the SST of the TWS.in SSTThe (∆ TWSSST) betweenrefers to the the TWS water and non-upwelling between 22–26° zone wereN and calculated. 116.5–121.5° The ∆SST E, as shown in Figure 1.calculation The non-upwelling was performed by zone subtracting refers the to average the offshore SST of the non-upwellingwater between zone from22–26° ◦ ◦ N and 124–124.5° theE (Figure SST of the 2a). TWS. The The subregions TWS refers to DSU, the water PTU, between and 22–26 TBUN are and also 116.5–121.5 illustratedE, as in shown in Figure1. The non-upwelling zone refers to the offshore water between 22–26 ◦N Figure 2a. and 124–124.5◦E (Figure2a). The subregions DSU, PTU, and TBU are also illustrated in Figure2a.
FigureFigure 2. The 2. The spatial spatial distribution distribution of the difference of the in diffe SST betweenrence in the SST TWS between and non− theupwelling TWS zoneand (nona) and−upwelling the SST gradient (b) of multi−year average summer sea surface temperature for 1982–2019 in the TWS. The calculation domain for zone (a) and the SST gradient (b) of multi−year average summer sea surface temperature for 1982– the Pingtan upwelling (PTU), Dongshan upwelling (DSU), Taiwan Bank upwelling (TBU), and non−upwelling zone (N) are illustrated2019 in as the red TWS. dashed The lines oncalculation (a). The overlaid domain white for contours the Pingtan represent aupwelling SST difference (PTU), of −2 ◦ C.Dongshan upwelling (DSU), Taiwan Bank upwelling (TBU), and non−upwelling zone (N) are illustrated as red dashed lines on (a). The overlaidUpwelling white contours intensity represent (R) is defined a SST following difference the methods of −2 °C. from Kuo et al. [38] and Shang et al. [24]: R = ∑ ∆Tidi Ai (1)
Remote Sens. 2021, 13, 1386 4 of 16
where Ai is the area of the upwelling region, ∆Ti is ∆SST, that is, the decrease in SST relative to the average SST of non-upwelling water, and di refers to the upwelling depth. Pond and Pickard [39] estimated di by 4.3Vi di = p (2) sin|ϕi|
where Vi is the wind speed and ϕi is the latitude of the upwelling zones. They are represented as the average wind speed and latitude of the PTU, DSU, and TBU zones. The ∆Ti Ai in the three upwelling zones was calculated pixel by pixel on waters colder than non-upwelling water by a threshold value. For the three upwelling subzones, the threshold value was chosen as 2.0 ◦C. Figure2b indicates that the SST difference of the −2.0 ◦C contour was consistent with the positions of the fronts between cold upwelling water and warm water.
2.3. Wind Data and Ekman Transport The monthly European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 global 0.25◦ × 0.25◦ reanalysis dataset was downloaded from the Asia-Pacific Data Research center (http://apdrc.soest.hawaii.edu/, accessed on 5 March 2021). ERA5 is the fifth generation ECMWF reanalysis for the global climate and weather, covering from 1979 to 2019. Two variables, the 10-m u wind component (Wx) and v wind component (Wy) were used in this study. The zonal and meridional components of wind stress (τx,τy) were calculated from the wind speed (W = (Wx,Wy)) according to Zhang et al. [28], as follows:
1/2 2 2 τx = ρaCd Wx + Wy Wx, (3)
1/2 2 2 τy = ρaCd Wx + Wy Wy (4)
−3 where ρa is the air density (1.22 kg m ) and Cd is the dimensionless drag coefficient from Trenberth et al. [40]: (0.49 + 0.065W) × 10−3 f or W > 10m s−1 −3 −1 Cd = 1.14 × 10 f or 3 ≤ W ≤ 10m s (5) 0.62 + 1.56W−1 f or W ≤ 3m s−1
The Ekman transport factors Qx and Qy were calculated as follows:
τy Qx = (6) ρw f
−τx Qy = (7) ρw f −3 where ρw is the density of sea water (1025 kg m ), f is the local Coriolis parameter (defined as twice the component of the angular velocity of the earth, Ω, at latitude θ, or f = 2Ω sin(θ), where Ω = 7.292 × 10−5 s−1). The cross-shore Ekman transport (Qcross), which also refers to upwelling index, is de- fined as the Ekman transport component in the direction perpendicular to the shoreline [41]. It was calculated as follows:
Qcross = Qx sin(ϕ) − Qy cos(ϕ) (8)
where ϕ is the mean angle between the shoreline and the equator. Positive values of Qcross indicate upwelling-favorable conditions, while negative values of Qcross indicate downwelling-favorable conditions [41]. Remote Sens. 2021, 13, 1386 5 of 16
2.4. Other Data Monthly level-3 chlorophyll a (Chl) data from the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the Aqua satellite (retrieved with the MODIS OC3M standard algorithm) were obtained from the Ocean Biology Processing Group (OBPG) at NASA’s Goddard Space Flight Center (http://oceancolor.gsfc.nasa.gov, accessed on 5 March 2021). The temporal coverage was between 2003 and 2019, and the spatial resolu- tion was 4 km. The numerical values of the monthly Multivariate ENSO (El Niño/Southern Os- cillation) Index (MEI), and El Niño Modoki index (EMI) were downloaded from Earth System Research Laboratory’s Physical Sciences Division (PSD) of National Oceanic and Atmospheric Administration (https://www.esrl.noaa.gov/psd/enso/mei/, accessed on 5 March 2021), and the Japan Agency for Marine-Earth Science and Technology (http://www.jamstec.go.jp/frsgc/research/d1/iod/modoki_home.html.en, accessed on 5 March 2021), respectively.
2.5. Statistical Method The Locally Weighted Scatterplot Smoother (LOWESS or LOESS) method was used in this study to detect non-linear trends. In some cases, linear regression cannot clarify the relationships between variables and cannot detect the trend of a data series. To overcome such limitations and provide qualitative results, the LOWESS method is found to be an effective regression approach for long-term environmental data with non-linear trends [42–44]. As a non-parametric smoothing procedure and locally weighted average model, the LOWESS method produces a smooth line through a scatterplot, which helps to understand the trends of variables [43]. Detailed information about this method can be found in studies by Cleveland [43] and Lee et al. [42].
3. Results The temporal variation of upwelling intensity (R) calculated based on SST and wind Remote Sens. 2021, 13, 1386 data in the three subregions of PTU, DSU, and TBU are shown in Figure3. The variability6 of 16 of R in the three subregions is almost the same. There were no significant differences between R of any two subregions using the independent t–test. 2000 1600 (a) LOWESS: r=0.34, p=0.028, n=38 C) o 1200 3 800 R(km 400 0 2000
(b) LOWESS: r=0.29, p=0.029, n=38 1600 C) o 1200 3 800 400 R(km 2000 0 1600 (c) LOWESS: r=0.35, p=0.021, n=38 C) o 1200 3 800 R(km 400 0 1980 1985 1990 1995 2000 2005 2010 2015 2020 Year FigureFigure 3. 3.Temporal Temporal variation variation in in upwelling upwelling intensity intensity (R (R)) from from 1982 1982 to to 2019 2019 for for the the Pingtan Pingtan upwelling (PTU,upwellinga), Dongshan (PTU, a), upwelling Dongshan (DSU, upwellingb), and (DSU, Taiwan b), Bank and Taiwan upwelling Bank (TBU, upwellingc). The (TBU, blue dotted c). The lines blue dotted lines and grey dashed lines represent the LOWESS fit lines and linear fit lines, respectively. and grey dashed lines represent the LOWESS fit lines and linear fit lines, respectively.
Statistical results show that the upwelling intensity decreased every year from 1982 to 2019 (Figure 3), but the linear trend of R did not pass the significance test. The LOWESS was then adopted to detect the non-linear trend of R over time. The regression fit lines from LOWESS are overlaid on Figure 3, as shown by the dotted lines. All of these regres- sion fit lines pass the significance test (p < 0.05). Figure 3 shows that the changes in R varied with a non-linear fluctuation over the period 1982–2019. The R increased slightly before 1996 and during 2003–2007, and decreased from 1996 to 2003 and after 2007. Four different periods during the 1980s (1982–1989), 1990s (1990–1999), 2000s (2000– 2009), and 2010s (2010–2019) were chosen to further investigate variation in upwelling intensity (Figure 4). Figure 4 indicates that the upwelling intensity in the three sub regions was the lowest in the 2000s, and the R decreased significantly from the 1990s to 2000s. From the 2000s to the 2010s, the R of the three subregions increased. However, the average value of R in the 2010s was much smaller than the R before 2000. We roughly separated the time series data into two periods separated by the year 2000. Table 1 shows the statis- tical results in changes of R before 2000 and after 2000. On average, the upwelling intensity (R) after 2000 was 35% less than that before 2000. Among the three subregions, the de- crease of R in PTU is the most obvious, with R decreasing by 46%.
1200 PTU DSU TBU 800 C) o
3 R(km 400
0 1982- 1989 1990- 1999 2000- 2009 2010- 2019 Period Figure 4. Variation in upwelling intensity (R) over four different periods.
Remote Sens. 2021, 13, 1386 6 of 16
2000 1600 (a) LOWESS: r=0.34, p=0.028, n=38 C) o 1200 3 800 R(km 400 0 2000
(b) LOWESS: r=0.29, p=0.029, n=38 1600 C) o 1200 3 800 400 R(km 2000 0 1600 (c) LOWESS: r=0.35, p=0.021, n=38 C) o 1200 3 800 R(km 400 0 1980 1985 1990 1995 2000 2005 2010 2015 2020 Year Figure 3. Temporal variation in upwelling intensity (R) from 1982 to 2019 for the Pingtan upwelling (PTU, a), Dongshan upwelling (DSU, b), and Taiwan Bank upwelling (TBU, c). The blue dotted lines and grey dashed lines represent the LOWESS fit lines and linear fit lines, respectively.
Statistical results show that the upwelling intensity decreased every year from 1982 to 2019 (Figure 3), but the linear trend of R did not pass the significance test. The LOWESS Remote Sens. 2021, 13, 1386 6 of 16 was then adopted to detect the non-linear trend of R over time. The regression fit lines from LOWESS are overlaid on Figure 3, as shown by the dotted lines. All of these regres- sionStatistical fit lines results pass show the that significance the upwelling test intensity (p < decreased 0.05). everyFigure year 3 from shows 1982 tothat the changes in R 2019varied (Figure with3), buta non-linear the linear trend fluc oftuationR did not over pass thethe significance period 1982–2019. test. The LOWESS The R increased slightly wasbefore then 1996 adopted and to detect during the non-linear2003–2007, trend and of R overdecreased time. The from regression 1996 fit linesto 2003 from and after 2007. LOWESS are overlaid on Figure3, as shown by the dotted lines. All of these regression fit lines passFour the different significance periods test (p < 0.05). during Figure the3 shows 1980s that (1982–1989), the changes in 1990sR varied (1990–1999), with 2000s (2000– a2009), non-linear and fluctuation 2010s (2010–2019) over the period were 1982–2019. chosen The Rtoincreased further slightly investigate before 1996 variation in upwelling andintensity during 2003–2007,(Figure 4). and Figure decreased 4 indicates from 1996 to that 2003 the and upwelling after 2007. intensity in the three sub regions Four different periods during the 1980s (1982–1989), 1990s (1990–1999), 2000s (2000– 2009),was the and 2010slowest (2010–2019) in the 2000s, were chosen and tothe further R decreased investigate variationsignificantly in upwelling from the 1990s to 2000s. intensityFrom the (Figure 2000s4). Figure to the4 indicates 2010s, the that R the of upwelling the three intensity subregions in the three increased. sub regions However, the average wasvalue the lowestof R in in thethe 2000s, 2010s and was the R muchdecreased smaller significantly than from the the R 1990s before to 2000s. 2000. From We roughly separated the 2000s to the 2010s, the R of the three subregions increased. However, the average value oftheR intime the 2010sseries was data much into smaller two than periods the R before separated 2000. We by roughly the year separated 2000. the Table time 1 shows the statis- seriestical dataresults into twoin changes periods separated of R before by the 2000 year 2000. and Table after1 shows 2000. the On statistical average, results the upwelling intensity in(R) changes after of2000R before was 2000 35% and less after than 2000. Onthat average, before the 2000. upwelling Among intensity the (R )three after subregions, the de- 2000 was 35% less than that before 2000. Among the three subregions, the decrease of R in PTUcrease is the of most R in obvious, PTU is with theR mostdecreasing obvious, by 46%. with R decreasing by 46%.
1200 PTU DSU TBU 800 C) o
3 R(km 400
0 1982- 1989 1990- 1999 2000- 2009 2010- 2019 Period Figure 4. Variation4. Variation in upwelling in upwelling intensity (Rintensity) over four (R) different over periods. four different periods.
Table 1. Statistical results in the change of R and Qcross before 2000 and after 2000.
RQ cross PTU DSU TBU PTU DSU TBU 1982–1999 610.0 ± 480.0 559.5 ± 461.3 564.8 ± 427.9 256 ± 160.3 174.4 ± 94.9 220.8 ± 87.0 2000–2019 326.1 ± 250.2 433.2 ± 329.3 356.6 ± 276.6 185.9 ± 84.3 149.0 ± 55.3 187.1 ± 52.8 Change rate −46% −22% −37% −27% −14% −15%
Similar variations were also found in the cross-shore Ekman transport (Qcross, Figure5) using linear and LOWESS method. During the study period, there was a slight increase in Qcross before 1996, then a decrease during 1996–2003. After 2003, the Qcross increased during 2003–2007, and slightly decreased after 2007. The average value of Qcross after 2000 was much lower than before 2000 (Table1), especially for the PTU. Remote Sens. 2021, 13, 1386 7 of 16
Table 1. Statistical results in the change of R and Qcross before 2000 and after 2000.
R Qcross PTU DSU TBU PTU DSU TBU 1982–1999 610.0 ± 480.0 559.5 ± 461.3 564.8 ± 427.9 256 ± 160.3 174.4 ± 94.9 220.8 ± 87.0 2000–2019 326.1 ± 250.2 433.2 ± 329.3 356.6 ± 276.6 185.9 ± 84.3 149.0 ± 55.3 187.1 ± 52.8 Change rate −46% −22% −37% −27% −14% −15%
Similar variations were also found in the cross-shore Ekman transport (Qcross, Figure 5) using linear and LOWESS method. During the study period, there was a slight increase
Remote Sens. 2021, 13, 1386 in Qcross before 1996, then a decrease during 1996–2003. After 2003, the Qcross increased7 of 16 dur- ing 2003–2007, and slightly decreased after 2007. The average value of Qcross after 2000 was much lower than before 2000 (Table 1), especially for the PTU.
600 )
-1 500 (a) LOWESS: r=0.33, p=0.018, n=38
km 400 -1 -1 s
3 300
(m 200
cross 100 Q 0 400 )
LOWESS: r=0.33, p<0.01, n=38 -1 (b) 300 km -1 -1 s
200 3 (m 100 cross Q 400 0 ) -1 (c) LOWESS: r=0.36, p<0.01, n=38 300 km -1 -1 s
3 200 (m 100 cross Q 0 1980 1985 1990 1995 2000 2005 2010 2015 2020 Year
FigureFigure 5. 5.Temporal Temporal variation variation of cross of cross−shore−shore Ekman Ekman transport transport (Qcross) (Q fromcross 1982) from to 20191982 forto PTU2019 (fora), PTU (a), DSUDSU ( b(b),), and and TBU TBU (c). (c The). The blue blue dotted dotted lines lines and greyand grey dashed dashed liens represent liens represent the LOWESS the LOWESS fit lines fit lines andand linear linear fit fit lines, lines, respectively. respectively.
4. Discussion 4. Discussion 4.1. Changes of Upwelling Intensity 4.1. Changes of Upwelling Intensity Figures3 and4, and Table1, show that from 1982 to 2019, the upwelling intensity in theFigures Taiwan 3 Strait and varied4, and withTable a 1, nonlinear show that fluctuation. from 1982 Although to 2019, thethe linearupwelling trend intensity of in decreasingthe TaiwanR didStrait not varied pass the with significance a nonlinear test, the fluctuation. average upwelling Although intensity the linear after 2000trend of de- wascreasing 35% lower R did than not that pass before the significance 2000 (Table1). test Similar, the upwellingaverage upwelling weakening intensity was found after 2000 alongwas 35% the Hainan lower and than Zhejiang that before coast of2000 the China(Table Sea 1). [ 45Similar,46]. Liu upwelling et al. (2013) weakening demonstrated was found thatalong since the the Hainan 1960s, the and upwelling Zhejiang intensity coast inof thethe northern China Sea South [45,46]. China SeaLiu has et decreased.al. (2013) demon- Other upwelling systems, such as the Iberian coastal upwelling [47] and low latitude strated that since the 1960s, the upwelling intensity in the northern South China Sea has Canary upwelling system [7], are reported to have a weakening in upwelling intensity. The decreasingdecreased. upwelling Other upwelling trend seen systems, in these areassuch isas quitethe Iberian different coastal from the upwelling north eastern [47] and low boundarylatitude Canary upwelling upwelling systems, whichsystem have [7], aare strengthening reported to of have upwelling a weakening [2,48]. in upwelling in- tensity.Figures The3 and decreasing5 indicate thatupwelling the changes trend in upwellingseen in these intensity areas of PTU,is quite DSU, different and TBU from the arenorth consistent eastern with boundary changes inupwelling cross-shore systems, Ekman transport. which have Further a statisticalstrengthening correlations of upwelling (Figure[2,48].6 ) show that there are significant correlation between R and Qcross in the three subregionsFigures (p ≤ 30.05), and especially5 indicate in that the PTU the andchanges DSU, whichin upwelling suggests intensity that Ekman of transport PTU, DSU, and isTBU the are dominant consistent mechanism with changes pumping in cross-shor cold watere upwardEkman transport. to the surface Further in the statistical TWS. corre- Several studies have also demonstrated that short-term or interannual variability of coastal lations (Figure 6) show that there are significant correlation between R and Qcross in the upwelling is closely correlated with changes in alongshore wind stress in the TWS [23,26]. threeIn subregions recent decades, (p the≤ 0.05), East Asian especially summer in monsoon the PTU has and been DSU, observed which to weaken suggests [49 ,that50]. Ekman Figuretransport5 and is Table the dominant1 reveal that mechanism the upwelling-favorable pumping cold Ekman water transport upward ofto the the three surface in the subregions in the TWS decreased over the last 40 years. Therefore, the decreasing along- shore wind might have contributed greatly to the weakening of the upwelling intensity after 2000. Remote Sens. 2021, 13, 1386 8 of 16
Remote Sens. 2021, 13, 1386 8 of 16 TWS. Several studies have also demonstrated that short-term or interannual variability of coastal upwelling is closely correlated with changes in alongshore wind stress in the TWS [23,26]. 2000 PTU: y=2.37*x- 57.66(n=38,r=0.77,p<0.0001) 2000 DSU: y=3.73*x- 108.81(n=38,r=0.72,p<0.0001)
1500 1500 C) C) o
o
3 3 1000 1000 R(km R(km
500 500
0 0 0 100 200 300 400 500 600 0 100 200 300 400 Q (m3 s-1 km-1) 3 -1 -1 cross Qcross(m s km ) (a) (b) 2500 TBU: y=2.06*x+37.51(n=38,r=0.41,p=0.011)
2000
C) 1500 o
3
R (km 1000
500
0 0 100 200 300 400 3 -1 -1 Qcross(m s km ) (c)
Figure 6. RelationshipFigure 6. Relationship between the between cross− theshore cross Ekman−shore transport Ekman (transportQcross) and (Q upwellingcross) and upwelling intensity (intensityR) in PTU (a), DSU (b), and TBU (c(R)). in PTU (a), DSU (b), and TBU (c).
In recent decades,4.2. Difference the East of Upwelling Asian summer Systems monsoon between PTU,has been DSU, observed and TBU to weaken [49,50]. Figure 5 and Table 1 reveal that the upwelling-favorable Ekman transport of the There is no significant difference in the trend of upwelling intensity in the three three subregions in the TWS decreased over the last 40 years. Therefore, the decreasing subregions (Figure3). However, some differences still exist in their inter-annual changes. alongshore windFor might example, have thecontributed highest R greaof PTUtly to occurred the weakening in 1998, of while the upwelling the maximums inten- of DSU and TBU sity after 2000. were in 1996. The upwelling intensity of PTU and DSU is more significantly affected by the wind than is TBU (Figure6). Moreover, the decrease of R in PTU after 2000 was much 4.2. Difference ofstronger Upwelling than Syst inems DSU between and TBU PTU, (Table DSU,1 ).and TBU There is no significantPrevious difference findings in have the trend shown of theupwelling different intensity short-term in the and three inter-annual sub- variation regions (Figure patterns3). However, of coastal some upwelling differences in still the northexist in and their south inter-annual part of the changes. western TWSFor [24,26]. These example, the higheststudies R suggestof PTU thatoccurred the upwelled in 1998, while water the of thesemaximums two upwelling of DSU and regions TBU comes from the were in 1996. Theadjacent upwelling East Chinaintensity Sea of and PTU the and South DSU China is more Sea. significantly The possible affected reasons by for the different the wind than isbehavior TBU (Figure of upwelling 6). Moreover, in the the three decrease subregions of R in PTU are supposedafter 2000 towas be much related to differing stronger than insources DSU and of upwelledTBU (Table water 1). or the control mechanism. Previous findingsUpwelling-favorable have shown the differen local windt short-term and topography and inter-annual are two key variation factors that drive the patterns of coastalupwelling upwelling in in the the Northern north and South south China part of Sea the [western51]. Topography TWS [24,26]. is theThese main mechanism studies suggest forthat triggering the upwelled the upwellingwater of these of the two Taiwan upwelling Bank regions [14]. However, comes from Figure the6 indicates that at a long-term time scale, the changes in cross-shore Ekman transport might play some role in the changes of upwelling intensity of TBU (r = 0.41 p = 0.011). For the PTU and
Remote Sens. 2021, 13, 1386 9 of 16
DSU, the southwest wind has been suggested to be an important factor for controlling the variation in upwelling intensity [13,15]. The simulation results based on a numerical model also reveal that the southerly wind plays a key role in shaping the upwelling strip in the PTU, while the local coastline geometry and topography are responsible for the spatial distribution of upwelling cold core [18]. It should be noted that there are unavoidable biases in the upwelling intensity ex- pressed by SST. Sometimes the TWS is affected by summer typhoons [52] that cause cold sea surface temperature. Furthermore, satellite-derived SST contains a certain degree of deviation because of clouds or the algorithm. However, the R comprehensively considers the influencing factors of SST and wind on upwelling [38], and the variation of upwelling indicated by R is very similar to that of Qcross. Therefore, the R is a valuable index for characterizing upwelling intensity in the TWS in summer.
4.3. Relationship with Canonical ENSO and ENSO Modoki Canonical ENSO events can affect the East Asia Monsoon through atmosphere-SST interaction and midlatitude-tropical interaction in the western North Pacific [53]. Previous studies have found that canonical El Niño events caused a stronger southwesterly wind in summer and weaker northeasterly wind in winter in the northwestern Pacific [53,54]. Kuo and Ho [34] and Shang et al. [33] suggest that the canonical ENSO events can affect wind patterns in the TWS, and therefore modulate the sea surface currents to result in SST changes. Hong et al. [26] showed that a delayed ENSO effect was a major mechanism in the TWS upwelling system. Our results show the same tendency. Both R and Qcross are highly correlated with MEI at a time lag of 3 to 12 months (Table2). It seems probable that the TWS upwelling system is modulated by other forces, such as the East Asian monsoon, and it may be overwhelmed by delayed tropical Pacific ENSO signals when the ENSO signals are strong enough [26,55].
Table 2. Lag correlation coefficients between Multivariate ENSO (El Niño/Southern Oscillation) Index (MEI) and R and Qcross, respectively.
Lag(month) R(p) Qcross(p) 0 −0.205(0.216) −0.218(0.187) −3 0.430(<0.01) 0.331(<0.05) −6 0.470(<0.01) 0.330(<0.05) −9 0.451(<0.01) 0.312(0.056) −12 0.526(<0.01) 0.443(<0.05)
Unlike the delay effect of canonical ENSO events on the changes in summer upwelling in the TWS (Table2), a negative linear correlation exists in the relationship between R and Qcross with EMI (Figure7). The correlation coefficient for R/EMI is −0.32 (p = 0.05) and for Qcross/EMI, it is −0.5 (p < 0.01). This result reveals that the El Niño Modoki has a more direct and much stronger impact on the upwelling system of TWS than the canonical El Niño has. During the positive phase of El Niño Modoki, a warm SST anomaly shifted from the east equatorial Pacific to the central equatorial Pacific, which suggests the associated low-level anomalous wind pattern is more distinct and closer to the Asian continent than the canonical ENSO events [56]. Therefore, El Niño Modoki might have a more significant influence on the East Asian summer climate than the canonical El Niño [56]. The East Asian summer monsoon tends to be weaker than normal during strong El Niño Modoki events [57,58] when northeasterly wind anomalies prevail over South China [59]. Strong El Niño Modoki events weaken offshore Ekman transport and are unfavorable for upwelling development. However, the situation is different in the case of La Niña Modoki, when the normal prevailing southwesterly winds are enhanced and the upwelling intensity increases. Remote Sens. 2021, 13, 1386 10 of 16
associated low-level anomalous wind pattern is more distinct and closer to the Asian con- Remote Sens. 2021, 13, 1386 10 of 16 tinent than the canonical ENSO events [56]. Therefore, El Niño Modoki might have a more significant influence on the East Asian summer climate than the canonical El Niño [56].
2000 R
) Qcross
-1 1600 km -1 s 3 1200 (m cross 800 C)/Q o
3
R(km 400
0 -1.5 -1 -0.5 0 0.5 1 EMI
FigureFigure 7.7.Relationship Relationship between between upwelling upwelling intensity intensity (R), (R),Qcross Q,cross and, and EMI EMI for the for coastal the coastal upwelling upwelling systemsystem of of the the Taiwan Taiwan Strait. Strait.
ToThe further East investigateAsian summer the impact monsoon of ENSO tends Modoki to be eventsweaker on than changes normal in upwelling during strong El intensities,Niño Modoki we definedevents [57,58] strong Elwhen Niño northeaste Modoki yearsrly wind as the anomalies years with prevail average over summer South China EMI > 0.5 and strong La Niña Modoki years as the years with average summer EMI < −0.5. [59]. Strong El Niño Modoki events weaken offshore Ekman transport and are unfavora- Between 1982 and 2019, the strong El Niño Modoki years included 1994, 2002, and 2004; the strongble for La upwelling Niña Modoki development. years included However, 1983, 1997, the 1998, situ 1999,ation andis different 2008. Statistics in the revealed case of La Niña thatModoki, the average when upwelling the normal intensity prevailing (R) and Qsocrossuthwesterlyin strong La winds Niña Modoki are enhanced years were and the 5.2upwelling and 2.2 times intensity higher increases. than they were in strong El Niño Modoki years. Overall,To further whether investigate considering the impact canonical of ElENSO Niño Modoki or El Niño events Modoki, on thechanges warm in and upwelling coldintensities, anomaly we events defined associated strong withEl Niño these Mo eventsdoki haveyears played as the important years with roles average in the summer interannualEMI > 0.5 and or long-termstrong La changesNiña Modoki of coastal years upwelling as the years in the with TWS. average Some researchsummer has EMI < −0.5. pointedBetween out 1982 that and the upwelling2019, the intensitystrong El has Niño different Modoki response years characteristics included 1994, to different 2002, and 2004; El Niño events [29]. Further quantitative analyses are needed to clarify the influence of the strong La Niña Modoki years included 1983, 1997, 1998, 1999, and 2008. Statistics re- ENSO events at their different stages. We do know that the occurrence of El Niño Modoki tendsvealed to that increase the underaverage global upwelling warming intensity [60]. Their (R) impacts and Q oncross the in circulation strong La or Niña ecosystem Modoki years ofwere the TWS5.2 and should 2.2 times be examined higher inthan more they detail were in future in strong studies. El Niño Modoki years. Overall, whether considering canonical El Niño or El Niño Modoki, the warm and 4.4.cold Impacts anomaly on SST events associated with these events have played important roles in the in- terannualPrevious or studieslong-term have changes pointed outof coastal that upwelling upwelling is a keyin the factor TWS. buffering Some oceanresearch has warmingpointed [out61,62 that]. A the general upwelling warming intensity was observed has different at most oceanic response and characteristics coastal locations to in different recentEl Niño decades events [63 ,[29].64]. However,Further warmingquantitative has been anal moreyses intenseare needed in offshore to clarify than nearshore the influence of watersENSO inevents most ofat thetheir upwelling different locations stages. [We61,65 do,66 know]. Researchers that the haveoccurrence demonstrated of El Niño that Modoki enhancing upwelling could hinder the warming due to the continuous pumping of water tends to increase under global warming [60]. Their impacts on the circulation or ecosystem from intermediate layers. In this study, the negative linear relationship between SSTA and upwellingof the TWS intensity should (Figure be examined8) in upwelling in more region detail of in TWS future also studies. indicated that increased upwelling could result in a decrease in SST. 4.4. Impacts on SST Previous studies have pointed out that upwelling is a key factor buffering ocean warming [61,62]. A general warming was observed at most oceanic and coastal locations in recent decades [63,64]. However, warming has been more intense in offshore than near- shore waters in most of the upwelling locations [61,65,66]. Researchers have demonstrated that enhancing upwelling could hinder the warming due to the continuous pumping of
Remote Sens. 2021, 13, 1386 11 of 16
Remote Sens. 2021, 13, 1386 11 of 16
water from intermediate layers. In this study, the negative linear relationship between Remote Sens. 2021, 13, 1386 water from intermediate layers. In this study, the negative linear relationship between11 of 16 SSTASSTA and and upwelling upwelling intensity intensity (Figure 8) in (Figure upwelling 8) region in up of wellingTWS also indicated region thatof TWS also indicated that increasedincreased upwelling upwelling could result could in a decreaseresult inin SST.a decrease in SST.
Figure 8. Relationship between upwelling intensity (R) and SSTA for the coastal upwelling system of the Taiwan Strait.
Figure 9 shows that SST increased significantly in the TWS coastal upwelling system (PTU and DSU) from 1982 to 2019. Statistics show that the warming rate of the TWS upwelling region was 0.27 °C/decade, and for the non-upwelling region it was 0.23 °C/decade. The warming rate of the coastal upwelling region was larger than for the non- FigureFigure 8. Relationship 8. Relationship between between upwelling intensityupwelling (R) and intensity SSTA for the(R) coastal and upwellingSSTA for system the coastal of upwelling system upwelling region. This finding is in contrast to observations from other upwelling regions the Taiwan Strait. suchof the as Benguela,Taiwan Strait.Canary, or California, where a moderate coastal warming or cooling was Figurelinked 9to shows the intensification that SST increased of coastal significantly upwelling [67,68]. in the TWS coastal upwelling sys- tem (PTUThe underlying and DSU) mechanism from 1982 for to SST 2019. warming Statistics was show complicated that the and warming should ratebe ana- of the Figure 9 shows that SST◦ increased significantly in the TWS coastal upwelling system TWSlyzedupwelling in future research region was involving 0.27 C/decade,numerical andmodeling. for the Nevertheless, non-upwelling the regionweakened it was ◦ 0.23upwelling(PTUC/decade. and after DSU)2000 The in warming thefrom TWS rate1982 (Figure of theto 3 and coastal2019. Table upwellingStatistics 1) could not region showslow was down that larger the the thanrise rate forwarming the rate of the TWS non-upwellingofupwelling SST, possibly region.regiondue to the This was reducing finding 0.27 vertical is °C/decade, in contrast transport to associated observationsand for with the from upwelling. non-upwelling other upwellingOn the region it was 0.23 regionsother°C/decade. hand, such the as warmingThe Benguela, warming could Canary, also rate produce or California, of thea deepening coastal where of a upthe moderate wellingupper mixed coastal region layer warming and was a larger or than for the non- coolingstronger was stratification, linked to thewhich intensification in turn reduce of the coastal upwelling upwelling intensity. [67,68 ]. upwelling region. This finding is in contrast to observations from other upwelling regions such1.2 as Benguela, Canary, or California, where a moderate coastal warming or cooling 0.8 (a)
wasC) 0.4 linked to the intensification of coastal upwelling [67,68]. o 0 The underlying mechanism for SST warming was complicated and should be ana-
SSTA( -0.4 lyzed-0.8 in future research involving numerical modeling. Nevertheless, the weakened y=0.027*x- 54.2(r=0.66,p<0.0001,n=38) upwelling-1.2 after 2000 in the TWS (Figure 3 and Table1.5 1) could not slow down the rise rate (b) 1
of SST, possibly due to the reducing vertical transport0.5 associatedC) with upwelling. On the o other hand, the warming could also produce a deepening0 of the upper mixed layer and a
-0.5 SSTA( stronger stratification, whichy=0.023*x- in turn 45.51(r=0.52,p<0.001,n=38) reduce the upwelling-1 intensity. -1.5 1.2 1985 1990 1995 2000 2005 2010 2015 Year 0.8 (a) FigureFigure 9.9. Variation in in SSTA SSTA for for coastal coastal upwelling upwelling region region (a) ( aand) and non non−upwelling−upwelling region region (b) during (b) during
C) 0.4 1982–2019.1982–2019.o 0 The underlying mechanism for SST warming was complicated and should be analyzed
SSTA( -0.4 in future research involving numerical modeling. Nevertheless, the weakened upwelling -0.8 after 2000 in the TWS (Figure3 and Table1) couldy=0.027*x- not slow 54.2(r=0.66,p<0.0001,n=38) down the rise rate of SST, possibly-1.2 due to the reducing vertical transport associated with upwelling. On the other 1.5 hand, the warming(b) could also produce a deepening of the upper mixed layer and a stronger 1 stratification, which in turn reduce the upwelling intensity.
0.5 C) o 0
-0.5 SSTA( y=0.023*x- 45.51(r=0.52,p<0.001,n=38) -1 -1.5 1985 1990 1995 2000 2005 2010 2015 Year Figure 9. Variation in SSTA for coastal upwelling region (a) and non−upwelling region (b) during 1982–2019.
RemoteRemote Sens. Sens.2021 2021, 13, 13, 1386, 1386 1212 of of 16 16
4.5.4.5. ImplicationsImplications forfor thethe UpwellingUpwelling EcosystemEcosystem IncreasedIncreased or decreased decreased upwelling upwelling has has profound profound implications implications for forbiogeochemical biogeochemical and andphytoplankton phytoplankton patterns patterns that thataffect affect the health the health of marine of marine ecosystems ecosystems [48,69–71]. [48,69 We–71 ].calcu- We calculatedlated the average the average Chl and Chl the and area the of area high of highChl (Chl Chl ≥ (Chl 1 mg/m≥ 1 mg/m3) over 3coastal) over coastalareas such areas as suchPTU asand PTU DSU, and then DSU, investigated then investigated their relationship their relationship with variations with variations in upwelling in upwelling intensity, intensity,as shown as in shown Figure in10. Figure 10.
2.4 30,000 ) 3 2 25,000 ) 2
1.6 20,000 HChl(km
1.2 15,000 Chl concentration(mg/m Chl
Chl conentration HChl
0.8 10,000 0 200 400 600 800 1000 R(km3 oC)
FigureFigure 10.10.Relationship Relationship betweenbetween upwellingupwelling intensityintensity ((R)R) andand chlorophyllchlorophyll aa (Chl)(Chl) concentrationconcentration andand the area of high Chl (HChl) for the coastal upwelling system of the Taiwan Strait. the area of high Chl (HChl) for the coastal upwelling system of the Taiwan Strait.
TheThe weakweak positive correlation correlation between between upwelling upwelling intensity intensity (R) (R and) and Chl Chl concentration concentra- tion(Figure (Figure 10) indicates 10) indicates that strong that strong upwelling upwelling can enhance can enhance phytoplankton phytoplankton growth growth by increas- by increasinging the nutrient the nutrient supply supplyfrom su frombsurface subsurface layers. The layers. correlation The correlation coefficients coefficients (r) were 0.37 (r) were(p = 0.14, 0.37 n (p == 17) 0.14, forn Chl= 17) concentration for Chl concentration and 0.47 (p and = 0.057, 0.47 (np == 17) 0.057, for nhigh= 17) Chl for area. high These Chl area.results These are resultsconsistent are consistentwith other with studies other from studies many from coastal many coastalupwelling upwelling regions, regions, which whichdemonstrate demonstrate that the that phytoplankton the phytoplankton production production can be modified can be modified by local bysupply local of supply nutri- ofents nutrients through through coastal coastal upwelling upwelling [72,73]. [72 The,73 ].weakened The weakened upwelling upwelling in the in TWS the TWSafter after2000 2000(Figure (Figure 3 and3 and Table Table 1) should1) should be benot not favora favorableble for for phytoplankton phytoplankton growth. growth. However, However, be- becausecause the the MODIS MODIS ocean ocean color color data data are are only only available available after 2002, we cannotcannot comparecompare thethe variationvariation inin ChlChl concentrationconcentration beforebefore andand afterafter 2000.2000. Therefore,Therefore, maintainingmaintaining aa longlong timetime seriesseries ofof consistentconsistent ocean-colorocean-color productsproducts isis veryvery importantimportant forfor evaluatingevaluating thethe impactimpact ofof climateclimate changechange onon biologicalbiological fields.fields. CoastalCoastal upwellingupwelling regionsregions are are “hot “hot spots” spots” with with high high primary primary production production [ 74[74,75].,75]. The The alterationalteration ofof phytoplanktonphytoplankton communitycommunity compositioncomposition waswas observedobserved totobe beassociated associated with with differentdifferent phasesphases ofof upwellingupwelling eventsevents [[76,77].76,77]. InIn thethe southernsouthern upwellingupwelling systemsystem ofof TWS,TWS, diatomsdiatoms dominatedominate inin thethe recentlyrecently upwelledupwelled andand maturemature water.water. AsAs thethe upwelledupwelled waterwater ages,ages, thethe community composition shifts shifts rapidly rapidly to to a acoexistence coexistence of ofdiatoms diatoms and and Synecho-Syne- chococcuscoccus [78].[78 Based]. Based on on this this pattern, pattern, changes changes in in coastal coastal upwelling associatedassociated withwith climateclimate changechange could could impact impact the the base base of of food food chain, chain, then then profoundly profoundly alter alter the the functioning functioning of marine of ma- ecosystems.rine ecosystems. The Taiwan The Taiwan Strait isStrait one ofis theone important of the important areas in southeastareas in southeast China for China fisheries for productionfisheries production [13]. Minor [13]. changes Minor inchanges local water in local properties water properties might significantly might significantly affect local af- fisheryfect local economics. fishery economics. Therefore, Therefore, understanding understanding the weakening/enhancing the weakening/enhancing trend of trend coastal of upwellingcoastal upwelling and the responseand the response of biological of biolog communitiesical communities and their and ecology their willecology provide will keypro- informationvide key information for predicting for predicting how marine how ecosystem marine ecosystem could react could to future react climate to future change. climate 5.change. Conclusions
Long-term changes in summer upwelling intensities in the Taiwan Strait were investi- gated using time series data on sea surface temperature and wind from 1982 to 2019. The
Remote Sens. 2021, 13, 1386 13 of 16
results reveal that the summer upwelling of the Taiwan Strait showed a decreased trend with a nonlinear fluctuation over the past four decades. The average upwelling intensity after 2000 was 35% lower than that before 2000. The long-term changes in upwelling intensity are closely correlated with the changes in offshore Ekman transport (Qcross). The offshore Ekman transport also experienced a declined trend after 2000. There was no significant difference in the trend of upwelling intensity in the three subregions of Pingtan upwelling, Dongshan upwelling, and Taiwan Bank upwelling zones. Unlike the delay effect of canonical ENSO events on the changes in summer upwelling, the ENSO Modoki index had a significant negative correlation with upwelling intensity of TWS. Statistics indicated that the average upwelling intensity (R) and Qcross in strong La Niña Modoki years were 5.2 and 2.2 times higher than they were in strong El Niño Modoki years. Strong El Niño Modoki events were not favorable for the development of upwelling. Moreover, the decreased upwelling did not slow down the warming rate of sea surface temperature and could cause the decline of chlorophyll a in the coastal upwelling system of TWS. SST and winds are important indicators for understanding variations in upwelling and the potential ecological response. Although changes in tide, stratification, currents, or other processes can contribute to the changes of upwelling, some of their effects in climate change scenarios are difficult to access or predict [66]. Thus, although more research is necessary, studies such as this one are a starting point for evaluating future trends in coastal upwelling and their potential ecological effects in the Taiwan Strait.
Funding: This research was funded by the National Key Research and Development Plan of China (Grant No. 2016YFE0202100) and the Key Program of NSF-China (Grant No. U1805241). Data Availability Statement: Publicly available data sets were analyzed in this study. AVHRR sea surface temperature data can be found here: http://pathfinder.nodc.noaa.gov (accessed on 5 March 2021). ERA5 wind data can be found here: http://apdrc.soest.hawaii.edu/ (accessed on 5 March 2021). MODIS chlorophyll a data can be found here: http://oceancolor.gsfc.nasa.gov (accessed on 5 March 2021). Acknowledgments: Anonymous reviewers contributed substantially to the improvement of this paper. Conflicts of Interest: The author declares no conflict of interest.
References 1. Chavez, F.; Pennington, J.; Castro, C.; Ryan, J.; Michisaki, R.; Schlining, B.; Walz, P.; Buck, K.; McFadyen, A.; Collins, C. Biological and chemical consequences of the 1997–1998 el niño in central california waters. Prog. Oceanogr. 2002, 54, 205–232. [CrossRef] 2. Bakun, A.; Field, D.B.; Redondo-Rodriguez, A.; Weeks, S.J. Greenhouse gas, upwelling-favorable winds, and the future of coastal ocean upwelling ecosystems. J. Glob. Chang. Biol. 2010, 16, 1213–1228. [CrossRef] 3. Bakun, A. Global climate change and intensification of coastal ocean upwelling. Science 1990, 247, 198–201. [CrossRef] 4. McGregor, H.; Dima, M.; Fischer, H.W.; Mulitza, S. Rapid 20th-century increase in coastal upwelling off northwest africa. Science 2007, 315, 637–639. [CrossRef] 5. Sydeman, W.; García-Reyes, M.; Schoeman, D.; Rykaczewski, R.; Thompson, S.; Black, B.A.; Bograd, S. Climate change and wind intensification in coastal upwelling ecosystems. Science 2014, 345, 77–80. [CrossRef] 6. Sousa, M.C.; Ribeiro, A.; Des, M.; Gomez-Gesteira, M.; de Castro, M.; Dias, J.M. Nw iberian peninsula coastal upwelling future weakening: Competition between wind intensification and surface heating. Sci. Total. Environ. 2020, 703, 134808. [CrossRef] [PubMed] 7. Wang, D.; Gouhier, T.C.; Menge, B.A.; Ganguly, A.R. Intensification and spatial homogenization of coastal upwelling under climate change. Nature 2015, 518, 390–394. [CrossRef] 8. Rykaczewski, R.R.; Dunne, J.P.; Sydeman, W.J.; García-Reyes, M.; Black, B.A.; Bograd, S.J. Poleward displacement of coastal upwelling-favorable winds in the ocean’s eastern boundary currents through the 21st century. Geophys. Res. Lett. 2015, 42, 6424–6431. [CrossRef] 9. Varela, R.; Álvarez, I.; Santos, F.; DeCastro, M.; Gómez-Gesteira, M. Has upwelling strengthened along worldwide coasts over 1982–2010? Sci. Rep. 2015, 5, 1–15. [CrossRef] 10. Relvas, P.; Luís, J.; Santos, A.M.P. Importance of the mesoscale in the decadal changes observed in the northern canary upwelling system. Geophys. Res. Lett. 2009, 36. [CrossRef] 11. Liu, Y.; Peng, Z.; Shen, C.C.; Zhou, R.; Song, S.; Shi, Z.; Chen, T.; Wei, G.; DeLong, K.L. Recent 121-year variability of western boundary upwelling in the northern south china sea. Geophys. Res. Lett. 2013, 40, 3180–3183. [CrossRef] Remote Sens. 2021, 13, 1386 14 of 16
12. Jan, S.; Wang, J.; Chern, C.-S.; Chao, S.-Y. Seasonal variation of the circulation in the taiwan strait. J. Mar. Syst. 2002, 35, 249–268. [CrossRef] 13. Hong, H.S.; Chai, F.; Zhang, C.Y.; Huang, B.Q.; Jiang, Y.W.; Hu, J.Y. An overview of physical and biogeochemical processes and ecosystem dynamics in the taiwan strait. Cont. Shelf Res. 2011, 31, S3–S12. [CrossRef] 14. Hu, J.Y.; Kawamura, H.; Hong, H.S.; Pan, W.R. A review of research on the upwelling in the taiwan strait. Bull. Mar. Sci. 2003, 73, 605–628. 15. Tang, D.; Kester, D.R.; Ni, I.H.; Kawamura, H.; Hong, H. Upwelling in the taiwan strait during the summer monsoon detected by satellite and shipboard measurements. Remote Sens. Environ. 2002, 83, 457–471. [CrossRef] 16. Jiang, Y.; Chai, F.; Wan, Z.; Zhang, X.; Hong, H. Characteristics and mechanisms of the upwelling in the southern taiwan strait: A three-dimensional numerical model study. J. Oceanogr. 2011, 67, 699–708. [CrossRef] 17. Hu, J.-y.; Kawamura, H.; Hong, H.-s.; Suetsugu, M.; Lin, M.-s. Hydrographic and satellite observations of summertime upwelling in the taiwan strait: A preliminary description. Terr. Atmos. Ocean. Sci. 2001, 12, 415–430. [CrossRef] 18. Chen, Z.; Yan, X.H.; Jiang, Y. Coastal cape and canyon effects on wind-driven upwelling in northern taiwan strait. J. Geophys. Res. Ocean. 2014, 119, 4605–4625. [CrossRef] 19. Cai, W.; Lennon, G. Upwelling in the taiwan strait in response to wind stress, ocean circulation and topography. Estuar. Coast. Shelf Sci. 1988, 26, 15–31. 20. Huang, B.; Xiang, W.; Zeng, X.; Chiang, K.-P.; Tian, H.; Hu, J.; Lan, W.; Hong, H. Phytoplankton growth and microzooplankton grazing in a subtropical coastal upwelling system in the taiwan strait. Cont. Shelf Res. 2011, 31, S48–S56. [CrossRef] 21. Xiao, H. Studies of coastal upwelling in western taiwan strait. J. Oceanogr. Taiwan Strait 1988, 7, 135–142. 22. Wang, Y.; Kang, J.-H.; Ye, Y.-Y.; Lin, G.-M.; Yang, Q.-L.; Lin, M. Phytoplankton community and environmental correlates in a coastal upwelling zone along western taiwan strait. J. Mar. Syst. 2016, 154, 252–263. [CrossRef] 23. Zhang, C.Y.; Hong, H.S.; Ru, C.M.; Shang, S.L. Evolution of a coastal upwelling event during summer 2004 in the southern taiwan strait. Acta Oceanol. Sin. 2011, 30, 1–6. [CrossRef] 24. Shang, S.L.; Zhang, C.Y.; Hong, H.S.; Shang, S.P.; Chai, F. Short-term variability of chlorophyll associated with upwelling events in the taiwan strait during the southwest monsoon of 1998. Deep Sea Res. Part II Top. Stud. Oceanogr. 2004, 51, 1113–1127. [CrossRef] 25. Tang, D.L.; Kawamura, H.; Guan, L. Long-time observation of annual variation of taiwan strait upwelling in summer season. Adv. Space Res. 2004, 33, 307–312. [CrossRef] 26. Hong, H.; Zhang, C.; Shang, S.; Huang, B.; Li, Y.; Li, X.; Zhang, S. Interannual variability of summer coastal upwelling in the taiwan strait. Cont. Shelf Res. 2009, 29, 479–484. [CrossRef] 27. Belkin, I.M.; Lee, M.A. Long-term variability of sea surface temperature in taiwan strait. Clim. Chang. 2014, 124, 821–834. [CrossRef] 28. Zhang, C.; Huang, Y.; Ding, W. Enhancement of zhe-min coastal water in the taiwan strait in winter. J. Oceanogr. 2020, 76, 197–209. [CrossRef] 29. Jacox, M.G.; Fiechter, J.; Moore, A.M.; Edwards, C.A. Enso and the c alifornia c urrent coastal upwelling response. J. Geophys. Res. Ocean. 2015, 120, 1691–1702. [CrossRef] 30. Roy, C.; Reason, C. Enso related modulation of coastal upwelling in the eastern atlantic. Prog. Oceanogr. 2001, 49, 245–255. [CrossRef] 31. Wolter, K.; Timlin, M.S. Measuring the strength of enso events—how does 1997/98 rank? Weather 1998, 53, 315–324. [CrossRef] 32. Ashok, K.; Behera, S.K.; Rao, S.A.; Weng, H.; Yamagata, T. El niño modoki and its possible teleconnection. J. Geophys. Res. Ocean. 2007, 112, C11. [CrossRef] 33. Shang, S.; Zhang, C.; Hong, H.; Liu, Q.; Wong, G.T.F.; Hu, C.; Huang, B. Hydrographic and biological changes in the taiwan strait during the 1997–1998 el nino winter. Geophys. Res. Lett. 2005, 32, 4. [CrossRef] 34. Kuo, N.J.; Ho, C.R. Enso effect on the sea surface wind and sea surface temperature in the taiwan strait. Geophys. Res. Lett 2004, 31, L13309. [CrossRef] 35. Saha, K.; Zhao, X.; Zhang, H.; Casey, K.; Zhang, D.; Baker-Yeboah, S.; Kilpatrick, K.; Evans, R.; Ryan, T.; Relph, J. Avhrr Pathfinder Version 5.3 Level 3 Collated (l3c) Global 4km Sea Surface Temperature for 1981-Present; NOAA National Centers for Environmental Information: Washington, DC, USA, 2018. 36. Qiu, C.; Wang, D.; Kawamura, H.; Guan, L.; Qin, H. Validation of avhrr and tmi-derived sea surface temperature in the northern south china sea. Cont. Shelf Res. 2009, 29, 2358–2366. [CrossRef] 37. O’Carroll, A.G.; Armstrong, E.M.; Beggs, H.M.; Bouali, M.; Casey, K.S.; Corlett, G.K.; Dash, P.; Donlon, C.J.; Gentemann, C.L.; Høyer, J.L. Observational needs of sea surface temperature. Front. Mar. Sci. 2019, 6, 420. [CrossRef] 38. Kuo, N.-J.; Zheng, Q.; Ho, C.-R. Satellite observation of upwelling along the western coast of the south china sea. Remote Sens. Environ. 2000, 74, 463–470. [CrossRef] 39. Pond, S.; Pickard, G.L. Introductory Dynamical Oceanography, 2nd ed.; Butterworth-Heinemann: Oxford, UK, 1983. 40. Trenberth, K.E.; Large, W.G.; Olson, J.G. The mean annual cycle in global ocean wind stress. J. Phys. Oceanogr. 1990, 20, 1742–1760. [CrossRef] 41. Cropper, T.E.; Hanna, E.; Bigg, G.R. Spatial and temporal seasonal trends in coastal upwelling off northwest africa, 1981–2012. Deep Sea Res. I 2014, 86, 94–111. [CrossRef] Remote Sens. 2021, 13, 1386 15 of 16
42. Lee, H.W.; Bhang, K.J.; Park, S.S. Effective visualization for the spatiotemporal trend analysis of the water quality in the nakdong river of korea. Ecol. Inform. 2010, 5, 281–292. [CrossRef] 43. Cleveland, W.S. Robust locally weighted regression and smoothing scatter plots. J. Am. Stat. Assoc. 1979, 74, 829–836. [CrossRef] 44. Richards, R.P.; Baker, D.B. Trends in water quality in leaseq rivers and streams (northwestern ohio), 1975–1995. J. Environ. Qual. 2002, 31, 90–96. [CrossRef] 45. Xie, L.; Zong, X.; Yi, X.; Li, M. The interannual variation and long-term trend of qiongdong upwelling. Oceanol. Limnol. Sin. 2016, 47, 43–51. 46. Yang, S.; Mao, X.; Jiang, W. Interannual variation of coastal upwelling in summer in zhejiang, china. Period. Ocean Univ. China 2020, 50, 1–8. 47. Pérez, F.F.; Padín, X.A.; Pazos, Y.; Gilcoto, M.; Cabanas, M.; Pardo, P.C.; Doval, M.D.; FARINA-BUSTO, L. Plankton response to weakening of the iberian coastal upwelling. Glob. Chang. Biol. 2010, 16, 1258–1267. [CrossRef] 48. García-Reyes, M.; Sydeman, W.J.; Schoeman, D.S.; Rykaczewski, R.R.; Black, B.A.; Smit, A.J.; Bograd, S.J. Under pressure: Climate change, upwelling, and eastern boundary upwelling ecosystems. Front. Mar. Sci. 2015, 2, 109. [CrossRef] 49. Xu, M.; Chang, C.P.; Fu, C.; Qi, Y.; Robock, A.; Robinson, D.; Zhang, H.m. Steady decline of east asian monsoon winds, 1969–2000: Evidence from direct ground measurements of wind speed. J. Geophys. Res. Atmos. 2006, 111, D24111. [CrossRef] 50. Zuo, Z.; Yang, S.; Kumar, A.; Zhang, R.; Xue, Y.; Jha, B. Role of thermal condition over asia in the weakening asian summer monsoon under global warming background. J. Clim. 2012, 25, 3431–3436. [CrossRef] 51. Wang, D.; Shu, Y.; Xue, H.; Hu, J.; Chen, J.; Zhuang, W.; Zu, T.; Xu, J. Relative contributions of local wind and topography to the coastal upwelling intensity in the northern south china sea. J. Geophys. Res. Ocean. 2014, 119, 2550–2567. [CrossRef] 52. Zhang, W.Z.; Shi, F.; Hong, H.S.; Shang, S.P.; Kirby, J.T. Tide-surge interaction intensified by the taiwan strait. J. Geophys. Res. Ocean. 2010, 115, C6. [CrossRef] 53. Wang, B.; Wu, R.; Fu, X. Pacific–east asian teleconnection: How does enso affect east asian climate? J. Clim. 2000, 13, 1517–1536. [CrossRef] 54. Pan, J.; Yan, X.-H.; Zheng, Q.; Liu, W.T.; Klemas, V.V. Interpretation of scatterometer ocean surface wind vector eofs over the northwestern pacific. Remote Sens. Environ. 2003, 84, 53–68. [CrossRef] 55. Li, N.; Shang, S.P.; Shang, S.L.; Zhang, C.Y. On the consistency in variations of the south china sea warm pool as revealed by three sea surface temperature datasets. Remote Sens. Environ. 2007, 109, 118–125. [CrossRef] 56. Chen, Z.; Wen, Z.; Wu, R.; Zhao, P.; Cao, J. Influence of two types of el niños on the east asian climate during boreal summer: A numerical study. Clim. Dyn. 2014, 43, 469–481. [CrossRef] 57. Yuan, Y.; Yang, S. Impacts of different types of el niño on the east asian climate: Focus on enso cycles. J. Clim. 2012, 25, 7702–7722. [CrossRef] 58. Weng, H.; Ashok, K.; Behera, S.K.; Rao, S.A.; Yamagata, T. Impacts of recent el niño modoki on dry/wet conditions in the pacific rim during boreal summer. Clim. Dyn. 2007, 29, 113–129. [CrossRef] 59. Feng, J.; Li, J. Influence of el niño modoki on spring rainfall over south china. J. Geophys. Res. Atmos. 2011, 116. [CrossRef] 60. Yeh, S.-W.; Kug, J.-S.; Dewitte, B.; Kwon, M.-H.; Kirtman, B.P.; Jin, F.-F.J.N. El niño in a changing climate. Nature 2009, 461, 511–514. [CrossRef] 61. Varela, R.; Lima, F.P.; Seabra, R.; Meneghesso, C.; Gómez-Gesteira, M. Coastal warming and wind-driven upwelling: A global analysis. Sci. Total. Environ. 2018, 639, 1501–1511. [CrossRef][PubMed] 62. Santos, F.; Gómez-Gesteira, M.; Varela, R.; Ruiz-Ochoa, M.; Días, J.M. Influence of upwelling on sst trends in la guajira system. J. Geophys. Res. Ocean. 2016, 121, 2469–2480. [CrossRef] 63. Casey, K.S.; Cornillon, P. Global and regional sea-surface temperature trends. J. Clim. 2001, 14, 3801–3818. [CrossRef] 64. Liao, E.H.; Lu, W.F.; Yan, X.H.; Jiang, Y.W.; Kidwell, A. The coastal ocean response to the global warming acceleration and hiatus. Sci. Rep. 2015, 5, 10. [CrossRef] 65. Santos, F.; de Castro, M.; Gómez-Gesteira, M.; Álvarez, I. Differences in coastal and oceanic sst warming rates along the canary upwelling ecosystem from 1982 to 2010. Cont. Shelf Res. 2012, 47, 1–6. [CrossRef] 66. Sousa, M.C.; Alvarez, I.; de Castro, M.; Gomez-Gesteira, M.; Dias, J.M. Seasonality of coastal upwelling trends under future warming scenarios along the southern limit of the canary upwelling system. Prog. Oceanogr. 2017, 153, 16–23. [CrossRef] 67. Arellano, B.; Rivas, D. Coastal upwelling will intensify along the baja california coast under climate change by mid-21st century: Insights from a gcm-nested physical-npzd coupled numerical ocean model. J. Mar. Syst. 2019, 199, 103207. [CrossRef] 68. Seabra, R.; Varela, R.; Santos, A.; Gómez-Gesteira, M.; Meneghesso, C.; Wethey, D.; Lima, F. Reduced nearshore warming associated with eastern boundary upwelling systems. Front. Mar. Sci. 2019, 6, 104. [CrossRef] 69. Xiu, P.; Chai, F.; Curchitser, E.N.; Castruccio, F.S. Future changes in coastal upwelling ecosystems with global warming: The case of the california current system. Sci. Rep. 2018, 8, 2866. [CrossRef] 70. Di Lorenzo, E. The future of coastal ocean upwelling. Nature 2015, 518, 310–311. [CrossRef] 71. Miranda, P.; Alves, J.; Serra, N. Climate change and upwelling: Response of iberian upwelling to atmospheric forcing in a regional climate scenario. Clim. Dyn. 2013, 40, 2813–2824. [CrossRef] 72. Chavez, F.P.; Messié, M. A comparison of eastern boundary upwelling ecosystems. Prog. Oceanogr. 2009, 83, 80–96. [CrossRef] 73. Lass, H.-U.; Mohrholz, V.; Nausch, G.; Siegel, H. On phosphate pumping into the surface layer of the eastern gotland basin by upwelling. J. Mar. Syst. 2010, 80, 71–89. [CrossRef] Remote Sens. 2021, 13, 1386 16 of 16
74. Dugdale, R.; Wilkerson, F. New production in the upwelling center at point conception, california: Temporal and spatial patterns. Deep Sea Res. Part A Oceanogr. Res. Pap. 1989, 36, 985–1007. [CrossRef] 75. Largier, J.L. Upwelling bays: How coastal upwelling controls circulation, habitat, and productivity in bays. Annu. Rev. Mar. Sci. 2020, 12, 415–447. [CrossRef][PubMed] 76. Mitchell-Innes, B.; Walker, D. Short-term variability during an anchor station study in the southern benguela upwelling system: Phytoplankton production and biomass in relation to specie changes. Prog. Oceanogr. 1991, 28, 65–89. [CrossRef] 77. Bettencourt, J.H.; Rossi, V.; Renault, L.; Haynes, P.; Morel, Y.; Garçon, V. Effects of upwelling duration and phytoplankton growth regime on dissolved-oxygen levels in an idealized iberian peninsula upwelling system. Nonlinear Process. Geophys. 2020, 27, 277–294. [CrossRef] 78. Zhong, Y.; Hu, J.; Laws, E.A.; Liu, X.; Chen, J.; Huang, B. Plankton community responses to pulsed upwelling events in the southern taiwan strait. ICES J. Mar. Sci. 2019, 76, 2374–2388. [CrossRef]