Two phytoplankton blooms near Luzon Strait generated by lingering Typhoon Parma


  • Hui Zhao,

    Corresponding author
    • Guangdong Key Lab. of Climate, Resource and Environment in Continental Shelf Sea and Deep Sea, College of Ocean and Meteorology, Guangdong Ocean University, Zhanjiang, China
    Search for more papers by this author
  • Guoqi Han,

    1. Northwest Atlantic Fisheries Centre, Fisheries and Oceans Canada, St. John's, NL, Canada
    Search for more papers by this author
  • Shuwen Zhang,

    1. Guangdong Key Lab. of Climate, Resource and Environment in Continental Shelf Sea and Deep Sea, College of Ocean and Meteorology, Guangdong Ocean University, Zhanjiang, China
    Search for more papers by this author
  • Dongxiao Wang

    1. State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
    Search for more papers by this author

Corresponding author: Hui Zhao, College of Ocean and Meteorology, Guangdong Ocean University, Zhanjiang 524088, China. (


[1] Two phytoplankton blooms near Luzon Strait triggered by Typhoon Parma in 2009 were investigated using remote sensing data and in situ observations. Parma was slow moving (a translation speed of ~2 m s−1) and relatively weak (a maximum sustained wind of ~30 m s−1) during its lingering path northwest of Luzon Island. After it reached a point (120.5°E, 20.3°N) west of Luzon Strait, Parma turned sharply back toward the northern Philippines along approximately the same course. Such long (~7 days) lingering typhoons are rather rare in the South China Sea (SCS). Before Parma, low Chl-a concentrations (<0.2 mg m−3) were observed in the northeastern SCS. After its passage, a strong offshore phytoplankton increase (Chl-a > 0.6 mg m−3) appeared west of the central Luzon Strait; a nearshore phytoplankton increase was also observed north of Luzon Island, together with high CDOM (color dissolved organic matter). During and after the typhoon, sea-surface cooling (~3°C), stronger wind (>20 m s−1), and heavy rainfall (>100 mm day−1) were seen in the above regions. The offshore bloom occurred where Parma's translation speed was the slowest (~1 m s−1). It may be caused primarily by the Ekman pumping which brought nutrients upward to the euphotic zone and also by the entrainment mixing. The nearshore bloom may be triggered by the heavy typhoon-induced rainfall, which supplied nutrients for the coast region north of Luzon Island. The rapid increase of CDOM in the nearshore region implied that terrestrial input may be the source of nutrients.

1 Introduction

[2] Typhoons can cause extremely strong winds, which can have dramatic effects on the upper oceans. They can have disastrous consequences during their passage over habitats of lands or farmlands. Due to limit of nutrients and influence of light, a maximum concentration appears at a subsurface layer which may vary from below the surface to near or below the bottom of the euphotic zone [Steele and Yentsch, 1960]. Therefore, typhoons may also play an important role in phytoplankton blooms and increased primary productivity in oligotrophic ocean waters [e.g., Lin et al., 2003; Babin et al., 2004; Walker et al., 2005; Zheng and Tang, 2007]. They can cause entrainment, strong vertical mixing, and upwelling as well as near-surface water's cooling on the right-hand side of the storm track in the Northern Hemisphere [Hazelworth, 1968; Dickey and Simpson, 1983; Stramma et al., 1986; Sanford et al., 1987; Price, 1998; Emanuel, 1999].

[3] Typhoons or tropical cyclones occur frequently in the South China Sea (SCS), more than seven times annually on average [Lin et al., 2003; Zheng and Tang, 2007; Zhao et al., 2008;]. They can trigger uptake of nutrients from subsurface layer and phytoplankton blooms near their paths through oceanic eddies, mixing, and upwelling [Chang et al., 1996, 2008; Chen et al., 2003; Lin et al., 2003; Zheng and Tang, 2007; Zhao et al., 2008]. Cyclones and typhoons have important effects on chlorophyll-a (Chl-a) and phytoplankton blooms, with estimated maximum contribution of 20–30% for the annual new production in the SCS [Lin et al., 2003]. Zhao et al. [2008] indicated that different kinds of typhoons, with different translation speeds and intensities, exert diverse impacts on intensity and area of phytoplankton blooms. The typhoon-induced increase of the Pearl River discharge [Zhao et al., 2009] can trigger an extensive phytoplankton bloom under the influence of favorable currents. Zheng and Tang [2007] investigated the impacts of a category 2 typhoon on two phytoplankton blooms in the northwest SCS, one nearshore and the other offshore. The offshore bloom was associated with the strongest maximum sustained wind of ~41 m s−1, while the inshore bloom occurred over the shallow continental shelf (depth <80 m), without any great river.

[4] Luzon Strait, the most important channel for the exchange of the SCS deep water with the water of the open northwestern Pacific Ocean, is about 350 km in width with a sill depth of about 1900 m. Previous investigations indicated that seasonal phytoplankton blooms and upwelling prevailed northwest of Luzon in winter [Chen et al., 2006; Gong et al., 1992; Shaw et al., 1996; Peñaflor et al., 2007; Lee Chen et al., 2007; Liu et al., 2007]. However, in the other seasons, the region is generally controlled by oligotrophic water with abundant light [Liu et al., 2002; Zhao and Tang, 2007]. Therefore, typhoons may play an important role in the increase of phytoplankton and primary productivity in the region.

[5] Parma at its strongest was a category 4 typhoon based on the Saffir-Simpson typhoon/hurricane scale ( It originated in the western Pacific (Figure 1) and traversed westward into the SCS. Parma became slow moving and relatively weak (not greater than category 1), while lingering near the northern Luzon Island for about 7 days in an area of 3° by 3° (Figures 1 and 2), where the offshore water depth is over 2000 m. Such long lingering typhoons are rather infrequent near Luzon Island in the SCS, and their influences on phytoplankton blooms have seldom been evaluated. Parma over its long lingering time may exert large influence on terrestrial nutrients runoff, mixing entrainment, and upwelling, and therefore phytoplankton in this region. In the present paper, we investigate two phytoplankton blooms (one offshore and the other nearshore) north of Luzon Island and the impacts of typhoon's translation speed and intensity during Parma, using satellite observations and in situ data. Dynamic mechanisms of the phytoplankton blooms are discussed based on meteorological and oceanographic data. In particular, the relative roles of the Ekman pumping versus the mixing entrainment are evaluated based on concurrent in situ temperature, salinity, and nutrient profiles in the deep SCS.

Figure 1.

(a) The study area (15°N–22°N, 118°E–125°E) and the track of typhoon Parma. (b) Track and intensity of Typhoon Parma (2009) in the study area. Shown in Figure 1a are (1) fast moving, intensifying wind speed; (2) fast moving, decreasing wind speed; and (3) slow moving, moderate wind speed. The three yellow asterisks in Figure 1b are the stations with CTD observations. LS: Luzon strait; LI: Luzon island.

Figure 2.

Change in maximum sustained wind speed (MSW) and translation speed of the typhoon. (a) MSW and sea-level pressure (SLP) of Typhoon Parma (box in Figure 2a, the stage influencing the study area). (b) Translation speed of the typhoon. In the red box of the lower panel, the mean translation speed is 1.6 m s−1, and the MSW is 28–31 m s−1 with a mean MSW of 29 m s−1. In the blue box of the lower panel (i.e., 4–5 October), the mean translation speed is 1.2 m s−1 for the radius range of 60 km. The gray box of the lower panel, the center of the typhoon, is located on Luzon Island.

2 Data and Methods

2.1 Satellite Products and Hurricane Data

[6] The Aqua MODIS-derived Chl-a product with 9 km resolution was obtained from the Distributed Active Archive Center (DAAC) of the National Aeronautics and Space Administration (NASA; Considering the influence of clouds during typhoon on satellite ocean color, we chose 8 day Level 3 ocean color data in the present study, which was calculated using the OC4 algorithm [O'Reilly et al., 1998]. The Aqua MODIS-derived (MODISA) color dissolved organic matter (CDOM) index is produced based on the algorithm of Morel and Gentili [2009], available at The 8 day product of CDOM index was used as a proxy to evaluate the influence from terrestrial materials and phytoplankton increase. Furthermore, the CDOM images were processed into mean images to discuss variations of CDOM during, before, and after Parma. In order to analyze the role of mixing, entrainment, and wind-induced upwelling on the offshore phytoplankton increase, MODISA-derived euphotic depth was also used in the present study (

[7] The absolute geostrophic velocity was obtained from the Colorado Center for Astrodynamics Research (CCAR) of the University of Colorado ( The data set was generated from model mean and altimeter measurements at the CCAR [Le Traon and Morrow, 2000].

[8] Daily rainfall rate and fusion of daily sea-surface temperature (SST) (TMI_AMSRE) ( were derived from the Tropical Rain Measuring Mission (TRMM) Microwave Imager (TMI) and the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) with a resolution of 0.25° by 0.25°. Due to the cloud-penetrating capacity of both TMI and AMSR-E, the two measurements together can overcome influence of cloudy conditions [Wentz et al., 2000]. Therefore, TMI_AMSRE and TRMM data can provide continuous SST/rainfall observations with better coverage before, during, and after a typhoon or hurricane.

[9] The surface wind data are based on the microwave scatterometer SeaWinds on QuikSCAT satellite that measured surface wind over the oceans [Liu et al., 2000]. The daily QuikSCAT data including ascending and descending passes downloaded from the NASA ( were used to study Typhoon Parma. The typhoon data used in this study were downloaded from the Unisys Weather website (, which is based on the best hurricane-track data from the Joint Typhoon Warning Center in the USA. The data include the maximum sustained wind (MSW) velocity, and the longitude and latitude of the hurricane center every 6 h. The translation speed (i.e., speed of movement) of a hurricane was thus estimated based on the 6 h position of its center in our analysis. The MSW speeds were labeled for the period of Typhoon Parma (Figure 1).

2.2 Methodology

2.2.1 Ekman Pumping Velocity (EPV) and Sea Surface Wind Vector (SSWV)

[10] To present the spatial-temporal variation of wind speed and wind-induced upwelling before and during Parma, daily sea surface wind vectors (SSWVs) were first processed into daily product averaged for the two ascending and descending passes, and Ekman Pumping Velocity (EPV) was estimated based on daily wind data [Stewart, 2002; Zhao and Tang, 2007]. Then, the daily product was averaged for 15–30 September and 4–5 October 2009. The time series of wind speed and EPV were obtained by averaging daily data over boxes (Figure 3), respectively.

Figure 3.

Surface wind vector (m s−1) and Ekman pumping velocity (EPV) (color shaded in 10−4 m s−1): (a) before typhoon and (b) during typhoon. Red Box: a sampling region of EPV (119°E–120°E, 19.5°N–20.5°N); and black box: a sampling region of wind speed (119°E–121°E, 19°N–21°N).

2.2.2 Chl-a Concentrations and CDOM

[11] Chl-a concentration and CDOM were first averaged over the two periods: the pre-Parma period (15–30 September 2009) and the post-Parma period (8 days after typhoon Parma's passage, i.e., 08–15 October 2009). The post-Parma period was so chosen since phytoplankton blooms generally appeared several days after a storm [Shi and Wang, 2007] and decayed gradually to the nominal prestorm level after several weeks, as well as there were generally relatively sparse valid observations of Chl-a due to cloudy weather conditions during Parma [Lin et al., 2003; Zheng and Tang, 2007; Zhao et al., 2008]. Their 8 day time series were produced based on the box average (boxes shown in Figure 3).

2.2.3 Rainfall, SST, and Geostrophic Current

[12] Similar to the means of QuikSCAT wind speed, the images of rainfall, SST, and geostrophic current (GC) were plotted to show changes of these conditions during the pre-Parma, Parma, and post-Parma periods. The products for the pre-Parma were estimated based on the average for 15–30 September 2009 (16–30 September 2009 for the GC). In view of their different response time scales to typhoons and data availability, here the rainfall rate was averaged for 02–09 October, SST averaged for 04–12 October, and GC averaged for 7–14 October for the Parma and post-Parma periods. These variables were further averaged over the boxes or point in Figure 5 to obtain their time series.

2.2.4 Time Series of Satellite and In Situ Data

[13] To investigate further the relationship between Chl-a concentration and oceanic conditions (including river/terrestrial discharge and slow-translation speed of the typhoon), we chose two boxes (119°E–120.5°E, 19.8°N–20.9°N) and (120°E–122.3°E, 18.3°N–19.3°N) in Figure 4b for the time series, where the variations of Chl-a and CDOM were more notable. The time series of Chl-a, CDOM, EPV, wind speed, rainfall rate, SST, and GC were averaged over the corresponding boxes (Figures 3-5) during 25 August to 24 October 2009, where the changes were more evident. Using World Ocean Atlas data 2009 (WOA 09) (, we produced a climatological vertical profile of nitrate for September by averaging over the offshore box (Figure 3c) to indicate the typical vertical distribution of nutrients. Three in situ profiles of temperature and salinity (depicted by three pentacles in Figure 1) were used to indicate changes in the upper layer in the offshore typhoon-induced region (available at: From salinity and temperature, potential density (sigma-theta) is calculated to estimate the mixed layer depth (MLD) using Levitus' method [Levitus, 1982]: starting at 10 m, search down the water column until the potential density has increased by 0.125 kg m−3.

Figure 4.

Left panels: images before Typhoon Parma; middle and right panels: images after Typhoon Parma. (a) Color dissolved organic material (CDOM) (no units) and (b) Chlorophyll-a (Chl-a) (mg m−3). The offshore box (119°E–120.5°E, 19.8°N–20.9°N) and the nearshore box (120°E–122.3°E, 18.3°N–19.3°N) in the lower panels are sampling regions of Chl-a and CDOM.

Figure 5.

Left panels: before Typhoon Parma; right panels: during/after Typhoon Parma. (a) TRMM rainfall (mm d−1); (b) GHRSST L4 RSS MW IR OI SST (°C); (c) absolute geostrophic current (GC; m s−1). The black boxes in Figures 5a and 5b delineate sampling regions of TRMM rainfall (120°E–122°E, 16°N–19°N) and SST (118.5°E–120.5°E, 18.5°N–21°N), respectively; the asterisk in Figure 5c is the sampling point of the GC (121.6°E, 8.8°N).

3 Results

3.1 Wind and EPV During Typhoon Parma

[14] Parma was a category 4 typhoon (Figure 1), which originated from a tropical depression in the northwest Pacific (147.50°E, 9.90°N) at 00:00 UTC on 27 September 2009, strengthened to a typhoon at 00:00 UTC on 30 September with the strongest wind speed near (130.90°E, 11.90°N) at 00:00 UTC on 01 October, and weakened to a category 1 typhoon (MSW: 41 m s−1) when it intruded into the northeastern SCS through Luzon Strait at 06:00 UTC on 03 October. Parma maintained generally at the level of a tropical storm or a tropical depression during its intrusion into the SCS. It gradually weakened to become a tropical storm after it lingered near (119.8°E, 19.9°N); it then retraced to Luzon Island. Its mean MSW was 30 m s−1 (at a level of strong tropical storm), with a slow mean translation speed of 1.6 m s−1 (Figure 2) and the slowest translation speed of 0.7 m s−1 (the MSW was only 28 m s−1) during its lingering time northwest of Luzon Island. Finally, Parma traversed the northern SCS from the coastline of central western Philippines to the Gulf of Tonkin with a mean MSW of ~20 m s−1 and a mean translation speed of ~3.1 m s−1.

[15] The wind vector image (Figure 3a) averaged for 15–30 September as the pretyphoon background state shows that the wind was generally weak (<8 m s−1) and northeasterly in the study area and was even weaker (<3 m s−1) south of the study area. Here, we chose an image of the wind vector averaged for 4–5 October (Figure 3b) to represent the wind during the typhoon. During Parma's intrusion into the northern Luzon Island, the relatively strong wind (>12 m s−1; Figure 3b) prevailed over the entire study area. The intensity of wind speed (Figures 1a and 3a) was stronger (>15 m s−1) along Parma's track and northwest of the study area, where there was a mean MSW of 34 m s−1 with a fast translation speed of 3.1 m s−1. There was a weakening tendency of MSW from 39 m s−1 at (121.20°E, 18.20°N) at 12:00 UTC on 3 October to 29 m s−1 at (119.80°E, 19.90°N) at 12:00 UTC on 4 October. However, there was a weaker stable MSW of about 28 m s−1 with a slower mean translation speed of 1.6 m s−1 when it retraced from the location of (119.8°E, 19.90°N) from 12:00 UTC on 4 October to 18:00 UTC on 5 October.

[16] The mean wind speed during Parma (Figure 3b) was only threefold higher than that before Parma in the study area. However, upwelling indicated by the EPV in the region increased significantly, with the EPV value up to 1–4 × 10−4 m s−1 near Parma's path (the region with yellow/red colors in Figure 3b), which is about 2 orders of magnitude higher than the peak value in the background EPV (<0.08 × 10−4 m s−1; Figure 3b). The region of high EPV (>2 × 10−4 m s−1) induced by Parma (Figure 3b) was located near (119.4°E, 19.8°N), where Parma lingered at its slowest speed.

3.2 Chl-a and CDOM

[17] Satellite-derived Chl-a (Figure 4b, left) averaged for 15–30 September 2009 was generally low (<0.1 mg m−3) before Parma's presence in the study area. Relatively high Chl-a (>0.3 mg m−3) was confined to the coastal zone (within about 20 km from the coastline). However, there were some drastic changes in Chl-a 1 week after Parma's passage. There were two significant bloom regions: one offshore northwest of Luzon Island and the other nearshore north of Luzon Island. The offshore Chl-a (Figure 4b, middle) increased quickly from 0.08 to 0.73 mg m−3 in the northwestern study area (averaged over an offshore area of 19,189 km2 in Figure 4b, middle) with a peak value of about 1.5 mg m−3. In the coastal region, the Chl-a increased from 0.2 to 1.3 mg m−3 (averaged over the coastal box of about 2.68 × 104 km2 in Figure 4b, middle), with a peak of 5.6 mg m−3. Then, the Chl-a (Figure 4b, right) decreased to the level of ~0.2 mg m−3 2 weeks after Parma in the offshore bloom area; however, the nearshore bloom decreased relatively slowly with a Chl-a level over 0.3 mg m−3 based on an average for 16–23 October. CDOM (Figure 4a, left) before Parma's intrusion was relatively low in the study area. CDOM (Figure 4a, middle) displayed obvious increase in most of the study area during 8–15 October after Parma's intrusion, especially in the coastal region. Two weeks after the intrusion, the CDOM (Figure 4a, right) displayed an increasing tendency in the two bloom regions.

3.3 Rainfall and SST

[18] Before Parma's passage, the mean rainfall was generally low (<15 mm d−1) in most of the study area, with a little higher rainfall (15–30 mm d−1) in the southernmost study area (Figure 5a, left). During the period of Parma's intrusion into the study area (Figure 5a, right), the rainfall averaged for 02–09 October around the typhoon's path increased to 3–10 times (>45 mm d−1; Figure 5a, right) of the pre-Parma rainfall (Figure 5a, left). There was heavier rainfall (>60 mm d−1) over Luzon Island, being the heaviest (>135 mm d−1) in the northwestern Luzon Island, roughly centered at (120.5°E, 16.5°N; Figure 5a, right). The SST averaged over 15–30 September (Figure 5b, left) indicated that high temperature (>29°C) prevailed in the region before Parma. The SST decrease was evident after Parma's intrusion. There was a low SST patch (<26.5°C) near Parma's path (Figure 5b, right) northwest of Luzon Island, roughly a decrease of 2–3°C.

3.4 Absolute Geostrophic Current

[19] There were a relatively strong cyclonic current (Figure 5c, left) west of Luzon Island and a strong northward current (i.e., the Kuroshio) north of Luzon Island based on the absolute geostrophic current averaged for 16–30 September before Parma's intrusion. A weak southward current (<0.1 m s−1) was also observed north of Luzon Island to 20°N. After the typhoon's passage, the northward current was intensified north of Luzon Island where the Cagayan River discharges. After Parma, the cyclonic current (Figure 5c, right) in the northwestern study area (118°E–121°E, 19°N–22°N) became much stronger too.

3.5 Time Series of Ocean Conditions

[20] The SST decrease was significant, with a maximum reduction of 3.2–3.7°C 3 days after the typhoon's passage on 8 October (Figure 6a). Then, the SST increased slowly. There was low rainfall (<10 mm d−1) during 15 September01 October (Figure 6b) (except on 25 September when another fast-moving typhoon passed by). However, intensive storm-induced rainfall accumulation of 662 mm for 02–08 October occurred over the northern Luzon Island during Parma's intrusion. The rainfall was evidently higher than that during the nontyphoon period. The time series of meridional component (Figure 6c) of the geostrophic current indicated that the northward current was strong (> 0.3 m s−1) during Parma; then the current speed reduced, and it reversed direction immediately after Parma's passage.

Figure 6.

Time series of ocean conditions for the chosen regions and location. (a) SST (°C); (b) rainfall (mm day−1); (c) the V-component of the geostrophic current (m s−1); (d) wind speed (m s−1); and (E) Ekman pumping velocity (10−4 m s−1).

[21] The wind speed (<8 m s−1) was generally weak with obvious diurnal variation during the pre- and post-Parma periods (Figure 6d). The wind speed was relatively higher (>12 m s−1) during the typhoon's lingering over the study area. Thus, the mean wind speed during this period was only ~1.5 times of that before Parma. However, the EPV (Figure 6e) was low during the periods without Parma and dramatically high during Parma, with an estimated cumulative vertical displacement of ~47.5 m based on the area-averaged EPV and Parma's lingering time (3 to 5 October).

[22] Before Parma's passage, there were generally low concentrations of CDOM (Figure 7; on average, ~1.4 for the offshore region and ~2.3 for the nearshore region) and Chl-a (on average, <0.08 mg m−3 for the offshore region and <0.4 mg m−3 for the nearshore region). The peak value of Chl-a (on average, 0.73 mg m−3 for the offshore region and 1.3 mg m−3 for the nearshore region) appeared in the first week (08–15 October 2009) after Parma's passage.

Figure 7.

Time series of 8 day-mean Chl-a and CDOM. (a, c) Averaged for the offshore box in Figure 4b (middle); and (b, d) averaged for the nearshore box in Figure 4b (middle). Note: the data shaded by the blue bars is unreliable in Figures 7b and 7d due to their sparse distribution.

[23] Peaks of CDOM (on average, 4.3 and 6.5 for the offshore and nearshore regions, respectively) occurred over 16–23 October, 2 weeks after Parma, with a 1 week lag to the peaks of Chl-a blooms. The CDOM in the offshore bloom region was low (<2) during the period from 29 September to 15 October and only increased significantly to its peak in the second week after Parma's passage. In contrast, the CDOM in the nearshore bloom region increased immediately to the level over 4.5 after Parma's intrusion (i.e., on 8–15 October), double the pre-Parma CDOM. Due to limited data availability, we did not analyze the mean values of Chl-a and CDOM averaged for 30 September to 7 October.

3.6 In Situ Observation

[24] The hydrographic profiles are available from the National Oceanographic Data Center (NODC) of the National Oceanic and Atmospheric Administration. The Conductivity-Temperature-Depth (CTD) data were downloaded from We used only the CTD data of three stations at (119.8°E, 21.09°N), (119.6°E, 21.03°N), and (119.11°E, 21°N) near the offshore bloom region on 24 September, 04 October, and 14 October, respectively. In this region, the climatological nitrate in September (Figure 8a) increases with depth below 50 m. The temperature and salinity profiles (Figures 8b–8d) show clear changes before, during, and after Parma. The MLD was ~20 m with high SST over 29°C on 24 September before Parma. The MLD deepened to about 56 m on 4 October with an SST of below 29°C. After Parma on October 14, although the MLD decreased obviously to 16 m, the isotherms of 20–27°C, the isohalines of 33.5–34 psu, and the isopycnals of 22–24 kg m−3 shoaled by about 50 m compared with those before/during Parma, roughly consistent with the cumulative upwelling distance suggested by the EPV. This consistency suggests that the Ekman pumping associated with the Parma's wind stress curls is the primary cause for the isotherm, isohaline, and isopycnal shoaling, nutrient upwelling, and therefore phytoplankton bloom.

Figure 8.

Vertical profiles of (a) the climatological nitrates in September, (b) temperature (°C), (c) salinity (psu), and (d) density (σt).

4 Discussion

[25] We observed two Chl-a blooms near Luzon Strait after the passage of Typhoon Parma: (1) an offshore Chl-a increase northwest of Luzon Island (the upper box in Figure 4b, middle) and (2) a nearshore Chl-a bloom near the Cagayan River north of Luzon Island.

4.1 Offshore Phytoplankton Bloom

[26] The increase of the offshore Chl-a occurred after Parma's passage, instead of an immediate response as the typhoon passed by, different from the quick drop in SST (Figures 5b (right) and 6a). The CDOM maxima lagging the Chl-a maxima by ~8 days was mainly produced by phytoplankton decomposition [Sathyendranath, 2000; Zhang et al., 2009]. Hu et al. [2006] found that CDOM maxima lagged pigment maxima by 2–4 weeks in the central North Atlantic Ocean through investigating a connection between chlorophyll and CDOM. The difference in the lag time between our study and Hu et al.'s [2006] may result from higher phytoplankton turnover in tropical oceans as well as different phytoplankton structures [Behrenfeld and Falkowski, 1997]. The lag phenomena (Figure 7d) implied also that the offshore Chl-a increase was not induced directly by entrainment mixing from a preexisting subsurface chlorophyll maximum. It was unlikely that the Chl-a increase was induced by horizontal transportation from the coastal water, since a lower Chl-a patch was observed between the coastal and offshore blooms. Thus, the offshore Chl-a increase was probably a phytoplankton bloom due to phytoplankton growth sustained by new nutrients brought into the euphotic zone from below, primarily through upwelling and supplemented by mixing entrainment.

[27] Why did the offshore bloom appear northwest of Luzon Island, where the maximum wind speed was about 30 m s−1 only? Typhoons were much stronger in previous studies on the responses of phytoplankton blooms to typhoons in the SCS [Lin et al., 2003; Zhao et al., 2008; Zheng and Tang, 2007]. However, according to the results of Zhao et al. [2008], the translation speed of a typhoon is also a key factor that affects phytoplankton bloom. Parma was a slow-translation storm in the study region; therefore, the cumulative effect of entrainment and upwelling could be significant. A recent study has shown that the typhoon size is also an important parameter to determine the ocean's response to a typhoon [Lin, 2012]. Parma has generally a gale force wind radius over 130 km in the region, according to data issued by the Japan Meteorological Agency (JMA) ( Consequently, the influence of Parma on the bloom in the region could last 2.5 days, in view of its mean slow-translation speed of 1.6 m s−1 and its roundabout route during the period. The accumulated displacement of typhoon-induced upwelling (based on the EPV integration from 3 to 5 October) was about 47.5 m in the offshore bloom region. According to the geostrophic adjustment theory [Gill, 1982], the potential wind-induced upwelling velocity and mixing are only fully developed after a time longer than the geostrophic adjustment time (i.e., the inertial period). The forcing time (~60 h) in the present study was much longer than the geostrophic adjustment time (T = 35 h for the latitude of 20°N where the center of the bloom was). Moreover, the preexisting cold-core eddy/cold water may strengthen the upper ocean dynamics and nutrient responses with significant increase of nutrient concentration after the typhoon passage [Lin et al., 2009; Liu et al., 2009]. Thus, the upwelling/mixing processes can be well established in the offshore region. According to the climatological vertical profile of WOA 09 (Figure 8a), nitrate in September is nearly constant above 50 m and increases with depth below 50 m. In other words, the additional upward transport of nutrients into the surface layer could only occur when the depth of mixing was deeper than 50 m. Upwelled water with rich nutrients from below 50 m depth during Parma as indicated by shoaling isohalines, isotherms, and isopycnals (Figures 8b–8d) can be critical to the offshore bloom. Therefore, we hypothesize that the wind-induced upwelling (Ekman pumping) transported deeper water with rich nutrients into the upper layer, and then the nutrients reached surface layer through the wind-induced entrainment mixing, triggering the surface phytoplankton bloom in the region.

4.2 Nearshore Chl-a Bloom

[28] The nearshore Chl-a bloom (lower box in Figure 4b, middle) appeared near the northern tip of Luzon Island, where the Cagayan River Estuary, the longest and largest in Luzon Island, lies. The sharply increasing rainfall (Figure 6b) in the northern Luzon Island was observed with the total rainfall of about 600 mm during 02–08 October, and the highest daily rainfall over 200 mm occurred on 3 October (Figure 6b). The huge precipitation could cause a significant increase of river discharge during Parma. Figure 4a (middle) may indicate that the greater volume of nutrient-rich water from terrestrial runoff with high concentration of CDOM was transported to the region. During Parma, the GC was strong (Figure 7c) and northward, and after Parma, the GC was weak or southward. These changes of surface circulation could firstly carry nutrient-rich water into the nearshore region and then cause high nutrient-rich water to spread northward, which was favorable to phytoplankton blooms. Due to the steep topography and favorable current conditions during Parma (Figure 7c), the transport of runoff from this watershed into the ocean would have been rapid. Moreover, the weak wind speed, EPV, and weak sea-surface cooling in the nearshore region (Figure 5b, right) all implied that wind-induced upwelling or entrainment mixing was weak. On the other hand, the runoff had likely been discharged from northern Luzon Island into this nearshore area, where phytoplankton and CDOM both increased significantly (Figure 4). Although the high level of CDOM may cause an overestimation of the Chl-a concentration, it suggests favorable conditions for phytoplankton with an increased load of nutrient of terrestrial origin.

[29] Moreover, the surface current north of Luzon Island was weak or southward before and after the typhoon, while a strong northward current during the typhoon was observed (Figure 6c). This northward current could transport the coastal water and the discharge from the estuary into the northern SCS. In comparison to seawater, terrestrial runoff contains a mass of nutrients [Yin et al., 2004] and CDOM [Milliman and Meade, 1983] orders of magnitude higher, because rivers and coastal waters near coastlines can not only carry CDOM primarily from soils but also contain plankton-derived CDOM produced in rivers and estuaries, as well as anthropogenic compounds from runoff, sewage discharge, and other effluents [Coble, 2007]. The higher CDOM (Figure 4a, middle) in the bloom region immediately after the typhoon suggests that the discharged water or coastal water with high CDOM may have intruded northward up to 20°N north of Luzon Island. Although data were not available due to sparse data in the region during 30 September to 7 October, the evident CDOM increase (Figure 7) for 8–15 October after the typhoon passage suggests that the nearshore bloom region was influenced by the terrestrial discharge. The CDOM maxima in the nearshore region lagged the Chl-a maxima by ~8 days, indicating a similar phytoplankton degradation rate to that in the offshore region. Thus, the significant increase in terrestrial discharge after the typhoon and the strong northward current both played important roles in triggering the bloom in the nearshore region. Despite the large amount of rainfall, the river discharge might not be able to spread as far from the coast without a typhoon-induced current (reflected by the Chl-a pattern; lower boxes in Figures 4b and 5c). Seaward-extending Chl-a filaments carried by typhoon-induced eddies have been observed before [Davis and Yan, 2004; Yuan et al., 2004; Walker et al., 2005]. Typhoon-induced rainfall discharge, therefore, can influence offshore water more than the nontyphoon rainfall discharge which tends to be confined to a narrower coastal zone.

5 Conclusion

[30] Two typhoon-induced phytoplankton blooms were observed near Luzon Strait where Parma was during the weak-intensity and slow-moving stage: one offshore west of the central Luzon Strait and other nearshore north of Luzon Island. The CDOM maxima in two bloom areas lagged the Chl-a maxima by ~8 days. The offshore bloom was likely triggered by wind-induced upwelling of nutrients and mixing entrainment in view of Parma's longer lingering time, with the former playing a more important role. The nearshore bloom was likely due to the increased discharge from the Cagayan River Estuary supported by typhoon-induced strong precipitation and favorable variation of current directions during the typhoon.


[31] The present research was supported by the National Natural Science Foundation of China (Grant number: 41006070, 41176011, and U0933001) and the Canadian Space Agency Government Related Initiative Program (GRIP). We thank NASA's Ocean Color Working Group for providing Modis and SeaWiFS data, Remote Sensing Systems for TMI-AMSRE sea-surface temperature and QuikSCAT wind-vector data, the Physical Oceanography Distributed Active Archive Center (PO.DAAC) for QuikScat wind stress, GES DAAC for TRMM accumulated rainfall data, and the Colorado Center for Astrodynamics Research (CCAR) Altimeter Data Research Group for sea-level anomaly data.