Geophysical Research Letters

Enhancement of phytoplankton primary productivity in the southern East China Sea following episodic typhoon passage



[1] The enhancement of primary productivity (PPenh) in the southern East China Sea (ECS) following 16 typhoon passages was investigated using ocean color data and a primary productivity model. PPenh tended to be higher when typhoons traversed slowly with trajectories that allowed strong southerly winds to prevail over Yonaguni Island. Such long-lasting southerly winds were believed to push the Kuroshio current axis shelfward, enhancing the upwelling of nutrients, hence promoting new productivity (NP). The importance of long-lasting southerly winds as a proxy for physical perturbations underlying PPenh was expressed by an empirical equation by which 88% of PPenh variation could be explained. Applying this equation, we assessed that typhoon passages accounted for a minimum of 0.6–11.8% of the ECS summer–fall NP, suggesting that typhoon passage over the southern ECS is an important phenomenon supporting NP in the ECS.

1. Introduction

[2] In the southern East China Sea (ECS), the impingement of the Kuroshio current onto the continental shelf results in year-round persistent upwelling and one of the most important nutrient sources in the ECS [Liu et al., 1992; Gong et al., 1995]. This upwelling of nutrients supports high primary productivity (PP, mgC m−2 d−1) and an abundance of ichthyoplankton [Chiu, 1991; Shiah et al., 1995], making the southern ECS as an important fishing ground.

[3] During the boreal summer–fall period, tropical cyclones (typhoons) frequently pass near the southern ECS. It has been suggested that cross-shelf typhoon passages in the vicinity of Taiwan push the Kuroshio current axis shelfward [e.g., Chen et al., 2003a, Morimoto et al., 2009] and lead to the upwelling of subsurface Kuroshio water, inevitably enhancing phytoplankton chlorophyll-a (Chl-a, mg m−3) and PP [Chen et al., 2003a].

[4] Although the ocean physical responses to typhoon passages have been investigated, large-scale biological responses have not been quantified in the southern ECS. Chang et al. [2008] investigated the influence of Typhoon Hai-Tang (July 2005) on Chl-a variation over a large spatial scale, but did not study phytoplankton PP. Chen et al. [2003a] used in situ data to investigate the influence of Typhoon Herb (July 1996) on PP variation, but did not cover a large spatial scale and thus do not reveal the spatial extent of the PP variation. Each of the aforementioned studies was based on a single typhoon event, and thus could not identify the main factors determining variations in biological response to typhoon passages. Thus, this study aims to investigate the influence of typhoon passages on PP variation in the southern ECS and to identify the probable factors influencing typhoon-enhanced PP variation.

2. Methods

[5] Phytoplankton biomass data before and after typhoon events were obtained from MODIS LAC Chl-a daily data (1-km resolution). To reduce the number of cloudy pixels, we also used SeaWiFS LAC (1-km, up to 2004) and GAC Chl-a daily data (4-km, from 2005) merged with MODIS Chl-a. Daily MODIS LAC sea surface temperature (SST, °C) data were also used for computing PP. Composite images of Chl-a and SST from several daily images were constructed to provide pre- and post-typhoon Chl-a and SST images (see auxiliary material).

[6] We computed pre- and post-typhoon PP by applying Gong and Liu's [2003] empirical model of PP = 2.512 [Chl-a PoptB Kd−1]0.957, where PoptB and Kd are the optimum Chl-a specific PP within the water column (mgC mg Chl-a−1 d−1) and the light attenuation coefficient (m−1), respectively. PoptB and Kd were computed using PoptB = −286.17 + 49.166 [SST] – 2.543 [SST2] + 0.0435 [SST3] [Gong and Liu, 2003] and Kd = 0.047 + 0.063 [Chl-a] [Siswanto et al., 2005], respectively. We chose Gong and Liu's [2003] PP model because many of the data from which it was constructed were collected from the southern ECS. PP enhancement (PPenh, Gg C) was computed by subtracting the pre-typhoon PP from the post-typhoon PP over the region of interest (see Section 4). Before subtraction, cloudy pixels were corrected with the simple formula of ([Tpxl/Cpxl] ICpxl), where Tpxl, Cpxl, and ICpxl are total pixels, cloud-free pixels, and integrated PP over all cloud-free pixels, respectively [see Siswanto et al., 2007].

[7] In order to determine the pre- and post-typhoon Kuroshio current axis positions, we used sea surface current velocity data measured with a high-frequency (HF) ocean surface radar system, the Long-Range Ocean Radar (LROR). The LROR was developed by the National Institute of Information and Communications Technology, Japan [Sato et al., 2004].

[8] We analyzed 16 typhoon passages (Figure 1a) between 2003 and 2007 for which before and after SeaWiFS and/or MODIS data were available. Typhoon variables including typhoon translation speed (TS, m s−1) and maximum sustained wind speed (MSW, m s−1) were obtained from the Japan Meteorological Agency (JMA). We also analyzed JMA meridional wind speed (V, m s−1) at Yonaguni Island (YI, Figure 1a), as during a typhoon event the southerly wind is important in driving the Kuroshio current onto the shelf [Morimoto et al., 2009].

Figure 1.

(a) The best tracks of 16 typhoons (red lines with dots) investigated in this study overlaid on the ECS bathymetric map. Green circle indicates the location of Yonaguni Island. Isobaths of 50, 100, and 200 m are also shown. (b, d, f, and h) Pre-Hai-Tang (12–16 July) and (c, e, g, and i) post-Hai-Tang (22–24 July) composite images of merged Chl-a, MODIS SST, estimated PP, and merged nLw555. (j and k) HF radar-measured surface currents for before (16 July) and after (21 July) Hai-Tang, respectively. Dotted black circle in (a) and dotted red circles in (c), (e), (g), and (i) denote the region over which PPenh was computed, and is considered to have minimum suspended sediment interference on satellite Chl-a retrieval. Yellow (or blue in SST images) contour line in Chl-a, nLw555, and PP images indicates 200-m isobath. White line with dots in (b)–(i) is Hai-Tang's best track. Hai-Tang position on 18 July 12:00 is marked with a blue dot over Taiwan. Red vectors in (j) and (k) denote the Kuroshio current axis with surface current velocities >100 cm s−1. The pre- and post-typhoon Chl-a, SST, nLw555, and PP for all 16 typhoon passages, as well as the Kuroshio current axis variation associated with 9 typhoon passages are presented in the auxiliary material.

3. Typhoon Passage-Enhanced Primary Productivity

[9] Phytoplankton Chl-a and hence PP in the southern ECS were enhanced following the 16 typhoon passages (see auxiliary material). The least PPenh occurred after the passage of Nock-Ten (October 2004), and the greatest followed the passage of Hai-Tang (July 2005). The PPenh caused by Hai-Tang also covered the largest area among the investigated typhoons. Before Hai-Tang, the Chl-a and PP in the southern ECS were generally around 0.3 mg m−3 and 500 mgC m−2 d−1 respectively, but after Hai-Tang had passed, Chl-a and PP were enhanced to a maximum of around 4.3 mg m−3 and 1780 mgC m−2 d−1 respectively (Figures 1b, 1c, 1f, and 1g).

[10] The most apparent PPenh after Hai-Tang coincided with the coolest SST (22.7°C Figure 1e) observed among post-typhoon SSTs (see auxiliary material). The minimum pre-Hai-Tang SST was 26.1°C (Figure 1d). This indicated that strong physical perturbations allowed the introduction of nutrient rich, cold water into the surface layer. These nutrients seemed to stimulate phytoplankton growth, resulting in the observed high PPenh. It is noteworthy that after Hai-Tang, the Kuroshio axis position (indicated by HF radar-measured ocean surface velocity of >100 cm s−1 in Figures 1j and 1k) had deviated somewhat from its position before Hai-Tang. The Kuroshio current axis before Hai-Tang lay in a northeast–southwest direction, but after Hai-Tang was mostly oriented in a north–south direction.

4. Main Factors Determining Variation in Primary Productivity Enhancement

[11] To identify factors that might cause the observed variations in PPenh, we integrated PPenh over the study region, as shown in Figure 1a, and then compared this with the averages of TS, MSW, and V. The reasons behind study region selection and the procedures used to derive mean TS, MSW, and V are detailed in the auxiliary material.

[12] To ensure that the estimated PP was not affected by suspended sediment from river discharge and resuspension caused by typhoon-driven mixing, we analyzed a normalized water-leaving radiance at 555 nm (nLw555, mW cm−2μm−1 sr−1) commonly used as an indicator of suspended sediment concentration [e.g., Warrick et al., 2004]. These nLw555 data were merged from SeaWiFS and MODIS data (see auxiliary material). On average, over the selected region, although the nLw555 after the 16 typhoon passages (0.50 mW cm−2μm−1 sr−1) was slightly higher than that before the passages (0.44 mW cm−2μm−1 sr−1), these nLw555 values were far from the levels that would indicate the presence of suspended sediment which might overestimate satellite Chl-a retrieval [see Tan et al., 2006; Siswanto et al., 2008]. Even during the highest PPenh after Hai-Tang, nLw555 was generally <1.0 mW cm−2μm−1 sr−1 and although this was higher than the pre-Hai-Tang nLw555 (∼0.5 mW cm−2μm−1 sr−1, Figures 1h and 1i) it nonetheless indicates that PP estimates were unlikely to be largely unaffected by suspended sediment.

[13] Over regions of low turbidity, PPenh was significantly correlated with TS (R2 = 0.76, p < 0.0001, Figure 2a), indicating that slower TS causes physical perturbations to occur over a longer period, thus allowing more PP-enhancing nutrients to be introduced to the euphotic zone. A non-linear relation between PPenh and TS (Figure 2a) seemed to be the best equation to reflect a non-linear response of phytoplankton to upwelled nutrients [see Dugdale, 1985]. PPenh was not significantly correlated with MSW (Figure 2c), as has been reported in previous studies [e.g., Babin et al., 2004; Siswanto et al., 2007]. Of interest is the finding that PPenh was significantly correlated with V measured at YI (R2 = 0.45, p < 0.01, Figure 2b). It was clear that the strong southerly winds that generally prevailed when a typhoon's trajectory crossed Taiwan from east to west, as in the case of Hai-Tang, greatly enhanced PP. In contrast, typhoons with trajectories that did not allow strong southerly winds to prevail, such as Nock-Ten, enhanced PP less.

Figure 2.

Scatter plots of PPenh against typhoon TS (a), V (b), and MSW (c). PPenh was computed by integrating the difference between pre- and post-typhoon PP over the study regions denoted in Figure 1a. Labels HT and NT indicate data for Typhoon Hai-Tang and Nock-Ten, respectively. The black curves in (a) and (b) are non-linear relationships the equations of which are mentioned in the graph. Line in (c) represents linear fit to the data. (d) Scatter plot comparing PPenh derived from satellite data and that computed using the constructed empirical equation. (e) Long-term variations in typhoon number and percentage of PPenh relative to ECS summer–fall NP within the period from 1980 to 2007. Note that because there were no continuous ocean color data during the 1980–1997 period, ECS summer–fall NP within this period was defined as the average summer–fall NP from 1998 to 2007.

[14] The higher PPenh associated with a typhoon's slower TS and stronger southerly wind indicated that southerly typhoon winds of long duration were important in determining PPenh. It is known that southerly typhoon winds of long duration are also important in pushing the Kuroshio axis onto the shelf [Morimoto et al., 2009]. The linkage of these two pieces of evidence can be explained as follows: long-lasting southerly winds force coastal upwelling along the east coast of Taiwan, generating an east–west sea level gradient that in turn leads to a northward geostrophic current. Such a northward geostrophic current accelerates the Kuroshio current, and thus the Kuroshio current axis shifts from its normal northeast–southwest direction to a north–south direction, causing it to move shelfward. This enhances the upwelling of subsurface Kuroshio water onto the shelf, increasing the nutrient load on the shelf [e.g., Chen et al., 2003a; Morimoto et al., 2009]. The upwelled nutrients would seem to non-linearly enhance PP [Dugdale, 1985] to a greater extent than when there is no shelfward movement of the Kuroshio current, resulting in a non-linear relationship between PPenh and V (Figure 2b).

[15] The highest PPenh was caused by Hai-Tang (10.71 Gg C), which had the slowest TS (3.18 m s−1), the highest southerly wind (12.33 m s−1, Figures 2a and 2b), and was followed by a clear shelfward shift of the Kuroshio axis (Figure 1k). The lowest PPenh (0.26 Gg C) was caused by Nock-Ten, which had a relatively fast TS (6.93 m s−1) and the strongest northerly wind (−10.46 m s−1). The northerly wind did not push the Kuroshio current axis shelfward, which meant that the upwelling of nutrients was unlikely to be enhanced [see Morimoto et al., 2009]. Other typhoon events (Bilis and Longwang) that caused a Kuroshio shelfward shift were also followed by clear Chl-a blooms, SST cooling, and PPenh bands (see auxiliary material).

5. Comparison of Typhoon-Enhanced Primary Productivity and New Productivity of the East China Sea

[16] We combined two non-linear relationships of PPenh against TS and PPenh against V into a single empirical equation indicating that the slower the TS and the higher the V, the higher PPenh will be. Rather than a linear model, we intuitively defined a non-linear model by placing PPenh – TS and PPenh – V equations as denominator and numerator, respectively as

equation image

to depict a non-linear response of phytoplankton to nutrient input [Dugdale, 1985].

[17] Iterative fitting routine to obtain optimum constant coefficients of the model was conducted by applying Type II reduced major axis linear regression (RMA) targeting the slope and intercept of the satellite- and equation-based PPenh scatter plots to 1.0 and 0.0, respectively. The RMA was used as satellite- and model-based PPenh are independent of each other [Laws and Archie, 1981]. With this equation, 88% of the variation in PPenh integrated over a selected region can be explained (Figure 2d), and it is thus possible to estimate PPenh even when ocean color data are not available. Sensitivity analysis indicated that PPenh was more sensitive to TS change rather than to V change, and the highest PPenh by Hai-Tang seemed to be largely associated with the slowest TS, rather than strongest V (data not shown).

[18] To understand the significance of the amount of PPenh, we compared it to new productivity (NP) over the entire region of the ECS (see auxiliary material) during the summer–fall period for each year from 1998 to 2007. The summer–fall NP was assessed first by integrating the monthly PP computed with SeaWiFS Chl-a and AVHRR SST data from June to October. Summer–fall PP was then multiplied by f-ratios of 0.10 and 0.15 respectively, for areas deeper and shallower than the 200-m isobath [Eppley, 1989; Chen et al., 2003b].

[19] The total PPenh during the typhoon season (June–October) from 1980 to 2007 was computed as follows: we derived V at YI and TS for all typhoons from June to October; applying the PPenh equation, we computed the PPenh for typhoons that had not been used to derive the PPenh equation and then summed all PPenh from June to October to obtain total PPenh. Comparing the total PPenh to the ECS summer–fall NP for each year within the 1980 to 2007 period, we concluded that PPenh due to typhoon passages in the southern ECS accounted for an average of 5.4% of the ECS summer–fall NP. In general, this amount of PPenh covaried with the number of typhoons (Figure 2e). The lowest PPenh (0.6%) was in 1993, a year with only 1 typhoon (Yancy, September 1993) which had strong northerly winds (−7.99 m s−1); the highest (11.8%) was in 2001, when a large number of typhoons occurred (six typhoons), one of which (Lekima, September 2001) had the slowest TS (1.43 m s−1). This range was considered to represent the minimum estimate of PPenh, as we did not consider the lifetime of a phytoplankton Chl-a bloom. Considering that a Chl-a bloom can last for several days or even weeks after a typhoon's passage [e.g., Babin et al., 2004; Siswanto et al., 2007], this range of PPenh could be several times larger, suggesting that typhoon passages over the southern ECS are important phenomena supporting NP in the ECS.

[20] Although, as also observed in the open ocean, PPenh after typhoon passages in the southern ECS might be associated with increased nutrients caused by mixing and/or upwelling, another mechanism also promotes upwelling of nutrients in the southern ECS. With prerequisite long-lasting southerly winds, the shelfward movement of the Kuroshio current axis seemed to be important in introducing more nutrients onto the shelf and hence further enhancing PP after typhoon passage, as previously suggested by Chen et al. [2003a].

6. Summary

[21] The increase in PP following typhoon passage in the southern ECS was revealed by satellite observation. Considering only region of low turbidity, we found that PPenh was significantly correlated with TS and V measured at YI. The slower TS and the stronger southerly winds that depended on the typhoon's trajectory led to greater PPenh, indicating that a long-lasting southerly wind is important in this process. We also found that high PPenh was associated with shelfward movement of the Kuroshio current axis, which is also believed to be driven by long-lasting southerly winds [Morimoto et al., 2009]. We therefore concluded that shelfward movement of the Kuroshio current axis with long-lasting southerly winds as prerequisite largely determined the variations in PPenh.

[22] The importance of long-lasting southerly winds in determining PPenh could be expressed by a single equation by which 88% of PPenh variation could be explained. Applying this PPenh equation, we concluded that typhoon-enhanced PP in the southern ECS accounted for ECS summer–fall NP within a minimum range of 0.6–11.8%, which could be several times larger depending upon the lifetime of the Chl-a blooms. This suggests that typhoon passage over the southern ECS is an important phenomenon supporting NP in the ECS.

[23] Although this study did not rely on numerical analysis that presumably would yield a detailed physical mechanism and hence improve the accuracy of the PPenh estimation, we offer an empirical approach to describe a long-lasting southerly wind as a proxy for the physical perturbation underlying PPenh variation due to typhoon passages. A great benefit of this method is that it can be applied independently of ocean color data, allowing us to examine the interannual variation in PPenh caused by typhoons in the southern ECS.


[24] This study was supported by a grant from the Japan Society for the Promotion of Science (JSPS) and also by the Collaborative Research Program of HyARC, Nagoya University, Japan. We are also grateful for helpful comments on this manuscript from two anonymous reviewers. We thank the Ocean Biology Processing Group (Code 614.2) at the GSFC, Greenbelt, Maryland, USA, for the production and distribution of the ocean color data.