Strong phytoplankton blooms are occasionally observed around a recurvature point of typhoon tracks in the western subtropical Pacific. These are noteworthy events in subtropical regions where both nutrient concentrations and biological production are persistently low. We investigated the response of phytoplankton to typhoon passage using a numerical model with/without biogeochemical processes. The model reproduced the observed patch-like phytoplankton bloom around a recurvature point of Typhoon Keith in 1997. The strong bloom is caused by the typhoon-centered upwelling of nutrient-rich water from below the euphotic layer, which supplies the nutrients required for phytoplankton growth, resulting in higher chlorophyll-a concentrations. Biogeochemical processes then play essential roles in determining the response after the passage of typhoons in subtropical regions.
 Many strong typhoons pass over the central region of the subtropical gyre in the western North Pacific. However, strong blooms are only rarely associated with typhoon passage. A strong phytoplankton bloom occurred around the recurvature point of Typhoon Keith on 5 November, 1997 at approximately 18°N, 136°E (Figure 1a). Its central pressure and maximum wind speed were 935 hPa and 45 m/s, and its average moving speed was 10 km/h (Figure 1b). This bloom had a radius of about 100 km and persisted for about one week after the typhoon passage, during which time the surface chlorophyll-a concentration estimated by Sea-viewing Wide Field-of-view Sensor (SeaWiFS) data increased from 0.04 mg/m3 to 0.24 mg/m3 (Figure 1c). Such a bloom was not found at the point where this typhoon had its maximum strength of 910 hPa on 3 November. Toratani  reported a similar rare bloom, which occurred in the stationary regions of Typhoon Ketsana in October, 2003.
 The upper oceanic layer in the western subtropical North Pacific is strongly stratified (e.g., World Ocean Atlas, 2009; available at http://www.nodc.noaa.gov/OC5/WOA09/pr_woa09.html). Nutrient concentrations are depleted ([NO3] < 1 μM) in the near-surface euphotic layer, and gradually increase with depth below 100 m. The chlorophyll-a concentration is approximately 0.05 mg/m3 at the surface, and a deep chlorophyll maximum (DCM) (∼0.2 mg/m3) occurs around the bottom of the euphotic layer (100–120 m depth), because photosynthesis depends on both nutrient concentration and light intensity. The concentrations and vertical structures vary little seasonally. Surface chlorophyll-a varies from 0.03 mg/m3 in spring to 0.08 mg/m3 in winter (Figure S1 in the auxiliary material). The mixed layer depth (MLD) remains shallower than the euphotic layer year-round: 35–40 m in summer and 80–90 m in winter [e.g., Longhurst, 1998; World Ocean Atlas, 2009]. Temperature gradually decreases with depth below the mixed layer.
 In subtropical regions, the differing vertical distributions of nutrient with temperature, i.e., nutricline much deeper than thermocline, might explain the rarity of phytoplankton blooms despite relatively more frequent decreases in sea surface temperature (SST). On the other hand, in the mid-latitudes, as both temperature decrease and nutrient increase with the depth below the mixed layer, tropical cyclones commonly cause both decreases in SST and increases in chlorophyll-a [e.g., Maeda, 1971; Price, 1981; Stramma et al., 1986; Sakaida et al., 1998; Son et al., 2006, 2007; Lin et al., 2009].
 In subtropical regions, surface chlorophyll-a concentrations after typhoon passages are higher than calculations based on the DCM concentration diluted with surface water, which suggests that biogeochemical processes contribute significantly to these blooms. Numerical simulations are useful for studying responses of chlorophyll-a and nutrients to typhoon passages, for which detailed observations by ships or satellites are often limited by severe weather or cloud cover. In a set of model experiments changing the strength, speed and size of storms, Wu et al.  found that the water mixing associated with moving storm passage induced significant increases in chlorophyll-a and nutrient concentration in the Labrador Sea. Liu et al.  reported that a patch-like bloom of phytoplankton after Hurricane Katrina's passage was due to a preexisting cold-core eddy caused by complex topography and currents. Their numerical models, however, did not explicitly include biogeochemical processes relevant to chlorophyll-a or nutrients, respectively.
 In this study, we focus on the role of biogeochemical processes during a typhoon-induced bloom by comparing simulation results with and without these processes, which were not included in previous studies. We particularly examine how the bloom depends on the moving speed and wind speed of the typhoon. We also discuss the relationships of the bloom both to upwelling at the typhoon's center and to deepening of the MLD by mixing.
2. Model Description
 We embedded biogeochemical processes in a three-dimensional physical model developed by Suzuki et al. . The model domain extends in the x-, y-, and z-directions to 1000 km, 2000 km, and 1800 m, respectively. The typhoon moves linearly toward positive y-direction at the center of the x-axis (Figure S2). The horizontal resolution is 10 km in both the x-, and y-directions. Vertically, the domain includes 18 layers between the surface and 150 m, whose thickness gradually increases with depth from 1 m (at the surface) to 30 m thick, and below this 12 layers from 37 m to 403 m thickness (Table S2). The improved Mellor-Yamada scheme (MYNN scheme [Nakanishi and Niino, 2004, 2006] is used for turbulence. For biogeochemical processes, we use a simple ecosystem model, based on North Pacific Ecosystem Model for Understanding Regional Oceanography (NEMURO) [e.g., Kishi et al., 2007], which includes nitrate, phytoplankton, zooplankton, and detritus as prognostic variables. Most parameters for biogeochemical processes were set to values typical for subtropical sites (Table S3). Maximum growth rates of phytoplankton and zooplankton were tuned to reproduce the observed vertical profiles of nitrate concentration and phytoplankton biomass.
 We assumed a middle-sized typhoon with idealized concentric wind stress. The radius of the maximum wind speed is fixed at 50 km from the center of the typhoon. For simplicity, we dealt with no heat and water fluxes through the sea surface, and no variations of light intensity and cloud cover affecting photosynthesis. Note that heat flux through the sea surface contributes much less to the SST cooling than do the mixing and upwelling [e.g., Price, 1981]. We obtained steady states, prior to typhoon passage, two weeks after the initial condition of the temperature, salinity, nitrate and chlorophyll-a concentrations in situ (137°E, 15–25°N) observed by the Japan Meteorological Agency (JMA) from 2003 to 2005 (temperature, salinity, nutrient, and phytoplankton as shown by Figure S1) [Japan Meteorological Agency, 2010].
 We calculated 17 cases with biogeochemical processes (hereafter termed BP cases) changing six different moving speeds, Ms from 1 m/s to 6 m/s (from 3.6 km/h to 21.6 km/h), and three different maximum wind speeds, Ws from 30 m/s to 50 m/s (see Table S1). As controls for comparison, two cases without biogeochemical processes (hereafter termed NO-BP cases) were calculated for moving speeds of 2 m/s and 6 m/s, respectively, with a wind speed of 40 m/s. We focused on slower moving speeds than the previous studies (e.g., 4 m/s to 12 m/s [Wu et al., 2007]), to study blooms around the recurvature point of the typhoon track.
 For slow moving speed (Ms = 2 m/s) and moderate maximum wind speed (Ws = 40 m/s), SST decreases by 2.4°C in Day 1 (one day after the typhoon passage), and returns to within 1.3°C of the pre-passage value by Day 10 (Figure 2a). The typhoon induces strong upwelling and vertical mixing. The upwelling brings deep water to the bottom of the mixed layer by Day 1 (Figure 2g). By Day 2, the strong wind deepens the mixed layer and mixes the upwelled water with the surface water (Figure 2b). The SST recovery is due to mixing with the surrounding waters, as the model does not account for heat exchange between the ocean and atmosphere. The actual SST recovery would be faster than in this model, due to solar heating and heat exchange with the atmosphere.
 We compared the BP and NO-BP cases during the same typhoon described above, to examine the effects of the biogeochemical processes. For both BP and NO-BP, surface nitrate and chlorophyll-a concentrations increase to 0.25 μM and 0.1 mg/m3 at Day 1, respectively, due to strong upwelling and vertical mixing (Figures 2c and 2e and Table S1). After Day 1, however, these concentrations for BP have different temporal profiles than for NO-BP, due to the nitrate uptake by phytoplankton in the former. The surface nitrate concentration for BP returns to its pre-typhoon level by Day 4, while that for NO-BP remains high even after Day 10 (Figure 2c). The chlorophyll-a concentration for BP rapidly increases and reaches a maximum concentration of 0.37 mg/m3 around Day 3, while that for NO-BP hardly changes after Day 1 (Figure 2e). The range of chlorophyll-a concentration simulated for BP (0.03 mg/m3 to 0.28 mg/m3 at Day 7) is comparable with that in the observations for Typhoon Keith (0.04 mg/m3 to 0.24mg/m3 at Day 7) (Figures 1d and 2e). Biogeochemical processes are responsible for most of the increase in chlorophyll-a after typhoon passage in subtropical regions.
 The increases in the surface nitrate and chlorophyll-a concentrations depends strongly on the typhoon moving speed, Ms; the chlorophyll-a concentration increases more drastically with decreasing moving speed than does SST (Figures 2 and S3). Compared with the surface chlorophyll-a concentrations of 0.03 mg/m3 before typhoon passage, typhoons with slow moving speeds result in large maximum values of 1.19 mg/m3 and 0.37 mg/m3 for Ms = 1 m/s and 2 m/s, respectively, while those with fast moving speeds have relatively low maximum values of 0.16 mg/m3 though 0.10 mg/m3 for Ms = 4 m/s though 6 m/s (Figures 2e and 2l and Table S1). The maximum wind speed, Ws, also affects chlorophyll-a and nitrate concentrations. When the maximum wind is strong (Ws = 50 m/s), surface chlorophyll-a concentrations reach 1.49 mg/m3 for Ms = 2 m/s, and 0.28 mg/m3 for Ms = 6 m/s, which are significantly greater than 0.37 mg/m3 and 0.1 mg/m3 for Ws = 40 m/s, respectively (Table S1). A weak typhoon can create a perceptible surface phytoplankton bloom if Ms is small. For a slow, weak typhoon (Ms = 1 m/s, Ws = 30 m/s), the chlorophyll-a concentration is 0.2 mg/m3. Strong blooms are produced by strong and/or stationary typhoons.
 Using the observed Ms and Ws (Figure 1c), we reproduced the satellite-observed chlorophyll-a data near the recurvature point of the typhoon track (Figure 1a). At the recurvature point, the observed increase in chlorophyll-a concentration due to typhoon passage is reproduced in the BP case but not in the NO-BP case (Figures 1d, 1e, and 1f). The simulated chlorophyll-a concentrations in BP case agree well with the observations on 9 to 11 November, although the simulated is slightly higher than the observed (Figure 1c). On the other hand, simulated chlorophyll-a in NO-BP is increased only by the upwelling of water located at DCM and is much lower than the observed. The time-integrated primary production produced by Typhoon Keith is estimated as 0.33 Mt C based on a C:cholorophyll-a of 1:50 (g/g). The value is of the same order as reported by Lin et al.  (0.8 Mt C), in which the biological production was enhanced by Typhoon Kai-Tak that stagnated in July 2000 in the South China Sea.
 In the absence of upwelling, deepening of the mixed layer shallower than 100m would cause SST to decrease but would not affect the surface nutrient concentration, because the SST gradually decreases with depth below the bottom of mixed layer, whereas the nitrate concentration is more uniformly depleted down to 100 m depth. Only when upwelling supplies nutrient-rich water from below 100 m to the mixed layer increases the surface nutrient concentration. To assess quantitatively whether the upwelled nitrate-rich water along the typhoon track is brought into the deepened mixed layer or not, we estimated: (1) the maximum accumulated upwelling (Umax), defined as the upwelling at 100 m depth which accumulates during the first upwelling phase of Ekman-pumping lasting for approximately one day, meaning the upward moving distance of the water mass that was at 100 m depth before typhoon passage (see Figures 2g and 2n), (2) MLD, and (3) their summation (Su+m = Umax + MLD) (Table S1). Since Su+m is clearly related to the decrease in SST and increases in nutrient and phytoplankton (Figure 3), Su+m is a convenient index for measuring the strength of response to the passage of typhoons having various moving speeds and maximum wind speeds. When Su+m is less than 100 m, SST is decreased, but nutrient and phytoplankton concentrations are hardly changed. On the other hand, when Su+m is much larger than 100 m, maximum concentrations of nutrient and phytoplankton increase with Su+m.
 Without biogeochemical processes, even if Su+m is much larger than 100 m, the surface chlorophyll-a concentration cannot reach the observed level (open triangle in Figure 3c), because the water transported from around the DCM into the mixed layer is diluted by mixing (i.e., the surface chlorophyll-a in NO-BP for the slow moving speed is less than 0.1 mg/m3). That is, biogeochemical processes are essential for the strong bloom when Su+m exceeds 100 m. On the other hand, when Su+m is less than 100 m, the observed increase in chlorophyll-a can be explained without biogeochemical processes. For a fast-moving (Ms = 6 m/s) typhoon of moderate maximum wind speed (Ws = 40 m/s), the chlorophyll-a increase for BP is more similar to that for NO-BP, compared to the slow-moving cases (Figures 2e and 2l).
 When Su+m is larger than 100 m, the DCM disappears and the maximum chlorophyll-a concentration occurs at the surface, where biological production is by far greatest. When Su+m is less than 100 m, the DCM remains at around 90 m depth and oscillates with a near-inertial period, with a slight decrease in its time-averaged depth and a slight increase in its concentration due to enhanced biological production as a result of improved light conditions (Figure 2m).
5. Conclusion and Remarks
 We investigated the phytoplankton response to the passage of typhoons using an idealized physical model coupled with biogeochemical processes. By assuming slow-moving typhoons, the model could reproduce the phytoplankton bloom, as observed by satellite around the recurvature point of Typhoon Keith in 1997. On the other hand, under the assumption of fast-moving typhoons, the model could not reproduce the dramatic increase observed in chlorophyll-a. Without biogeochemical processes, the increase in chlorophyll-a is much smaller than the observed, even for slow-moving typhoons. This clearly shows that, in addition to the supply of nutrients by upwelling and mixing, biological processes are essential determinants of the chlorophyll-a increase after typhoon passage in subtropical regions. The occurrence of phytoplankton blooms is determined by whether the sum of the mixed layer depth and total upward shift due to upwelling exceeds the nutricline depth, which is approximately 100 m in the western subtropical North Pacific.
 Our results imply that weak, slow-moving typhoons, which occur more frequently than stronger typhoons, may strongly impact subtropical ecosystems. Further studies are needed to estimate the total effects of typhoons on the annual biological production over the whole western subtropical North Pacific, and to investigate the structure of phytoplankton species at the local scale from the viewpoint of biodiversity, which may also be impacted by typhoons [Son et al., 2007]. Recent simulation studies of global warming scenarios suggest that stronger typhoons may become more frequent while the total number of typhoons may decrease [Oouchi et al., 2006]. Future studies should also investigate the potential impact of stronger typhoons on total biomass and production in the subtropical ocean.
 This study was supported by the Grant-in-Aid for Scientific Research in Priority Areas “Western Pacific Air-Sea Interaction Study (W-PASS)” under grants 18067003 and 18067004 from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. The authors were partially supported by the Global COE Program from MEXT and by Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency (JST). This research is a contribution to SOLAS, IGBP. We referred to in situ data, which are provided by Marine Division, Global Environment and Marine Department of JMA, for the initial conditions given by Hisayuki Y. Inoue. We also thank to Mitsuo Uematsu, Atsushi Tsuda, Shigenobu Takeda, Koji Suzuki, Naoki Yoshie, Masahiko Fujii and Takafumi Hirata for discussions, and S. Lan Smith for proofreading. We dedicate this paper to Walter Brain Barker.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.