Hypoxic Bottom Waters as a Carbon Source to Atmosphere During a Typhoon Passage Over the East China Sea

A high‐resolution mooring record from the Changjiang River plume (45‐m depth) is used to investigate how air‐sea CO2 flux responds to typhoon in the productive plume. With strong wind, surface partial pressure of carbon dioxide (pCO2) increased sharply from 369 to 606 μatm due to entrainment of high‐CO2 subsurface water. Though it was followed by pCO2 decrease of 250 μatm and Chl a increase days after the typhoon, the typhoon caused a net CO2 efflux overall. The maximum CO2 efflux (+111.6 mmol·m−2·day−1) is much greater than that under non‐typhoon condition (−2.3 to −11.7 mmol·m−2·day−1). Based on historical typhoon records, we estimate typhoon‐induced CO2 efflux to be +0.27 Tg C/year, which can cancel 18% of summer CO2 influx in the East China Sea shelf. It may likely occur in other coastal waters. Ignoring such contribution may induce large bias in estimating regional air‐sea CO2 flux.


Introduction
Despite constituting only 7% of the global ocean surface area, shelf seas account for 14-30% of the global primary production, 80% of the organic matter burial, and 50% of the calcium carbonate deposition (Gattuso et al., 1998); they are an important component of the global carbon cycling. Previous studies suggested that shelf waters are an important sink of atmospheric CO 2 , taking up CO 2 at a rate of 0.21-0.45 Pg C/year (P = 1 × 10 15 ; Borges et al., 2005;Cai et al., 2006;Chen et al., 2013;Laruelle et al., 2018). A large fraction of carbon absorbed from the atmosphere is respired in subsurface waters, which releases CO 2 into the water column (Cai et al., 2011;Chou et al., 2009). Thus, to maintain the net carbon sink of shelf seas, CO 2 -undersaturated surface layer and CO 2 -rich subsurface water should be separated (Chou et al., 2009;Thomas et al., 2004).
However, episodic events such as tropical cyclones (we use typhoons thereafter as our study area is the East China Sea shelf) can well mix the water column and cause significant CO 2 efflux in coastal waters within days (Crosswell et al., 2014;Huang & Imberger, 2010;Nemoto et al., 2009). Note that we will likely have more intense typhoons in a warmer future according to climate models (Webster et al., 2005). In shelf seas, CO 2 efflux during a typhoon is usually much higher than that of normal weather due to combined effect of mixing or upwelling of subsurface water with high dissolved inorganic carbon (DIC; Mathis et al., 2012;Ye et al., 2017) and extreme wind speed (Huang & Imberger, 2010;Nemoto et al., 2009). Such influence can be amplified when bottom shelf waters are hypoxic and enriched with CO 2 (Cai et al., 2011;Rabalais et al., 2014;Xue et al., 2015;Yu et al., 2014). Surface pCO 2 in the hypoxic inner East China Sea shelf is predicted to increase by 312 μatm after vertical mixing based on model simulation (Chou et al., 2009), which is much higher than that observed in the South China Sea shelf (increase by 20 μatm; Ye et al., 2017). Field observations in the Neuse River Estuary-Pamlico Sound suggested that extreme wind can trigger intense carbon efflux of +4,080 mmol C·m −2 ·day −1 in shallow bottom hypoxic waters (+2.4 mmol C·m −2 ·day −1 before typhoon; Crosswell et al., 2014). In spite of this, observations of pCO 2 variation and air-sea CO 2 flux during typhoons in the hypoxic shelf are still scarce.
The inner East China Sea shelf is an ideal place to study how typhoon influences air-sea CO 2 flux in hypoxic shelf waters. It represents one of the most studied eutrophic coastal waters and is also a typical sink of atmospheric CO 2 in summer (Guo et al., 2015;Tseng et al., 2014;Tsunogai et al., 1999;Zhai & Dai, 2009). Due to respiration of sinking organic carbon, increasing DIC and large-area hypoxia has been frequently observed in bottom waters Wang et al., 2016). CO 2 -undersaturated upper layer and carbon-rich subsurface waters are separated by strong stratification in summer (Ni et al., 2016;Zhai & Dai, 2009). A previous study reported large CO 2 efflux triggered by wind mixing in early autumn, which weakened stratification of the water column (Li et al., 2018). It is likely that carbon-rich subsurface waters in the inner East China Sea could also have significant impact on air-sea CO 2 flux in summer, considering that more than five typhoons visit this shelf each year.
In this study, we use time series of buoy data to evaluate how surface pCO 2 and air-sea CO 2 flux respond to vertical mixing in the inner East China Sea shelf and biological processes afterward. To the best of our knowledge, no high-resolution time series observations of pCO 2 variation during typhoon passage in the inner East China Sea shelf have ever been reported. In addition, we calculate annual air-sea CO 2 flux induced by typhoons in the whole East China Sea shelf, based on 22-year historical typhoon track record.

Buoy Deployment and Sample Measurements
Typhoon Chan-hom in 2015 is a Category-2 typhoon with wind velocity higher than 40 m/s. It entered the East China Sea on 7 July and left on 14 July. The variations of sea surface pCO 2 before, during, and after the typhoon were obtained using a moored buoy (3-20 July). It was deployed at a Changjiang River plume site, which was located on the typhoon track (water depth~45 m, 122.8°E, 30.6°N, supporting information Figure S1). Details of buoy observations were presented in Li et al. (2018) and Wang, Chen, Ni, et al. (2017). Our buoy also measured sea surface salinity, sea surface temperature (SST), surface dissolved oxygen (DO), surface chlorophyll a (Chl a), and 2-m wind. As our buoy wind direction data were not good enough, we also collected wind direction from National Centers for Environmental Prediction/National Center for Atmospheric Research reanalysis 10-m wind data (124°E, 31.3°N). In addition, bottom water temperature was measured. All sensors collected data every 15 min, except for pCO 2 which was measured at 30-min intervals. Details of instruments deployed are presented in the supporting information Text S1.
A Multi Water Sampler (Hydro-Bios) was used to collect discrete water samples in August 2013. In August 2017, water samples were collected with 5-L Niskin bottles mounted on a rosette assembly. Temperature and salinity were recorded using a Seabird SBE 917 profiler. DIC and total alkalinity (TA) samples were collected, stored and measured according to Cai et al. (2004). The precisions of the DIC and TA analyses were both ±2 μmol/kg. Certified reference materials from A.G. Dickson (Scripps Institution of Oceanography) were used to calibrate the analyses results. Discrete pCO 2 value of stations are calculated using the program CO2SYS based on the inputs of DIC, TA, salinity, and temperature (Pierrot et al., 2006). DO was determined using the Winkler titration method. Chl a samples were measured with a 10-AU Field Fluorometer (Turner Designs) after extraction with 90% acetone.

Satellite-Retrieved SST
We collected remote sensing SST data 2 days before and after the typhoon in the East China Sea from Remote Sensing System (http://www.remss.com/). The SSTs are Microwave Optimally Interpolated daily SST product with 9-km resolution. It combines the through-cloud capabilities of the microwave data with the high-spatial-resolution and near-coastal capability of the infrared SST data.
SSTs before and after 115 typhoons passing the East China Sea from 1998-2018 were collected. The sea surface coolings by typhoon mixing were evaluated by calculating ΔSST (posttyphoon SST minus SST before typhoon).

Typhoon-Induced Air-Sea Carbon Flux
Air-sea CO 2 flux is estimated using F gas = k × s × ΔpCO 2 where k is the gas transfer velocity, s is carbon dioxide solubility (Weiss, 1974), and ΔpCO 2 is the difference between sea surface pCO 2 and air pCO 2 , which is assumed to be a constant of 396 μatm (calculated from monthly air pCO 2 in July 2015 at Korea's Tae-ahn Peninsula site, ftp://aftp.cmdl.noaa.gov). The gas transfer velocity (cm/hr) is calculated according to Wanninkhof (2014).
We also estimate annual typhoon-induced air-sea carbon flux in the East China Sea. Total efflux of a single typhoon can be estimated as follows: total efflux (g C) = daily flux (mmol·m −2 ·day −1 ) × area (m 2 ) × duration (day) × 12 g/mol. Daily flux can be estimated using F gas = k × s × ΔpCO 2 . First, we categorize the East China Sea shelf into inner shelf and outer shelf, as ΔpCO 2 and air-sea fluxes induced by typhoons have sharp differences between the two regions. The reasons for such large differences are explained in section 3.2. We set ΔpCO 2 = 200 μatm in the inner shelf and ΔpCO 2 = 14 μatm in the outer shelf for every typhoon passage. Details of ΔpCO 2 chosen can be seen in the supporting information.
Because typhoon wind speed decreases from typhoon core to typhoon periphery, we decide not to use a single wind speed for a particular typhoon. In this study, we adopt a wind speed gradient: maximum wind speed, 25.7 m/s (50-knot wind, 1 knot = 0.514 m/s) and 15.4 m/s (30 knots; supporting information Table  S1). Each wind speed covers a specific influenced area which can be calculated by typhoon radius (R 30 or R 50 , typhoon radius with 30-or 50-knot wind). The details of area calculation can be seen in the supporting information Text S1.

Results and Discussion
Our 18-day mooring records showed large fluctuations of pCO 2 , from pretyphoon to posttyphoon period. Prior to the typhoon, the inner shelf was under strong northerly wind (maximum wind velocity > 11.5 m/s) on 7 July, which triggered strong vertical mixing already. We define the period of 3-7 July as Period I. Discussion of the effect of vertical mixing on pCO 2 is mainly based on the data in Period I (Figure 1). The typhoon passed our buoy during Period II (7-13 July). Period III (13-20 July) covers the time when the typhoon wind ceased and stratification was formed. As we focus on daily variation, high-frequency pCO 2 , DO, and Chl a data are firstly averaged into daily data.

Large Surface pCO 2 Variation During Typhoon Passage: Vertical Mixing and Biological Production
At the beginning of Period I (4-7 July), there was strong northerly wind (maximum wind velocity > 11.5 m/s). SST decreased by 1.32°C (to 20.53°C) within 3 days (4 to 7 July) when bottom temperature increased by 1.05°C (to 20.65°C, Figure 1b). Surface DO decreased by 35 μmol/L. Meanwhile, surface pCO 2 increased from 369 to 606 μatm (Figure 1c). Such pCO 2 increase (237 μatm) is similar to that predicted by Chou et al. (2009;242 μatm) but much higher than that reported in the South China Sea (20 μatm; Ye et al., 2017).
Large surface pCO 2 variation has been frequently observed in the shelf, which results from multiple processes: mixing and advection of water mass, biological activity, temperature variation, air-sea CO 2 exchange, etc. (DeGrandpre et al., 1998;Li et al., 2018). Surface temperature and bottom temperature approached each other (4 to 7 July, Figure 1b), indicating the mixed layer had deepened due to strong vertical turbulent mixing. Similar observation was reported by Nemoto et al. (2009) (Figure 2a). Such entrainment can explain the pCO 2 increase and DO decrease during Period I. Although we do not have subsurface samples before this vertical mixing event, vertical profiles in our past cruises always showed high pCO 2 and low oxygen in the subsurface water near our buoy site (Figure 2b). High-pCO 2 subsurface water has also been observed frequently in the inner shelf by other studies, due to intense respiration of organic matter (Chou et al., 2009;. The concurrent occurrences of vertical mixing and pCO 2 increase suggest that pCO 2 variation during Period I was probably controlled by entrainment of high-pCO 2 subsurface water. The dominant effect of vertical mixing or upwelling of DIC-enriched subsurface water on surface pCO 2 has also been observed in the East China Sea shelf (Li et al., 2018) and other coastal waters (Kortzinger et al., 1997;Mathis et al., 2012)  In the shallow coastal waters, stirring of sediment pore water DIC and in situ respiration of organic matter during typhoon may also be important in influencing pCO 2 (Crosswell et al., 2014). In this study, the sharp pCO 2 increase was probably dominated by vertical mixing of the two water masses, not by in situ respiration, as indicated by the significant negative relationship between pCO 2 and temperature ( Figure 2c). We can quantify pCO 2 increase due to mixing in the inner shelf. If we assume a thoroughly mixed water column of Stations 18-21 and 29 in Chou et al. (2009; the inner shelf), the DIC of surface waters will increase (ΔDIC) by 93-196 μmol/kg (using the trapezoidal rule). Using the Revelle Factor (RF) = 10 (Zhai & Dai, 2009), pCO 2 increase (ΔpCO 2 ) can be calculated by ΔpCO 2 ≈ ΔDIC/DIC × RF × pCO 2 = 124-295 μatm. In contrast, most studies suggested biological activity consumes DIC and decreases pCO 2 in the surface waters of the East China Sea shelf (Tseng et al., 2014;Zhai & Dai, 2009). Net production of DIC is generally reported in the subsurface water (Chou et al., 2009). If we adopt net community production of 30-64 mg·m −3 ·day −1 in the water column, as reported in Chen et al. (2003), DIC increase will be 7.8-15.6 μmol/kg assuming seawater density of 1.024 kg/m 3 and duration of 3 days. It is an order magnitude smaller than that of vertical mixing.
The typhoon traveled to our buoy site on 11 July; pCO 2 peaked during the following 2 days (576 to 602 μatm, Period II). It was probably because vertical mixing during 4-7 July had mixed the water column thoroughly (Figure 1b). From 7 to 9 July, advection of less saline water decreased the salinity by 2.4. We did not observe sharp pCO 2 variation similar to that during Period I. It can be explained as the horizontal gradient of pCO 2 was much smaller than that of the vertical pCO 2 gradient. In addition, vertical mixing also reduced the horizontal gradient of pCO 2 .
On 13 July, 2 days after the typhoon, relaxation of wind and increasing surface temperature favored the formation of stratification (Figures 1b and 1f). For the following 6 days (13-19 July, Period III), daily pCO 2 decreased by 230 μatm (from 602 μatm, Figure 1c). Supersaturated DO and increasing Chl a were both observed, while daily salinity stayed relatively constant in the coastal waters (26.25-27.27, Figures 1a, 1d,   Figure 2. Relationships between temperature and salinity (a) of buoy record, vertical profiles of DO (b), relationship between pCO 2 and temperature (c), and between npCO 2 and saturated DO (%; d) of buoy record. Colors in panels a and b denote pCO 2 value; colors in panels c and d denote Chl a. Filled circles are daily averaged data. In panel b, depth profiles of DO and pCO 2 were obtained near our buoy site. DO = dissolved oxygen. and 2a), suggesting strong influence of biological production. In addition to the mooring record, our cruise data on 17 July also showed high Chl a at the surface (3.61 μg/L), and even at 30 m (2.60 μg/L; supporting information Figure S3). Increased Chl a several days after the typhoon has been previously reported in the East China Sea shelf (Hung et al., 2010), South China Sea (Lin et al., 2003), and Gulf of Mexico (Shi & Wang, 2007). Furthermore, we observed significant relationship between pCO 2 and Chl a with a slope value of −34.49 (supporting information Figure S2), which is close to that obtained using underway data in the East China Sea shelf (−49.8;Tseng et al., 2014). Very significant relationship between pCO 2 and DO was also observed (Figure 2c), which was similar to previous studies (Li et al., 2018;Zhai & Dai, 2009). Such significant relationships suggest the controlling effect of biological production on surface pCO 2 during Period III. Biological activity after the typhoon could be related to nutrient supplies from vertical mixing (Hung et al., 2010;Shi & Wang, 2007) or rainfall. Heavy precipitations after a typhoon can carry terrestrial nutrients into coastal waters by river discharge (Meng et al., 2017), besides direct wet deposition. However, their effects on air-sea carbon flux are largely unknown. Intensive biological activity after the typhoon likely links to bottom hypoxia, as previous observations showed that DO of bottom water under the Changjiang River plume decreased by 109-137 μmol/L 1-2 weeks after wind mixing events (Ni et al., 2016;.

Hypoxic Subsurface Water as a Potential CO 2 Source to the Atmosphere
The maximum daily average air-sea CO 2 flux during vertical mixing was +111.6 mmol·m −2 ·day −1 using the gas transfer velocity parameterization of Wanninkhof (2014; Figure 1e), which is much greater than that under nontyphoon condition and in reverse direction (air to sea, −2.3 to −11.7 mmol·m −2 ·day −1 in four seasons; Guo et al., 2015;Zhai & Dai, 2009). Biological production several days after the typhoon decreased pCO 2 significantly and made the buoy site a weak carbon sink (−0.5 mmol·m −2 ·day −1 ; Figure 1e). However, the buoy site was still a strong source of atmospheric carbon if the whole typhoon period (11-20 July) was considered (Figure 1e). Large-scale biological bloom triggered by along-shore plume extension was observed 2-3 weeks after typhoon Chan-hom (Zhang et al., 2018). It can possibly reduce the typhoon-induced CO 2 efflux. Unfortunately, the effect of such posttyphoon phenomenon on typhooninduced air-sea carbon flux is difficult to be evaluated quantitatively. Recent studies showed that gas transfer velocity could vary significantly during strong wind due to bubble-mediated gas exchange (Prytherch et al., 2010). If using the expression of Prytherch et al. (2010), which spans a larger range of wind speed, the daily efflux could be up to +258.4 mmol·m −2 ·day −1 (Figure 1e). It should be significant in influencing regional air-sea CO 2 estimate in the East China Sea shelf.
We cannot simply apply such extreme carbon efflux to the entire shelf. Typhoon-induced carbon efflux in the inner East China Sea shelf (+111.6 mmol·m −2 ·day −1 ) in this study is higher than that in Wu (2015; +48.8 mmol·m −2 ·day −1 ) and in Nemoto et al. (2009;+21.2 mmol·m −2 ·day −1 ) in the outer shelf (supporting information Table S2), though typhoon wind speed in this study is lower than that in both of those studies. The increasing trend of carbon efflux per typhoon from the outer shelf to the inner shelf indicates complex spatial variation of carbon efflux during a typhoon. It can be explained as ΔpCO 2 (sea pCO 2 − air pCO 2 ) after vertical mixing in hypoxic waters of the inner shelf (192-207 μatm) is much higher than that in the outer shelf (13-48 μatm; Figure 3d and supporting information Table S2). Such extremely high ΔpCO 2 after a typhoon is probably related to bottom hypoxia in the inner shelf.
As explained in section 3.1, increasing sea surface pCO 2 in the inner East China Sea shelf during wind event was caused by mixing of surface waters with carbon-enriched subsurface waters. Extra CO 2 is added into the subsurface waters in the inner shelf by enhanced respiration of organic carbon, which consumes oxygen and releases CO 2 stoichiometrically (Cai et al., 2011;. Thus, hypoxic waters could have much more DIC when compared with nonhypoxic waters. The hypoxic subsurface waters (in the inner shelf) generally have DIC of~2,100 μmol/kg, and pCO 2 of~650 μatm (calculated using CO2SYS). In the summer of 2017, we even observed hypoxic waters with oxygen concentration < 20 μmol/kg near our buoy station (123°E, 31°N;personal communication with Yanyi Miao in October, 2017), which is the lowest value ever reported. The calculated pCO 2 of the bottom water is 1,022 μatm (Figure 3d), which is more than twice the air pCO 2 (396 μatm). In addition, hypoxia and carbon accumulation could be enhanced due to the combined effect of global warming and regional eutrophication (Rabalais et al., 2014;Wang et al., 2016). In contrast, the bottom waters of the outer shelf showed DIC concentration of~2,020-2,040 μmol/kg, and pCO 2 of~419-550 (Chou et al., 2009;.

Extensive CO 2 Efflux During Typhoon Passage
Remote sensing ΔSST (posttyphoon SST minus SST before typhoon) results showed that typhoons Chanhom (2015), Rumbia (2018), and Bolaven (2012) induced large cold patches along their tracks in the East China Sea shelf (Figures 3a, 3b and 3c). In addition, large carbon effluxes were observed during the passage of the three typhoons (supporting information Table S1). Statistical analysis of 115 typhoons during past 20 years also showed that the SST of 52% typhoons decreased by 2 to 4°C, indicating that subsurface waters were brought to the surface by typhoon passage in the shelf (Figure 3e).
Here, we estimate CO 2 efflux contributed by typhoon in the inner and outer East China Sea shelves. The boundary of the inner shelf (shallower than 50 m) is defined based on the historical hypoxia area of Wei et al. (2017) and the inner shelf boundary of Guo et al. (2015; Figure 3d). The calculation method is presented in section 2.3. Briefly, total efflux of a single typhoon can be estimated as follows: carbon efflux (g C) = daily flux (mmol·m −2 ·day −1 ) × duration (day) × area (m 2 ) × 12 g/mol. In air-sea CO 2 flux estimate, we use ΔpCO 2 of +200 μatm (in this study) for the inner shelf and + 14 μatm for the outer shelf (Nemoto et al., 2009; supporting information Table S2). Wind speed, duration, and area data are calculated based on the Japan Meteorological Agency best track data.
The calculated annual average typhoon-induced CO 2 efflux (averaged during 1997-2018) is +0.27 Tg (1 Tg = 1 × 10 12 g; Figure 3f), using the gas transfer velocity of Wanninkhof (2014). It can cancel out 18% of the summer influx in the shelf (−1.48 Tg, area of 453 × 10 3 km 2 ; Tseng et al., 2014;Figure 3f). Our results suggest that the air-sea carbon flux induced by typhoons could likely weaken the carbon sink of the East China Sea shelf in summer. In previous studies, typhoons were reported to enhance carbon efflux (Bates et al., 1998;Nemoto (c); ΔpCO 2 (sea pCO 2 − air pCO 2 ) and air-sea carbon fluxes during six typhoons in the inner and outer shelves (d); statistic analysis of ΔSST during typhoons from 1998 to 2018 (e); annual typhoon-induced carbon efflux from 1997 to 2018 (f). In d, flux data were obtained from our study, Nemoto et al. (2009), andWu (2015). The dash box and grey shade denote the area of the inner shelf and outer shelf in estimation, respectively. Bottom pCO 2 in the inner shelf (cruise in 2017) and outer shelf (cruise in 2007, obtained from Chou et al., 2009) are also shown. The blue dash line shows the 200-m isobath. In e, the lowest ΔSST for every typhoon was chosen for analysis. In f, red dash line and blue dash line are the annual-averaged typhoon-induced carbon efflux for the whole shelf and average carbon efflux under non-typhoon condition (obtained from Tseng et al., 2014), respectively. SST = sea surface temperature. et al., 2009). Carbon efflux contributed by typhoon in the western subtropical North Pacific accounts for 69-96% of the summer carbon efflux (Nemoto et al., 2009).
Such estimate still contains large uncertainties. First, ΔpCO 2 could have large variation due to different preconditions of ocean before typhoon, especially in the inner shelf. CO 2 in subsurface waters increases continuously during the development of hypoxia. Also, the string of sediment during strong winds could release DIC of pore water into the water column (Crosswell et al., 2014). Second, according to Hung et al. (2010), responses of Chl a and particulate carbon flux to typhoons could be distinct in the East China Sea. Though the role of posttyphoon phytoplankton blooms in influencing air-sea carbon flux is minor when compared to that of vertical mixing in our study, they can possibly reduce the effects of typhoons on releasing the CO 2 out of the ocean during other typhoon events or even enhance the carbon sink of the East China Sea shelf. Third, typhoon-induced rainfall brings massive terrestrial matter into coastal waters (Dadson et al., 2005;He et al., 2014). We still do not know much about respiration of such terrestrial organic carbon in the water column and the nutrient transports and whether they influence air-sea carbon efflux. Fourth, the relationship between gas transfer velocity and wind has large uncertainty as gas transfer velocity data during strong winds are still very scarce (Crosswell et al., 2014;Prytherch et al., 2010).
Such multiple processes and factors call more field observations of carbon system during typhoon events. And a combination of high-frequency mooring observations, remote sensing observations, numerical models, and ship-based observations is needed to produce accurate and precise typhoon-induced air-sea CO 2 flux estimates. Notwithstanding, our observation data in the inner shelf provide an opportunity to assess typhoon-induced air-sea CO 2 flux in the East China Sea. Strong efflux due to entrainment of high pCO 2 subsurface water suggested that subsurface DIC was a significant potential source of atmospheric carbon in the East China Sea shelf. It may likely occur in other coastal waters. There will be large biases in regional air-sea CO 2 flux if such potential sources are ignored in estimation.