Simulating the impact of typhoons on air‐sea CO₂ fluxes on the northern coastal area of the South China Sea

This study simulated four typhoons: Hato (1713), Mangkhut (1822), Nida (1604), and Merbok (1702). Track data for these typhoons were obtained from the Central Meteorological Observatory Typhoon Network (http://Typhoon.nmc.cn/web.html). To verify the model results and assess its ability to capture pCO_2sea characteristics, monthly average pCO_2sea retrieval data from 2003 to 2019 with a spatial resolution of 1 km (https://zenodo.org/records/7743187), provided by Song et al (2023), were used.

As figure 1 shows, this study utilized a coupled model (Luo et al 2023) based on the Coupled Ocean-Atmosphere-Wave-Sediment Transport modeling system. The model consists of a Weather Research Forecasting (WRF) module for the atmosphere and a Regional Ocean Modeling System (ROMS) module for the ocean (Warner et al 2008, Warner et al 2010). The WRF domain covers the Pearl River Watershed and northern part of the South China Sea, with a horizontal resolution of approximately 12 km. The ROMS domain includes the northern part of the continental shelf of the South China Sea and Pearl River Estuary, with a horizontal grid spacing ranging from 2.8 km to 40.7 km and an average resolution of approximately 4 km. The model had 30 vertical layers with a higher resolution near the surface and bottom boundaries. The vertical s-coordinate function was based on Shchepetkin and McWilliams (2009). The physical model used a recursive Multidimensional Positive Definite Advection Transport Algorithm three-dimensional advection scheme for tracers, fourth-order horizontal advection of tracers, third-order upwind advection of momentum, and the vertical mixing turbulence closure scheme of Mellor and Yamada (1982). The viscosity/diffusion coefficient for second-harmonic level mixing was set to 10⁻⁵ m²s⁻¹. The numerical integration had an internal mode time step of 900 s and an external mode time step of 60 s, as determined by the Courant–Friedrichs–Lewy criterion. The Model Coupling Toolkit facilitates the interaction between WRF and ROMS. WRF transfers fluxes to ROMS, including variables such as sensible heat fluxes, precipitation fluxes, surface stress components, and surface pressure. ROMS provides SST information to WRF. Furthermore, this model is coupled with a nitrogen cycle model that considers carbonate chemistry (Fennel et al 2006, 2008, 2011).

Zoom In Zoom Out Reset image size

Figure 1 indicates that in the northern coastal area of the South China Sea, the pCO_2sea in nearshore waters is lower than that in the inner estuary and offshore waters. This difference may be attributed to the intricate biological processes in this area (Song et al 2023). Therefore, we selected the coastal waters of the Pearl River Estuary as our research region. The area enclosed by the black dashed line represents the selected study area, while the red line represents the vertical profile line positioned at the center of the region.

This study examines the impact of temperature and non-temperature factors on the pCO_2sea during typhoons, based on the research conducted by Takahashi et al (2002):

$$\begin{align}{\text{npC}}{{\text{O}}_{{\text{2nt}}}} & = {\text{pC}}{{\text{O}}_{{\text{2sea}}}} \nonumber\\ & \quad\times \text{exp} \left[{\text{0}}{\text{.0423}} \times \left({\text{SS}}{{\text{T}}_{{\text{mean}}}} - {\text{SST}}\right)\right],\end{align} \tag{ 1 }$$

$$\begin{align}{\text{npC}}{{\text{O}}_2} & = {\text{pC}}{{\text{O}}_{2{\text{sea mean}}}} \nonumber\\ & \quad \times \text{exp} \left[{\text{0}}{\text{.0423}} \times \left({\text{SST}} - {\text{SS}}{{\text{T}}_{\text{mean}}}\right)\right].\end{align} \tag{ 2 }$$

We represented the changes in pCO_2sea caused by non-temperature and temperature effects using the variables npCO_2nt and npCO₂, respectively. The term pCO_2sea refers to the partial pressure of CO₂ at the ocean surface, whereas SST represents the SST in degrees Celsius. The subscript 'mean' indicates the average value.

During a typhoon, factors such as SST, Dissolved Inorganic Carbon (DIC), Total Alkalinity (TA), and SSS can be altered, leading to changes in the pCO_2sea through vertical mixing (Takahashi et al 2002, 2009, McKinley et al 2006, Ji et al 2022). To determine the factors controlling the changes in pCO_2sea during the typhoon period, the following equation was used (Takahashi et al 1993):

$$\begin{align}{\text{dpC}}{{\text{O}}_{2sea}} & = \frac{{\partial {\text{pC}}{{\text{O}}_{2sea}}}}{{\partial {\text{T}}}}{\text{dT}} + \frac{{\partial {\text{pC}}{{\text{O}}_{2sea}}}}{{\partial {\text{DIC}}}}{\text{dDIC}} \nonumber\\ & \quad + \frac{{\partial {\text{pC}}{{\text{O}}_{2sea}}}}{{\partial {\text{ALK}}}}{\text{dALK}} + \frac{{\partial {\text{pC}}{{\text{O}}_{2se{\text{a}}}}}}{{\partial {\text{S}}}}{\text{dS,}}\end{align} \tag{ 3 }$$

where dpCO_2sea is the change in pCO_2sea, and T, ALK, S denote SST, TA, and SSS, respectively.

The effect of each factor on pCO_2sea can be determined by the thermodynamics of seawater CO₂ (Takahashi et al 1993).

$$\begin{equation}\left(\partial {\text{pC}}{{\text{O}}_2}/\partial {\text{T}}\right)/{\text{pC}}{{\text{O}}_2} = {\text{0}}{\text{.0423}}\;^\circ {{\text{C}}^{ - {\text{1}}}},\end{equation} \tag{ 4 }$$

$$\begin{equation}\left(\partial {\text{pC}}{{\text{O}}_2}/\partial {\text{DIC}}\right)/\left({\text{DIC}}/{\text{pC}}{{\text{O}}_2}\right) = {\text{8,}}\end{equation} \tag{ 5 }$$

$$\begin{equation}\left(\partial {\text{pC}}{{\text{O}}_2}/\partial {\text{ALK}}\right)/\left({\text{ALK}}/{\text{pC}}{{\text{O}}_2}\right) = - {\text{7}}{\text{.4,}}\end{equation} \tag{ 6 }$$

$$\begin{equation}\left(\partial {\text{pC}}{{\text{O}}_2}/\partial {\text{S}}\right)/\left({\text{S}}/{\text{pC}}{{\text{O}}_2}\right) = {\text{0}}{\text{.93}}.\end{equation} \tag{ 7 }$$

Air‐sea CO₂ fluxes (${{\text{F}}_{C{O_2}}}$) in the model was calculated as follows:

$$\begin{align}{{\text{F}}_{{\text{C}}{{\text{O}}_{\text{2}}}}} = {\text{K}} \times {{\text{K}}_{\text{H}}} \times \left({\text{pC}}{{\text{O}}_{{\text{2sea}}}} - {\text{pC}}{{\text{O}}_{{\text{2air}}}}\right)\end{align} \tag{ 8 }$$

where K is the gas transfer velocity and K_H is the solubility of CO₂ in seawater (Weiss 1974, Wanninkhof 1992). pCO_2sea and pCO_2air are the partial pressures of CO₂ in the ocean and the atmosphere, respectively. A positive ${{\text{F}}_{{\text{C}}{{\text{O}}_{\text{2}}}}}$ indicates fluxes of CO₂ from the ocean to the atmosphere, meaning that the ocean acts as a source of CO₂ to the atmosphere. Conversely, negative ${{\text{F}}_{{\text{C}}{{\text{O}}_{\text{2}}}}}$ indicates fluxes of CO₂ from the atmosphere to the ocean, indicating that the ocean acts as a CO₂ sink.

We employed MAPE to assess the disparity between the model results (y_sim) and inversion data (y _i ) within the selected study area:

$$\begin{align}{\text{MAPE}} = \frac{1}{n} \times \mathop \sum \limits_{i = 1}^1 \left| {\frac{{{y_i} - {y_{{\text{sim}}}}}}{{{y_i}}}} \right| \times 100\% .\end{align} \tag{ 9 }$$

We used the weighted average method to calculate time series:

$$\begin{align}{\text{Weighted Average }} = \frac{{\mathop \sum \nolimits _{{\text{i}} = 1}^{\text{n}}\left( {{{\text{X}}_i} \times {{\text{A}}_i}} \right)}}{{\mathop \sum \nolimits _{{\text{i}} = 1}^{\text{n}}{{\text{A}}_i}}}\end{align} \tag{ 10 }$$

where X_i represents the value at each grid point of model outputs, A_i is the corresponding area of each grid point, n is the total number of grid points.

Previous studies have demonstrated a good performance in the physical and ecological simulations of our model (Luo et al 2023). In this study, we compared the simulated typhoon paths with the observed typhoon paths and used inverted ocean pCO_2sea data to further evaluate the simulation ability of the coupled model. The results are shown in figure 2.

Zoom In Zoom Out Reset image size

The simulated paths of typhoons are in good agreement with the observations (figure 2(j)). Typhoons Hato and Mangkhut mainly pass through the left side of the Pearl River Estuary, whereas typhoons Nida and Merbok mainly pass through the right side of the estuary. Both inverted and simulated data indicate that the pCO_2sea is high in the inner Pearl River Estuary and offshore waters and low in nearshore waters (figures 2(a)–(j)). These results demonstrate that the coupled model can effectively capture the spatial variations of the pCO_2sea. It is worth noting that the simulated pCO_2sea by the coupled model appears to be slightly higher than the inversion data, especially at the boundaries. This discrepancy may be attributed to a deficiency in oceanic boundary conditions.

As shown in figure 3, three time periods were determined for the analysis: when the typhoon entered the study area, when the typhoon was in the study area, and when the typhoon left the study area.

Zoom In Zoom Out Reset image size

When Typhoon Hato entered the study area, the distribution of CO₂ fluxes was uniform, and the study area was a weak CO₂ source (figure 3(b)). When Typhoon Hato was in the study area, the entire study area became a stronger source of CO₂ (figure 3(f)), and became a weak CO₂ source when Typhoon Hato left the study area (figure 3(j)). Throughout all three periods, the pCO_2sea values were high near Lingdingyang Bay (figures 3(c), (g) and (k)), possibly because of the effect of carbon-rich water from the Pearl River Estuary. When Typhoon Hato was in the study area, there was an obvious increase in pCO_2sea and CO₂ fluxes. During this period, the temperature of nearshore waters drops more significantly than that of offshore waters.

When typhoon Mangkhut entered the study area, the distribution of sea-air CO₂ fluxes was uniform, and the study area was a weak CO₂ source overall (figure 4(b)). When the typhoon was in the study area, with a lower wind speed and corresponding smaller CO₂ fluxes at the center of the typhoon, the entire study area became a strong CO₂ source (figure 4(f)). When the typhoon left the study area, it became a weak CO₂ source. When typhoon Mangkhut entered the study area, the pCO_2sea in coastal waters was lower than that in offshore waters. During the typhoon's presence, there was a uniform increase in the pCO_2sea, accompanied by an even drop in temperature, which is different from typhoon Hato.

Zoom In Zoom Out Reset image size

As shown in figure 5, when Typhoon Nida was in the study area, the entire area was a strong CO₂ source. The eastern part of the study area exhibits a more pronounced transfer of CO₂ from the ocean to the atmosphere. When the typhoon left the study area, it became a weak CO₂ source. During this period, pCO_2sea increased in the eastern part of the study area and slightly decreased in the western part; the decrease in SST was more significant in the eastern part of the nearshore waters. This could be because the typhoon mainly passed through the eastern side of the study area, resulting in stronger upwelling of deep water in the eastern part.

Zoom In Zoom Out Reset image size

When Typhoon Merbok entered the study area, the study area was a weak CO₂ source overall, became a stronger CO₂ source when the typhoon was in the study area, and then recovered to a weak CO₂ source when the typhoon left (figure 6). Notably, the CO₂ fluxes were lower in the center of the typhoon, where the wind was weak. When Typhoon Merbok entered the study area, the pCO_2sea in the nearshore waters was lower than that in the offshore waters. When the typhoon left, there was a more significant increase in the pCO_2sea on the eastern side of the nearshore waters. In addition, more significant changes in SST were observed in the eastern part of the study area.

Zoom In Zoom Out Reset image size

Before the passage of the typhoon, there was a positive correlation between temperature and pCO_2sea for Hato, Nida, and Merbok, but the situation was different for Typhoon Mangkhut (figure 7). One possible reason for this difference is that the pCO_2sea of the open ocean is higher than that in the study area. Before the passage of Typhoon Mangkhut, high pCO_2sea seawater from the open ocean flowed into the study area, causing a dilution of chla and a reduction in carbon consumption in the surface seawater, which resulted in a slight increase in pCO_2sea when the SST decreased. However, the specific reasons for this phenomenon require further investigation. In addition, U10 varied from 5 to 10 m s⁻¹, and the CO₂ fluxes ranged from 0 to 10 mmol⁻¹m⁻²d⁻¹. Generally, this region was a weak source of CO₂ during this period.

Zoom In Zoom Out Reset image size

During the passage of typhoons, npCO_2nt increased, while npCO₂ decreased (figure 8), which means that pCO_2sea driven by temperature increased, while pCO_2sea driven by non-temperature factors decreased.

Zoom In Zoom Out Reset image size

It is worth noting that, as the wind speeds increased and decreased, the typhoons seemed to have a two-step impact on the pCO_2sea and CO₂ fluxes (figure 8). In the first phase, during the initial passage of typhoons, the region experienced lower CO₂ content and higher SST. As U10 increased, the CO₂ fluxes increased. During this period, pCO_2sea increased for Hato, Mangkhut, and Merbok, whereas pCO_2sea decreased with fluctuations in typhoon Nida. In the second step, the concentration of CO₂ on the sea surface increased, the SST had already cooled to some extent, and the U10 and CO₂ fluxes also decreased. The growth of pCO_2sea slowed down for typhoons Mangkhut and Merbok, while the pCO_2sea showed a slight decrease for typhoon Hato. However, Typhoon Nida consistently caused a smooth decrease in pCO_2sea. Overall, the influence of the non-temperature factors on the change in pCO_2sea in the first step was more significant than that in the second step.

Notably, the pCO_2sea and npCO_2nt levels initially decreased and then increased for typhoon Merbok. Additionally, when npCO_2nt increased, pCO_2sea continued to decrease for a short period. This suggests that SST played a primary role in pCO_2sea during the initial limited time period of Typhoon Merbok. Furthermore, there was no significant slowdown in the temperature-induced pCO_2sea relative to non-temperature factors during the second step of Typhoon Nida (figure 8), indicating that SST may not only be affected by deeper water but may also be influenced by other factors such as heavy rainfall, which could cause a decrease in SST (Deng et al 2019, Feng et al 2021).

Considering the entire process, except for typhoon Nida, pCO_2sea increased, with non-temperature factors having a more significant impact on pCO_2sea than temperature factors. In addition, the region acts as a significant CO₂ source during typhoons.

After the passage of typhoons, pCO_2sea increased and showed a positive correlation with SST, indicating that SST was the main factor in the change in pCO_2sea. The U10 and CO₂ fluxes decreased to a lower positive value, indicating that the study area was restored to a weak CO₂ source. However, neither pCO_2sea nor SST fully recovered within 36 h of the typhoon's passage.

Changes in pCO_2sea are primarily influenced by temperature, DIC, TA, and salinity (Takahashi et al 1993). DIC, TA, and salinity increased with depth, whereas temperature decreased. During the passage of typhoons, vertical mixing and coastal upwelling occurred (figure 9), which could transport CO₂-rich and cool deep water to the surface. Coastal upwelling could bring lower temperatures deeper water to the nearshore (figure 9). Coastal upwelling was more pronounced during typhoons Hato, Nida, and Merbok, while vertical mixing was more pronounced during Mangkhut. This explains why temperature changes along the coast were higher than those in the offshore waters during Typhoons Hato, Nida, and Merbok, while SST changes were more uniform during Typhoon Mangkhut (figures 3–6).

Zoom In Zoom Out Reset image size

For typhoon Hato, the cooling of surface seawater resulted in a decrease of 35 μatm in the surface pCO_2sea. The changes in pCO_2sea caused by DIC and TA were +120 and −100 μatm, respectively. SSS caused an increase of 30 μatm in the pCO_2sea (figures 10(a)–(d)). The ratios of the changes in pCO_2sea caused by DIC, TA, and SSS to those caused by SST were 3.4, 2.9, and 0.9, respectively.

Zoom In Zoom Out Reset image size

For Typhoon Mangkhut, the cooling of surface seawater caused a decrease in pCO_2sea of 37 μatm. Surface DIC and TA contributed to a +72 and −37 μatm change in pCO_2sea, respectively, whereas salinity caused a +11 μatm change inpCO₂ (figures 10(e)–(h)). Furthermore, the ratios of changes in pCO_2sea caused by DIC, TA, and salinity to those caused by SST were 1.9, 1.0, and 0.3, respectively.

For Typhoon Nida, the cooling of the surface seawater led to a decrease in pCO_2sea by 30 μatm. The changes in pCO_2sea by surface DIC and TA were 43 and −34 μatm, respectively. Additionally, SSS caused a change of 8 μatm in the pCO_2sea (figures 10(i)–(l)). Furthermore, the ratios of the changes in pCO_2sea caused by DIC, TA, and salinity to those caused by SST were 1.4, 1.1, and 0.3, respectively.

For Typhoon Merbok, the cooling of the surface seawater led to a decrease in pCO_2sea by 14 μatm. Surface DIC and TA contributed to +25 and −12 μatm change in pCO_2sea, respectively. Salinity caused a +4.5 μatm change in pCO_2sea (figures 10(m)–(p)). The ratios of the changes in pCO_2sea caused by DIC, TA, and salinity to those caused by SST were 1.8, 0.9, and 0.3, respectively.

Overall, during the passage of typhoons, pCO_2sea driven by DIC and TA continuously increased, whereas pCO_2sea driven by TA and SST continuously decreased. Specifically, typhoon Nida caused a decrease in pCO_2sea (figure 7), and the increased amplitude of pCO_2sea driven by SST was similar to that driven by TA, suggesting that both TA and SST were important factors in the change in pCO_2sea. Conversely, for typhoons Hato, Mangkhut, and Merbok, pCO_2sea increased continuously (figure 7), and the changes in pCO_2sea driven by DIC were significantly greater than those driven by SSS, which indicates that DIC played a more prominent role in the changes in pCO_2sea.

Interestingly, for typhoon Hato, the ratio of the variation amplitude of pCO_2sea caused by DIC, TA, and SSS to the variation amplitude of pCO_2sea caused by SST was significantly higher than that of typhoon Mangkhut. This suggests that these non-temperature factors played more significant roles during Typhoon Hato than during Typhoon Mangkhut. Except for the initial oceanic conditions, this difference is also possibly due to the stronger vertical mixing during the passage of typhoon Mangkhut, whereas coastal upwelling was more pronounced during the passage of typhoon Hato (figures 3, 4 and 9). Coastal upwelling initially intensified and then weakened, with a vertical displacement of deep coastal waters, contrasting the vertical mixing of surface and deep waters. Furthermore, in terms of vertical distribution, DIC, TA, and SSS were higher along the coast than offshore at the same depth (figure 11). Coastal upwelling brought higher DIC, TA, and salinity to the nearshore ocean surface than vertical mixing.

Zoom In Zoom Out Reset image size

The impacts of the four typhoons on the CO₂ fluxes were significant. Before the passage of typhoons, the fluxes were approximately 5 mmol⁻¹ m⁻²d⁻¹, increased to a range of approximately 23–54 mmol⁻¹ m⁻²d⁻¹ during the passage of typhoons, and then decreased to approximately 6.0 mmol⁻¹ m⁻²d⁻¹ after typhoons passage. The maximum average CO₂ flux occurred during typhoon passage and increased approximately 6–14 times than that before typhoon passage (figure 12(a)). However, the change in ΔpCO₂ (pCO_2sea-pCO_2air) during the typhoon passage did not always increase (figure 12(b)). For example, ΔpCO₂ increased during the passage of typhoons Hato and Mangkhut, but slightly decreased during Typhoons Nida and Merbok. However, U10 always increased during the passage of typhoons (figure 12(c)). Therefore, the increase in CO₂ flux during the passage of typhoons is primarily due to the increase in wind speed (U10).

Zoom In Zoom Out Reset image size

Overall, although each typhoon has its own characteristics and may have different impacts on the pCO_2sea and CO₂ fluxes, these four typhoons significantly increased the amount of CO₂ emissions.