Acceleration of Warming, Deoxygenation, and Acidification in the Arabian Gulf Driven by Weakening of Summer Winds

Abstract

The Arabian Gulf (AG) exports hypersaline, dense waters into the Sea of Oman (SOO), replaced by fresher inflowing surface waters from the Indian Ocean. We investigate the impact of recent AG warming on its exchange with the SOO and the implications this has on the AG biogeochemistry. Using an eddy-resolving hindcast model simulation, we analyze the hydrography and biogeochemistry of the AG and the SOO from 1980 to 2018. Our study reveals that changes in summer surface winds have accelerated AG warming and weakened it in the SOO, reducing the density gradient and water exchange between the two seas during late summer. This has led to nutrient buildup, increased productivity, and heightened deoxygenation and acidification in the AG. These findings underscore how subtle wind changes can exacerbate the vulnerability of marginal seas to climate change and stress the need to properly represent regional winds in global climate models.

Key Points

• Recent changes in surface winds have accelerated warming in the Arabian Gulf (AG) while dampening it in the Sea of Oman (SOO)

• The faster warming in the AG has reduced exchange with the SOO, leading to a buildup of nutrients and biomass in the Gulf

• This increased productivity has caused a rise in respiration, thereby accelerating deoxygenation and acidification in the Gulf

Plain Language Summary

The Arabian Gulf (AG) (also known as Persian Gulf) produces dense, salty water that flows into the SOO, while it receives fresher water from the Indian Ocean. This study investigates how the recent rapid warming of the AG affects this exchange with the SOO and its impact on the Gulf's environment. Using a computer simulation to model the AG's evolution from 1980 to 2018, we discovered that changes in surface winds have warmed the AG and weakened its connection to the SOO during summer. This led to nutrient accumulation, increased micro-algae growth, decreased oxygen levels, and increased water acidity in the AG. These findings highlight how minor changes in wind patterns can exacerbate the effects of climate change in specific seas, emphasizing the need to improve the representation of local winds in climate models.

1 Introduction

The Arabian Gulf, also known as the Persian Gulf (hereafter AG), is a shallow semi-enclosed sea subject to a hyper-arid climate, characterized by intense evaporation that far exceeds both precipitation and runoff (Reynolds, 1993). This results in the prevalence of large areas of hypersaline waters (Vaughan et al., 2019) and an inverse estuary circulation, in which the dense Gulf water (Gulf Deep Water; GDW) is discharged at depth along the southern side of the Strait of Hormuz (hereafter the Strait) into the Sea of Oman (hereafter SOO), and is replaced by a surface inflow of fresher, lower-density waters from the Indian Ocean (Indian Ocean Surface Water; IOSW) along the northern side of the Strait (Chao et al., 1992; Reynolds, 1993; Swift & Bower, 2003). As direct observations of this exchange and its variability are scarce (Johns et al., 2003; Swift & Bower, 2003), numerous modeling studies have explored its dynamics, revealing its tight coupling to the density contrast between the inflowing IOSW and the outflowing GDW (Campos et al., 2020; Kämpf & Sadrinasab, 2006; Lorenz et al., 2020; Pous et al., 2015; Thoppil & Hogan, 2009; Yao & Johns, 2010). The recent rapid warming of the AG relative to the Arabian Sea (Al Senafi, 2022; Al-Rashidi et al., 2009; Hereher, 2020; Strong et al., 2011), which is expected to further accelerate in the future, is likely to impact the density gradient between the two seas and hence alter transport through the Strait (Kämpf & Sadrinasab, 2006; Paparella et al., 2022; Swift & Bower, 2003). Yet, little is known about the effects of such changes on the AG environment.

The AG is generally considered oligotrophic throughout most of the year because the new waters entering the Gulf mainly flow in at or near the surface, and thus are generally depleted in nutrients, except during winter when convective mixing in the northern Arabian Sea brings nutrients to the surface, triggering a winter bloom that enriches the Gulf source waters in nutrients and organic matter, fertilizing the Gulf in late winter (Al-Yamani & Naqvi, 2019). Consequently, the Gulf has been assumed to be relatively well oxygenated in its pristine state (Al-Yamani & Naqvi, 2019). Moreover, given its high levels of alkalinity, the Gulf is believed to have a high buffering capacity against ocean acidification, a process that can threaten the growth and maintenance of marine calcifiers, including coral reef calcifying organisms (Izumi et al., 2022; Purkis et al., 2011). Yet, recent observations challenge these assumptions. First, observational and modeling evidence suggests that summer near-bottom hypoxia has become regular, more intense, and widespread over recent decades (Al-Ansari et al., 2015; Lachkar et al., 2022; Saleh et al., 2021). Second, observations indicating rapid progression of ocean acidification in the AG have been reported (Uddin et al., 2012). The concomitant occurrence of ocean deoxygenation and ocean acidification may further exacerbate the vulnerability of the AG ecosystems, including its coral reefs, to ongoing warming (Burt & Paparella, 2023; Burt et al., 2019; de Verneil et al., 2021; Purkis et al., 2011).

The drivers behind these biogeochemical changes and the eventual role of the rapid warming of the AG in their emergence remain poorly understood. In particular, the role of altered exchange between the AG and the SOO in these changes remains unexplored. This paper addresses the following key questions: (a) What are the implications of the rapid warming of the AG for the Gulf outflow? (b) How do these changes contribute to recently reported biogeochemical changes in the Gulf such as hypoxia expansion and acidification? and (c) what are the mechanisms responsible for the faster warming in the AG relative to the SOO? While the lack of comprehensive observational surveys of the biogeochemical properties of the Gulf waters makes documenting and understanding the ongoing biogeochemical changes challenging, here we employ a state-of-the-art coupled-physical biogeochemical model of the Gulf and the Arabian Sea region to reconstruct the evolution of the hydrography and biogeochemistry between 1980 and 2018. We demonstrate that changes in surface wind have contributed to amplifying the warming of the AG relative to the SOO, thereby reducing the density gradient across the Strait and the exchange between the two seas over the study period. Consequently, water residence times in the Gulf increased alongside respiration, leading to exacerbated deoxygenation and acidification in its deeper portions.

2 Methods

The circulation model is based on the Regional Ocean Modeling System (ROMS) (Shchepetkin & McWilliams, 2005). The model uses non-local K-profile parameterization scheme for vertical mixing (Large et al., 1994). Covering the Indian Ocean from 31.5$${}^{\circ}$$S to 31$${}^{\circ}$$N and 30$${}^{\circ}$$E to 120$${}^{\circ}$$E, the model employs a horizontal resolution of 1/10° and 32 sigma-coordinate vertical layers, with enhanced resolution near the surface. Biogeochemical processes are simulated using a nitrogen-based NPZD model with two nutrient components (nitrate and ammonium), one phytoplankton, one zooplankton, and two detrital classes (Gruber et al., 2006). Additionally, the model incorporates a module describing the oxygen cycle (Lachkar et al., 2021) and a carbon module with dissolved inorganic carbon (DIC), total alkalinity (TA), and calcium carbonate state variables (de Verneil et al., 2022). A more detailed description of the model is available in Supporting Information S1. The hindcast simulation is forced by ECMWF ERA-Interim 6-hourly heat fluxes, air temperature, pressure, humidity, precipitation, and winds spanning January 1980 to December 2018. Initial and lateral boundary conditions for various parameters are derived from ECMWF Ocean Reanalysis System 5 (ORAS5; Zuo et al., 2019), World Ocean Atlas 2018 (Garcia et al., 2019), and GLODAP version 2 (Lauvset et al., 2016). Atmospheric pCO₂ data is obtained from Mauna Loa (Joos & Spahni, 2008; Keeling et al., 2005). The model is spun-up for 69 years, after which two simulations are conducted: a control hindcast run forced with increasing atmospheric carbon and interannually varying momentum, freshwater, and heat fluxes from 1980 to 2018, and a constant climate (CC) simulation forced with climatological forcing (repeated normal year) and increasing CO₂. The CC simulation serves to quantify model drift and disentangle the roles of climate change and rising atmospheric CO₂ levels in reported acidification. Our analysis reveals a negligible model drift in the study area. Further details of the model setup and the evaluation of model drift are provided in Text S2 in Supporting Information S1.

We evaluate the model's performance in reproducing key aspects of the region's hydrography and biogeochemistry using the limited available data (Figures S1–S8 in Supporting Information S1). Overall, we find that despite some local discrepancies, our model generally captures the essential hydrographic features of the Gulf region, including the seasonal progression of temperature, salinity, and the Gulf outflow. Similarly, our model also aligns relatively well with available data regarding oxygen levels, vertical distribution of chlorophyll, seasonal variability in biological production, as well as the state of the carbonate system (DIC, TA, pCO₂, and pH). A detailed description of the model evaluation is provided in the SI (Text S3, Tables S1–S2, and Figures S1–S8 in Supporting Information S1).

To characterize the hydrography of the AG and the SOO, we averaged physical properties such as temperature, salinity, and potential density across each sea and calculated their differences. To capture maximum variability, these variables were considered at the surface. The Gulf water outflow was calculated as the zonal eastward current, integrated vertically and meridionally at the Strait (56.4$${}^{\circ}$$E). To characterize the biogeochemistry of the AG, we analyzed vertically integrated net primary production, total nitrogen transport into the AG, and O₂ and pH near the seafloor (deepest model layer), where they reach their lowest values. Finally, to investigate long-term changes in the hydrography and biogeochemistry of the AG and the SOO, data were deseasonalized by removing monthly climatologies from the original time series. Linear trends were determined from the slope of the least squares regression line. The statistical significance of these trends was assessed using a Student's t-test.

3 Results

3.1 Seasonal and Interannual Variability in the AG and the SOO

To put the long-term trends in the AG in broader context, we first characterize the seasonal and interannual variability of both the physical and biogeochemical properties within the AG and the SOO. The temperature in both the AG and the SOO exhibits much stronger seasonal variability than salinity (Figure 1a–1c). Consequently, the seasonal cycle in density in both seas appears to be primarily driven by temperature fluctuations throughout the year rather than variations in salinity. An analysis of the sensitivity of in-situ density to temperature and salinity variations indicates that temperature fluctuations account for around 90% of the seasonal amplitude in density in both the AG and the SOO. While both the AG and the SOO display strong seasonal variability in temperature, the amplitude of this variability is notably stronger in the AG compared to the SOO. This leads to the amplitude of the variability in surface density being nearly twice as large in the AG relative to the SOO (Figure 1e). Furthermore, the Gulf exhibits significantly higher spatio-temporal variability on interannual timescales vis-a-vis the SOO, with a standard deviation 3 to 4 times larger for winter temperature and 5 to 10 times larger for salinity throughout the year (Figures 1a and 1c). These differences are also reflected in surface density, which displays stronger interannual variability in the AG relative to the SOO, with a standard deviation 3 to 4 times larger in the former (Figure 1e). Even when excluding spatial variability by averaging variables across each basin, interannual variability in spatially-averaged properties remains 50% to nearly twice higher in the AG relative to the SOO for most of the year (Table S3 in Supporting Information S1).

Figure 1

Seasonal variability in the Arabian Gulf (AG) and Sea of Oman (SOO). Seasonal cycle of sea surface (a) temperature, (c) salinity, (e) density, (g) density gradient (superimposed on the Gulf outflow at the Strait), as well as (b) net primary production (NPP), (d) oxygen, (f) pH, and (h) transport of total nitrogen through the Strait. Oxygen and pH are shown near the seafloor below 40 m in the AG and at 100 m in the SOO. NPP is vertically integrated while the density gradient is calculated from the difference between the AG- and the SOO-averaged surface densities. The solid lines depict the seasonal averages while the dashed lines show interannual variability envelope defined by $$\pm $$ SD. Time series in A-F are area-averaged properties (AG and SOO spatial definitions are shown in Figure 2). Temperature and salinity variations in panels (a and c) are scaled to their contributions to density variations shown in panel (e). The shaded area depicts the first 2 months of the following annual cycle.

The strong variability in AG water density results in notable seasonal changes in the density gradient between the AG and the SOO, with the density contrast between the two seas peaking in winter and reaching its lowest point in summer (Figure 1g). The density gradient across the Strait is the primary driver of the overturning circulation, which transports light surface waters from the SOO into the Gulf and dense, deep Gulf waters into the SOO (Swift & Bower, 2003). Consequently, fluctuations in this gradient are expected to cause fluctuations in the strength of the Gulf water outflow (e.g., Kämpf & Sadrinasab, 2006; Paparella et al., 2022; Swift & Bower, 2003). Our analysis of the Gulf outflow at the Strait indeed reveals seasonal variability similar to that of the density gradient, albeit with a time lag of 2–3 months (Figure 1g). Maximum outflow occurs in spring, while minimum outflow is observed in late summer and early autumn (Figure 1g). This finding aligns with evidence from previous observational and modeling studies that have investigated the seasonal variability in Gulf outflow (e.g., Johns et al., 2003; Kämpf & Sadrinasab, 2006; Lorenz et al., 2020).

The pronounced variability in hydrography is accompanied by similarly strong seasonality in biogeochemistry in both the AG and the SOO (Figure 1b–1h). Despite faster photosynthetic growth rates in summer driven by higher temperatures, the stronger winter vertical mixing and the net inflow of nutrients from the SOO result in greater biological production in the Gulf during winter than in summer (Figure 1b–1h). Similarly, in the SOO, biological production is higher in winter due to increased nutrient supply from depth because of winter convective mixing in the northern Arabian Sea (Naqvi et al., 2002). Finally, both dissolved oxygen and pH exhibit strong seasonality at depth in both seas, with maximum values occurring in late winter and minimum values in late summer when oxygen deficit and carbon excess driven by respiration accumulate at depth due to limited vertical mixing (Figure 1d–1f).

3.2 Contrasting Rates of Warming in the AG and the SOO

We model a rapid warming trend in both the AG and the SOO, consistent with observations (Figures 2c and 2d). While much of the northern Arabian Sea has experienced warming in recent decades, the rate of warming in the AG is nearly 50% faster than in the rest of the Arabian Sea (Figure 2b). During winter, the warming in the AG is not statistically significant and is comparable in magnitude to that of the SOO (Figures 2c and 2d). However, during summer the warming in the Gulf is statistically significant and up to three times faster than that observed in the SOO (Figures 2c and 2d). This results in an enhanced temperature gradient between the two seas for most of the year, except during winter (Table S4 in Supporting Information S1). Along with changes in the surface salinity gradient, this leads to uneven variations in the density contrast between the two seas (Figure 3a, Table S4 in Supporting Information S1). Over the study period, the density contrast slightly increases in early winter, primarily due to enhanced salinity contrast, but significantly decreases during summer due to the faster warming of the AG compared to the SOO (Table S4 in Supporting Information S1). Consequently, the Gulf outflow strengthens slightly (+3% per decade) in winter but weakens significantly (up to −10% per decade) in late summer and early autumn (Figure 3a).

Figure 2

Warming and surface wind changes in the Arabian Gulf (AG) and Sea of Oman (SOO). (a) Average summer (JJA) sea surface temperature (SST; in $${}^{\circ}$$C) in the northern Arabian Sea as simulated in the model over the study period (1980–2018). (b) Linear trends in summer (JJA) SST (in $${}^{\circ}$$C per decade) in the AG and northern Arabian Sea. The hatching indicates statistically significant trends at 95% confidence interval. (c and d) Trends in AG-averaged SST (c) and SST gradient between the AG and the SOO (d) during winter (blue), summer (red) and annual-mean (purple) based on the ROMS simulation and from different data products. (e and f) Trends in AG-averaged surface wind speed (e) and the difference in surface wind speed trends between the AG and the SOO (f) during winter (blue), summer (red) and for the month of August (pink) based on different atmospheric reanalyses products. White stars and triangles indicate statistically significant trends at 95% and 90% confidence levels, respectively.

Figure 3

Long-term changes in the Arabian Gulf (AG) hydrography and biogeochemistry. (a) Trends (1980–2018) in AG-averaged SST (red), salinity (blue), density (purple), density gradient with the Sea of Oman (violet) and Gulf outflow (green). (b) Trends (1980–2018) in AG-averaged net primary production (red), oxygen (blue), pH (violet) and transport of total nitrogen through the Strait (green). Oxygen and pH are shown for the deepest model layer in the AG (near the seafloor). Filled (open) circles indicate statistically significant (non-significant) trends at 95% confidence interval. Multiple variables on each chart share the horizontal axis (months of the year) but are shown on different vertical axes. The shaded area depicts the first 2 months of the following annual cycle.

3.3 Rapid Changes in the Biogeochemistry of the AG

Changes in the biogeochemical properties of the AG also display significant variations among the seasons (Figure 3b). For example, nitrogen supply to the Gulf shows a notable increase in early winter (December), coinciding with a significant intensification of water exchange (Figure 3b). However, nitrogen supply to the Gulf also experiences a significant increase during late summer and early autumn (September to October) despite a weakening of the Gulf inflow/outflow driven by warming (Figure 3b). To comprehend this apparent paradox, one must consider not only the strength of the overturning circulation but also the vertical gradient of nutrients at the Strait (Figure S10 in Supporting Information S1). In contrast to winter, strong summer stratification induces a pronounced vertical gradient in nitrogen at the Strait, with very low concentrations near the surface and significantly higher levels below 20 m. Consequently, the net nitrogen flux associated with the overturning circulation in summer results in a loss for the AG (outflow waters have a higher nitrogen content than inflow water). Conversely, in winter, surface inflow waters have a higher nitrogen content compared to outflowing waters (Figure S10 in Supporting Information S1), leading to a net supply of nitrogen to the Gulf (Figure 3b). Consequently, the significant decrease in outflow during late summer, as well as the increase in early winter, both contribute to an increase in the supply of nitrogen to the Gulf. This buildup of nutrients and biomass in the AG, combined with faster photosynthetic growth rates driven by higher temperatures, leads to enhanced biological production throughout most of the year (Figure 3b). The increase in productivity enhances respiration, particularly in the benthos, resulting in increased consumption of oxygen and the release of carbon dioxide at depth near the seafloor (Figure S11 in Supporting Information S1). As the stratification increases in summer, the supply (release) of O₂ (CO₂) from (to) the surface diminishes, causing depletion (accumulation) of O₂ (CO₂), and thus an expansion of hypoxia and low-pH waters near the bottom. This increased acidification, driven by enhanced respiration, compounds the background acidification caused by rising atmospheric CO₂ levels (CC), amplifying it by up to 50% in late summer (Figure 3b).

3.4 Drivers of the Rapid Warming of the AG

An examination of the heat budget in the AG throughout the study period indicates that most of the simulated temperature changes are driven by alterations in atmospheric fluxes, while the lateral transport of heat to and from the SOO has a minor impact (Figure 4a). To understand the processes behind changes in atmospheric forcing, we analyze changes in the individual components of the atmospheric heat fluxes (Figure 4b–4d). Changes in downward longwave radiation, likely due to increased water vapor in the lower atmosphere (Figure S12d in Supporting Information S1), contribute to summer warming in both seas, while changes in sensible heat fluxes and incoming shortwave radiation play a negligible role (Figure 4b and Figure S12a in Supporting Information S1). In contrast, changes in latent heat fluxes amplify warming in most of the AG but dampen it in the SOO (Figure 4c), thus explaining much of the differential warming between the two seas. It's important to note that the contribution of latent heat fluxes to the warming of the Gulf varies spatially and temporally due to high interannual variability. This is consistent with previous works that found that interannual fluctuations in total heat fluxes over the Gulf are dominated by fluctuations in the latent heat fluxes (e.g., Paparella et al., 2019; Pous et al., 2015). Generally, the contribution of latent heat fluxes is more pronounced in the central Gulf and during late summer (August) (Figure S13 in Supporting Information S1). In winter, evaporative cooling decreases in both the AG and the SOO, contributing to warming trends in both seas (Figure S13 in Supporting Information S1). These long-term trends in latent heat fluxes are primarily driven by changes in surface wind speed in the region (Figures 2e and 2f; Figures S12c and S14 in Supporting Information S1). While northwesterly Shamal winds have weakened over the AG during both winter and summer, surface winds over the SOO have weakened during winter and increased during summer, contributing to the differential warming in the region in recent decades (Figure 2; Figures S14 and S15 in Supporting Information S1). The weakening of the Shamal winds may be linked to the strengthening of the Arabian thermal heat low over the Arabian Peninsula during summer in response to increased surface temperatures as recently reported by Fonseca et al. (2022).

Figure 4

Drivers of Arabian Gulf (AG) warming. (a) Heat budget in the AG showing the contribution of atmospheric fluxes (orange) and lateral fluxes (magenta) to the net temperature change (black) over the study period. (b–d) Trends in downward radiation (b), latent (c), and sensible (d) heat fluxes over the study period (in W m$${}^{-2}$$ per decade; positive fluxes indicate ocean heat gain). Hatching indicates areas where trends are statistically significant at 95% confidence level.

4 Discussion

4.1 Comparison With Previous Works

Our study reveals significant variability in the exchange flow between the AG and the SOO, primarily driven by fluctuations in atmospheric forcing over the AG. This finding aligns with the research conducted by Lorenz et al. (2020) and Pous et al. (2015), who concluded that interannual variability in Gulf outflow is primarily influenced by surface fluxes.

We found the warming rate of the Gulf to be on average around 0.26$${}^{\circ}$$C per decade but with a strong spatiotemporal variability with local warming rates ranging from less than 0.2$${}^{\circ}$$C per decade in winter to above 0.5$${}^{\circ}$$C per decade during summer in much of the western Gulf. Other studies also reported important warming in the Gulf with rates varying between 0.2$${}^{\circ}$$C and 0.7$${}^{\circ}$$C depending on the region and period considered (Al Senafi, 2022; Al-Rashidi et al., 2009; Hereher, 2020).

Finally, in a recent modeling study, Vasou et al. (2024) studied the changes in the heat content of the AG over the period between 1993 and 2021. While they showed that the interannual variability in the heat content is dominated by the surface heat fluxes, they also suggested that the long-term warming of the basin is primarily driven by enhanced heat transport from the Arabian Sea because of a simulated increase in the annual mean volume of waters being exchanged at the Strait. Contrary to Vasou et al. (2024), our findings indicate that long-term warming trends in the AG are driven by changes in atmospheric heat fluxes, similar to seasonal and interannual variability. We did not observe an increase in the overturning circulation of the AG. Instead, we found a significant decrease in the volume of Gulf outflow in summer and only a slight increase in winter. This aligns with the warming-induced reduction in density contrasts between the two seas, the primary determinant of outflow strength according to theory (e.g., Bryden & Stommel, 1984; Pratt & Lundberg, 1991) and previous studies of the Gulf circulation (e.g., Kämpf & Sadrinasab, 2006; Lorenz et al., 2020; Paparella et al., 2022; Pous et al., 2015; Swift & Bower, 2003). The recent decrease in density gradient and exchange strength is further corroborated by evidence from ORAS5 reanalysis and multiple versions of the SODA reanalysis, all indicating a notable reduction in density contrast between the AG and the SOO, along with decreased Gulf outflow intensity in recent decades, particularly pronounced in summer (Figure S16 in Supporting Information S1). It is important to mention that simulated surface temperature trends in Vasou et al. (2024) significantly underestimate observed trends in the AG and overestimate them in the SOO (refer to their Figure S1 in Supporting Information S1).

4.2 Caveats and Limitations

While large-scale changes in atmospheric conditions in the Gulf region are relatively robust across multiple data-based products (Figure 2 and Figure S14 in Supporting Information S1), they still harbor significant uncertainties, particularly at smaller, local scales. These uncertainties arise from the limited availability of direct observations and the relatively low spatial and temporal resolution of existing atmospheric forcing products. To enhance confidence in modeling the AG circulation and biogeochemical changes, it is important to employ better-resolved atmospheric forcing, validated via local in-situ observations, in future studies. Furthermore, the simulated alterations in Gulf outflow, productivity, and nutrient availability require confirmation through in-situ observations. Repeated measurements of critical physical parameters, such as the intensity of water exchange through the Strait, and vital biological parameters, such as chlorophyll and nutrient concentrations, are essential for documenting ongoing changes in this region, which is both under-sampled and under-studied. Finally, employing higher resolution models capable of fully resolving both mesoscale and submesoscale eddies and filaments could be crucial for enhancing our comprehension of the dynamics and vulnerability of the Gulf to ongoing global change.

4.3 Implications and Recommendations

Our findings indicate that both the physical and biogeochemical properties of the Gulf exhibit significant variability across seasons and years. Consequently, measurements taken over short periods may not accurately capture the climatological conditions in the AG. This highlights the importance of employing high-resolution monitoring or continuous sampling methods, such as through the use of oceanographic moorings. Moreover, the monitoring of key variables, such as the strength of the Gulf outflow, is particularly crucial due to its critical role in the ecology and biogeochemistry of the Gulf, as evidenced in this study. The demonstrated strong relationship between Gulf outflow variability and density contrasts between the two seas underscores the potential for monitoring this gradient as a proxy to gain insights into exchange flux and its variability.

The export and subduction of Gulf waters into the northern Arabian Sea have been shown to deeply affect the intensity of the Arabian Sea's oxygen minimum zone (OMZ). Notably, the reduction in the intensity of the Gulf outflow due to surface warming has been linked to a recent decline in the ventilation of the upper layers of the OMZ (Lachkar et al., 2021), a trend expected to intensify in the future (Ditkovsky et al., 2023; Lachkar et al., 2023; Vallivattathillam et al., 2023). These changes are likely to impact biological productivity and the carbon cycle in the northern Arabian Sea by affecting denitrification and, consequently, nutrient availability in the region (Lachkar et al., 2016; Lévy et al., 2022). In this study, we demonstrate that a reduction in the Gulf outflow not only significantly influences the biogeochemistry of the Arabian Sea but also deeply impacts the biogeochemistry of the Gulf itself.

The findings of this study have important implications for biodiversity and socio-economics in the Gulf. As a marginal marine system characterized by already extreme environmental conditions (Burt et al., 2020), further pressures such as those identified here have great potential to result in sudden, non-linear impacts on marine organisms and ecosystems (Bouwmeester et al., 2020). Marine organisms in the Gulf are considered to live very near to their thermal tolerance thresholds in summer, and we are already witnessing an increasing frequency and severity of coral bleaching and mass mortality events during summers when low wind conditions permit temperatures to rise by just 2$${}^{\circ}$$C above the normal summer maximum (Burt et al., 2019; Riegl et al., 2018). As ectothermic fauna whose metabolic oxygen demand is directly tied to temperature, fishes are likely to face considerable physiological strain from the need to consume more oxygen under extreme temperatures while simultaneously being challenged by the growing extent of hypoxia that has been identified in the Gulf (de Verneil et al., 2021; Lachkar et al., 2022; Vaughan et al., 2021). The physiological costs of accommodating the naturally extreme temperatures and salinity in the Gulf have previously been implicated in reducing the size and productivity of Gulf fish and therefore fisheries yields (Ben-Hasan et al., 2024); further environmental pressure may exacerbate these effects and have direct negative impacts on fisheries - a resource sector second only to oil in this region (Van Lavieren et al., 2011).

The mechanisms identified in this research may also be at work in other semi-enclosed marginal seas that have experienced rapid warming in recent decades, such as the Mediterranean Sea and the Red Sea. For example, García-Lafuente et al. (2021) documented a weakening of the Mediterranean outflow at the Strait of Gibraltar between 2004 and 2020, attributed to a warming-induced loss of buoyancy, particularly strong in the western Mediterranean. In the rapidly warming Red Sea (e.g., Chaidez et al., 2017), evidence of recent outflow changes is less clear (Xie et al., 2019), although paleo data suggest a weakening of the overturning circulation during warm periods (e.g., Hubert-Huard et al., 2023). These changes are likely to have significant effects on the nutrient budgets and biogeochemistry of these seas (e.g., Richon et al., 2019).

5 Summary and Conclusions

As a shallow, semi-enclosed marginal sea, the AG is notably sensitive to atmospheric forcing, leading to heightened seasonal and interannual variability compared to the SOO. Our analysis confirms that the AG has experienced rapid warming, exceeding that of the neighboring SOO. The primary driver of warming in both seas is enhanced downward radiation. However, weakening winds over the AG and strengthened southeasterly winds over the SOO contribute to accelerating warming in the former and dampening it in the latter. Consequently, this process increases the temperature gradient and reduces the density gradient between the two seas, thereby slowing down the water exchange between the Gulf and the Arabian Sea. This reduction in Gulf water outflow has led to an increased accumulation of nutrients and biomass in the Gulf over the recent decades, intensifying respiration and causing depletion of O₂ and an increase in water acidity, particularly pronounced in the deeper parts of the Gulf. Our findings underscore the importance of local changes in atmospheric conditions, particularly surface winds, in modulating global anthropogenic perturbations at regional scales, particularly for marginal and semi-enclosed seas.

Acknowledgments

This research was supported by funding provided by Tamkeen through grants CG009 to the Mubadala ACCESS center and CG007 to the Water Research Center, as well as funding support from Mubadala Philanthropies under XR016; their support is greatly appreciated. Computations were performed at the High Performance Computing (HPC) cluster of NYUAD, Jubail. We extend thanks to the NYUAD HPC team for technical support.

Open Research

Data Availability Statement

The atmospheric forcing data is available in European Centre for Medium-Range Weather Forecasts (2012). The lateral boundary conditions is available in Copernicus Climate Change Service, Climate Data Store (2021). The ROMS model code is available in Auclair et al. (2018). The model outputs are described and available in Lachkar (2024).