Extreme precipitation events induce high fluxes of groundwater and associated nutrients to coastal ocean

. Current Submarine Groundwater Discharge (SGD) studies are commonly conducted under aquifer baseflow conditions, neglecting the influence of episodic events that can significantly increase the supply of nutrients and water. This limits our understanding of the social, biogeochemical, and ecological impacts of SGD. In this study, we evaluated the influence of an extreme precipitation event (EPE) on the magnitude of both the terrestrial and marine components of SGD. To 15 do so, three seawater sampling campaigns were performed at a Mediterranean ephemeral stream-dominated basin after an extreme precipitation event (~90 mm in few hours) and in baseflow conditions. Results indicate that the groundwater flows of terrestrial and marine SGD after the extreme precipitation event were 1 order of magnitude higher than those in baseflow conditions. SGD induced by extreme precipitation events, which only take place a few days per year, represented up to one third of the annual discharge of groundwater and associated nutrients at the study site. This work accentuates the need to 20 account for episodic increases in the supply of water and nutrients when aiming at providing reliable annual SGD estimates, particularly in the current context of climate change, since the occurrence of such events is expected to increase worldwide.


Introduction
Submarine groundwater discharge (SGD) -the flow of terrestrial and marine groundwater to the coastal ocean -is one of the primary processes regulating the transfer of solutes from land to ocean (Santos et al., 2021). The significance of this process 25 at local, regional and global scale stems mainly from its role in modulating the water and chemical budgets of oceans, controlling coastal ecosystems, and contributing to the well-being of coastal societies Lecher et al., 2015;Luijendijk et al., 2020). In the last three decades, there have been many studies focusing on quantifying SGD and associated solute fluxes in multiple sites across the globe, including coves, bays, estuaries, and entire basins (e.g., Beck et al., 2008;Kwon et al., 2014;Tamborski et al., 2020). However, most of the SGD investigations are conducted under baseflow 30 conditions, that is, in the absence of any meteorological, hydrological, or oceanographical event (e.g., storms, monsoons, sea-

Field methods
Three samplings were conducted in the southern section of Maresme County during 2019 and 2020. The two first samplings (hereinafter P1 and P2, chronologically) were performed shortly after an EPE with an accumulated precipitation rate of ~90 85 mm in one day. The EPE took place on October 22 nd 2019 and P1 and P2 were conducted on October 25 th and 29 th , 2019 (~4 and 8 days after the rainfall event, respectively). The third sampling (named BF, after "baseflow") was conducted on March 11 th 2020, was not affected by any rainfall event, and was therefore considered to have been conducted under baseflow conditions (accumulated rainfall of 18 mm in the prior 40 days). Seawater samples were collected at different stations from three perpendicular transects to the coastline corresponding to the ephemeral streams of Argentona, Cabrera de Mar, and 90 Vilassar de Mar (Transects T1, T2, and T3, respectively; Fig. 1). Each transect consisted of 7 offshore stations distributed along the first 1000 m. The central transect (T2) had three additional stations at 1500, 2000, and 4000 m from the coastline.
Coastal seawaters were collected directly from the shore by filling 25 L water containers. At each station, surface and deep (only selected stations) seawater samples were collected by placing a submersible pump at ~0.5 m depth and ~1 m above the seabed, respectively. 95 Samples were collected for Ra isotopes ( 223 Ra, 224 Ra, 226 Ra, and 228 Ra; 25 -120 L each sample), which are widely applied tracers of SGD (Garcia- Orellana et al., 2021) and nutrient analysis. Depth profiles of salinity and temperature were performed at each station by using a YSI 600XL probe. Groundwater samples for Ra isotopes (10 -25 L) and nutrients were collected periodically from 2015 to 2020 in several piezometers at the experimental site of the Medistraes project, located in the coastal alluvial aquifer of the Argentona ephemeral stream (corresponding to Transect 1). The experimental site consists of 16 100 piezometers located at 30 to 100 m from the coastline with screened depths of 15 to 25 m, with 2 m screened intervals for each (see Folch et al., 2020;Palacios et al., 2019, for more details about the experimental site; Fig. 1).
Each piezometer was purged with a submersible pump to remove at least three times the volume of stagnant water before sampling. Continuous in-situ groundwater level, conductivity, and temperature time-series were monitored at a shallow piezometric well (N3-15; 15 m depth, 2 m screened interval, from 11 to 13 m, 80 m from the coastline) using a CTD diver. 105 Salinity and groundwater temperature, as well as seawater samples, were measured in-situ with two handheld probes (HANNA HI98192 and WTW COND 330I). Rainfall data was obtained from a meteorological station from the Meteorological Catalan Service (Servei Meteorològic de Catalunya; SMC) at the municipality of Cabrils (see Fig. 1).

Field methods
Samples collected for Ra isotopes both in seawater and groundwater were weighted and filtered through MnO2-impregnated 110 acrylic fibers, at a controlled flow rate (< 1 L min -1 ) to ensure the quantitative adsorption of Ra onto the fibers (Moore and Reid, 1973). Fibers were then washed with Ra-free deionized water and partially dried to a fiber-water ratio of 1:1 (Sun and Torgersen, 1998). Each fiber was measured twice with the Radium Delayed Coincidence Counter (RaDeCC) (Moore and Arnold, 1996). Short-lived Ra isotopes ( 223 Ra, T1/2 = 11.4 d; 224 Ra, T1/2 = 3.66 d) were quantified using the first RaDeCC https://doi.org/10.5194/hess-2021-594 Preprint. quantification of 223 Ra with the RaDeCC system, due to cross-talk effect (Diego-Feliu et al., 2020). The second measurement, performed one month after sample collection, was used for quantifying the unsupported activity of 224 Ra (excess 224 Ra; 224 Raex), by accounting for the activity of its parent, 228 Th, in the fiber. The quantification of 224 Ra was made following the guidelines and limits proposed by Diego-Feliu et al. (2020) in order to avoid interferences inherent to the detection system, while 224 Ra uncertainties were estimated following Garcia-Solsona et al. (2008). The MnO2-fibers were subsequently incinerated, 120 grounded, and transferred to gamma counting vials. After radioactive equilibration (~21 d), the activities of long-lived Ra isotopes ( 226 Ra, T1/2 = 1,600 y; 228 Ra, T1/2 = 5.75 y) were measured using a HPGe gamma spectrometer. The photopeaks of 214 Pb (352 keV) and 228 Ac (911 keV) were used to quantify the activities of 226 Ra and 228 Ra, respectively.

Meteorological and hydrological context 130
The temporal evolution of groundwater level, conductivity, significant wave height, mean sea level (MSL), and accumulated precipitation from October 2019 to April 2020 are shown in Fig. 2. Three major precipitation events occurred in October 2019 (~90 mm), December 2019 (~100 mm), and January 2020 (~160 mm), which had a direct impact on groundwater level and conductivity. These events are considered extreme precipitation events following the threshold value derived from the 99 th wet-day percentile (~75 mm). After each EPE, the groundwater table from the shallow piezometer (N3-15) rose between 60 135 and 130 cm, gradually recovering the previous values 7 to 10 days after the event. The magnitude in the increase of the groundwater level (60, 70, and 130 cm, respectively) correlated with the amount of accumulated rainfall corresponding to each EPE. Precipitation events were followed by a drastic reduction of groundwater conductivity at the shallow piezometer, reaching minimum values of ~1 mS cm -1 , and a subsequent gradual increase before the next precipitation event (Fig. 2). As inferred by electric resistivity tomography profiles shown in Palacios et al. (2019), conductivity variations in the piezometric wells of the 140 study site were not derived from dilution with low-conductivity rainwater, but associated with the movement of the mixing zone due to EPE. Significant wave height fluctuations occurred, associated mainly with changes in wind and atmospheric pressure during the EPEs, increasing rapidly from 2 to 5 m above the baseline value of approximately 0.5 m. Similarly, the MSL presented oscillations linked to the EPE, which are usually associated with atmospheric fronts and strong winds, and also to seasonal meteorology, with higher MSL values from October to December than from January to April. 145 groundwater level relative to the values of March 2020. The data from the buoy and CTD-diver was smoothed by using a low-pass filter (12 h averaged). Red lines indicate the groundwater and seawater samplings performed at the study site (P1, P2, and BF) and grey bands indicate the EPEs that occurred during the monitoring period (10 days after the event are included in the band).

Radium and nutrient concentrations
The activities of Ra isotopes in groundwater samples measured during the 2015-2020 period in the Medistraes site ranged from 10 to 940, 10 to 550, and 1 to 50 Bq m -3 for 224 Ra, 228 Ra, and 226 Ra, respectively (see supplementary information (SI); Fig. S1).
The activities of Ra increased with groundwater salinity, presenting some variations that are mostly associated with differences in the geological matrix (Beck and Cochran, 2013;Webster et al., 1995). The seawater activities of 224 Ra and 228 Ra isotopes 150 generally decreased with increasing distance offshore for all transects and seawater samplings (

Pathways of submarine groundwater discharge
Submarine Groundwater Discharge incorporates a set of water flow processes involving the discharge of fresh groundwater and the circulation of seawater through permeable sediments (Garcia- Orellana et al., 2021;Michael et al., 2011;Santos et al., 2012). The driving forces and pathways of these processes likely determine the extent of the chemical reactions occurring in the subterranean estuary (Moore, 1999). Therefore, considering all the different SGD pathways concurrently occurring in a 170 specific study site is fundamental for deriving reliable estimates of SGD and associated nutrient fluxes (Garcia- Orellana et al., 2021). Here, we define terrestrial groundwater discharge as the combined discharge of meteoric groundwater and densitydriven circulated seawater, and marine groundwater discharge as those processes solely involving the circulation of seawater through permeable sediments (i.e., beach-face circulation, porewater exchange).
Radium isotopes represent one of the most common techniques for quantifying the fluxes of groundwater and associated 175 nutrients to the coastal ocean (Garcia- Orellana et al., 2021;Taniguchi et al., 2019). These isotopes can additionally be instrumental for differentiating SGD pathways, since their enrichment rates strongly depend on the transit time of groundwater through the coastal aquifer (e.g., Michael et al., 2011). Coastal seawater samples collected during the three samplings performed in Maresme County were enriched in both 224 Ra and 228 Ra relative to offshore waters (Fig. 3), suggesting the occurrence of a land-based Ra source. Whilst the enrichment in 224 Ra may result from any groundwater 180 discharge, regardless of spatiotemporal scale, due to its short half-life ( 224 Ra is enriched in all SGD pathways), coastal waters enriched in 228 Ra may be indicative of long-scale SGD pathways (e.g., terrestrial groundwater discharge; Rodellas et al., 2017;Tamborski et al., 2017a).
The activity ratio (AR) of 224 Ra/ 228 Ra can similarly be used to evaluate the temporal scale of SGD pathways . Thus, the 224 Ra/ 228 Ra AR found in coastal seawater samples after the EPE from October 2019 decreased from a 185 baseline value of 6 to approximately 4. This decrease is simultaneously followed by an increase in absolute 228 Ra activities, which are two-times higher than those in baseflow conditions (Fig. 3). Both trends may indicate that the relative contribution of the terrestrial component of SGD, which is characterized by 224 Ra/ 228 Ra ARs close to the equilibrium value (1.0 to 2.2; , increased during the occurrence of the EPE. Thus, based on the application of Ra isotopes and the estimates of the second sampling (P2; 8 days after the event) are comparable to those from baseflow conditions. Besides the similitudes in SGD estimates from these two samplings (P2 and BF), they also present similar groundwater levels, conductivities, and Darcy's flow estimates ( Fig. 2 and Fig. B1). Thus, the temporal extent of the EPE effects on SGD is 200 consistent with the recovery of groundwater level, which commonly occurs from 7 to 10 days after the rainfall ceases (Fig. 2).
In baseflow conditions, the terrestrial component of SGD (including fresh and brackish density-driven discharge) represented 60% of the total SGD (Fig. 4). The relative contribution of this SGD component increased after the rainfall event of October 2019 to up to 75% of the total SGD. This is consistent with the variation on the 224 Ra/ 228 Ra AR in coastal seawater after the EPE (see Section 4.1.1) and coherent with Darcy's flow calculations (Appendix B). These estimates of the relative contribution 205 of the terrestrial component are generally much larger than previous estimates for the Mediterranean Sea (1 -25%, Rodellas et al., 2015), global estimates (10%, Kwon et al., 2014;0.06%, Luijendijk et al., 2020), and local studies (5 -55%; Alorda- Kleinglass et al., 2019;Beck et al., 2008;Kiro et al., 2014;Knee et al., 2016;Rodellas et al., 2017;Tamborski et al., 2017a).
This difference most likely emphasizes the role of alluvial aquifers in ephemeral streams-dominated areas as a preferential pathway of terrestrial groundwater discharge (Jou-Claus et al., 2021). It should be noted however that any comparison between 210 studies should be considered as semi-quantitative, firstly because the diverse geological settings where SGD studies are conducted result in diverse SGD pathways and implications, and secondly because different studies use disparate methods that can capture different components of SGD depending on their spatial and temporal scales (Taniguchi et al., 2019).

SGD-driven nutrient fluxes
SGD-driven nutrient fluxes were estimated by considering the Ra-derived flows of terrestrial and marine SGD and the 215 respective nutrient concentration in groundwater from both fractions (see Appendix A). Total SGD-driven fluxes in baseflow conditions for dissolved inorganic nitrogen (DIN), dissolved inorganic phosphorus (DIP), and dissolved silicate (DSi) derived from median SGD estimates in the study site were 16.2, 0.06, and 5.4 ·10 3 mol km -1 d -1 , respectively (Fig. 5). The median fluxes, normalized by the study site area, were lower compared with median SGD-derived nutrient fluxes estimated worldwide for DIP and DSi, but significantly higher for DIN (2.7 times higher; Santos et al., 2021). The DIN:DIP ratio was 390:1, much 220 higher than the Redfield ratio of 16:1, but comparable with SGD-derived input in the Mediterranean Sea (80:1-430:1; Rodellas et al., 2015) and in studies worldwide (259±1090:1; Santos et al., 2021). The high loads of N and the disproportionate ratio DIN:DIP relative to the Readfield ratio in the study site may result from the lixiviation of nitrogen from agricultural activities (representing ~15% of the total land-use; Rufí-Salís et al., 2019), and the attenuation of P along groundwater flow paths due to adsorption onto Mn/Fe oxides present in the coastal aquifer (Robinson et al., 2018;Spiteri et al., 2007Spiteri et al., , 2008b. 225 After an EPE, the supply of all nutrients increased due to the higher terrestrial and marine SGD associated with these episodes (Fig. 5). Fluxes after the EPE (P1) were 9 times higher for DIN and 7 times higher for DIP and DSi than those in baseflow conditions. The predominant pathway for DIN, DIP, and DSi discharge to the coastal ocean was the terrestrial component of SGD. This pathway represented ~60% of the total inputs of DIP and DSi in baseflow conditions and up to ~75% after an EPE (Fig. 5). Nitrogen inputs during EPE and in baseflow conditions were chiefly governed by the discharge of terrestrial SGD 230 (~99% of the total DIN inputs; Fig. 5). The significant difference between the supply of nitrogen through terrestrial and marine SGD relies on the high concentrations of nitrate (~1,000 µM) in coastal aquifer freshwater (see SI; Fig. S2), which exceed the maximum groundwater concentration for drinking water set by the World Health Organization (WHO, 2011). Contrastingly, marine SGD is a relevant source of nitrite and ammonia, representing ~70% and ~40% of the total fluxes, respectively. It should be noticed that nutrient fluxes were estimated by multiplying the volumetric water flux of terrestrial and marine SGD 235 by the minimum nutrient concentration from a set of onshore samples, selected following the criteria used for the Ra endmembers, as explained in the appendices (see appendix A.2.4). Since it was not possible to directly collect the discharging groundwater, by using onshore samples we are implicitly assuming that no nutrient transformation occurred between the sampling point and the discharge point, within the subterranean estuary (Cook et al., 2018). It should also be noted that these SGD-derived nutrient estimates may be biased due to the groundwater endmember selection, since nutrient concentrations in 240 discharging groundwaters may vary during EPE due to dilution, increasing lixiviation of fertilizers, or enhancement of biogeochemical reactions in the mixing zone of coastal aquifers (Spiteri et al., 2008a).

Episodic events
Although several studies have focused on understanding the seasonal dynamics of SGD (Charette, 2007;Gwak et al., 2014;245 Michael et al., 2005;Rodellas et al., 2017), limited research has been done on SGD driven by episodic events (Adyasari et al., 2021;Wilson et al., 2011). This is mainly because of the inherent difficulties related to monitoring and sampling during and after these extreme events. Some studies have already shown that SGD may vary in direct or delayed response to meteorological and oceanic episodic events such as sea-level anomalies (Gonneea et al., 2013), waves (Bakhtyar et al., 2012;Rodellas et al., 2020;Sawyer et al., 2013), hurricanes (Hu et al., 2006), typhoons (Cho et al., 2021;Sugimoto et al., 2016), and 250 temperature inversion (Moore and Wilson, 2005). Regarding SGD induced by precipitation events: Sugimoto et al. (2016) reported high values of SGD eleven days after a precipitation event in Obama Bay (Japan); Gwak et al. (2014) suggested that SGD in the II-Gwang watershed (South Korea) was partially triggered by intensive precipitation events; and Uddameri et al.
(2014) indicated that precipitation events associated with the Emili hurricane contributed to the SGD in Baffin Bay (USA). At our study site, the significant increase (1 order of magnitude; Fig. 4) in both the terrestrial and marine SGD after EPE, may be 255 mediated by different processes: (1) increase of terrestrial groundwater discharge due to aquifer infiltration of rainfall, subsequent increase of hydraulic gradient, and displacement of stored water towards the sea (Anwar et al., 2014;Palacios et al., 2019;Santos et al., 2012;Yu et al., 2017), (2) increase in the exchange of seawater and density-driven discharge due to movements of the fresh-saltwater mixing zone, and (3) increase of shoreface circulation of seawater and porewater exchange due to the effect of sea level rise and waves associated with the EPE (Fig. 2). 260 The higher total SGD driven by EPE, especially the increase of terrestrial relative to marine SGD during these episodic events ( Fig. 4), induces the transport of large amounts of nutrients from the freshwater fraction of coastal aquifers into the coastal ocean. These fluxes, which are substantially higher than those in baseflow conditions (Fig. 5), may represent a significant periodic and episode-related nutrient input of particular relevance in sites where surface water renewal is limited (e.g., coastal lagoons and/or semi-enclosed bays), EPE occurs frequently (e.g., Mediterranean region), and the response to EPE is fast due 265 to the geological and geomorphological characteristics of the coastline (e.g., alluvial aquifers in ephemeral stream-dominated areas, such as Maresme County). However, in other areas the response to EPEs may be much slower. For instance; aquifers with a high thickness of vadose zones, confined aquifers with recharge areas far from the coastal zones, or systems with soils with water deficits (among other factors), may smooth or delay the effects of EPE. The biological and ecological implications of these events-which may include eutrophication, formation of red and green tides, and mass fish death (Hu et al., 2006;Lee 270 et al., 2010;Montiel et al., 2019;Valiela et al., 1990;Zhao et al., 2021)-are far from being understood, and require further attention.

Annual SGD estimates
A proper understanding of the temporal patterns of SGD in local to regional-scale studies is essential for deriving annual estimates for predicting reliable ocean budgets of nutrients and other dissolved compounds (Luijendijk et al., 2020;Santos et 275 al., 2021). However, most SGD studies are conducted in periods with stable meteorological conditions to evaluate baseflow SGD and associated nutrient, metal, or contaminant fluxes. Based on the results obtained from monitoring the EPE occurred at Maresme County in October 2019, the discharge of groundwater associated to this single event accounted for 13% (IQR: 5 -40%) and 8% (IQR: 5 -18%) of the annual terrestrial and marine fraction of SGD, respectively. Moreover, the nutrient inputs resulting from this event (lasting only 8 days; ~2% of the year) represented 13% (IQR: 5 -40%) for DIN, and 11% 280 (IQR: 5 -30) for DIP and DSi, of the yearly supply of nutrients at the study site. The increase in SGD-driven nutrient fluxes during these events may be mediated, on the one hand by the total SGD increase, but also because this increase is more significant for terrestrial SGD, which presents higher concentrations of nutrients (see supplementary information; Fig. S2), relative to marine SGD. These results suggest that annual estimates based on samplings conducted in baseflow conditions may systematically underestimate SGD and associated nutrients, particularly at study sites affected by EPEs or other episodic events 285 that can significantly impact SGD. Periodic and seasonal samplings may be taken as snapshots, and only representative of the time periods with similar environmental conditions. A better characterization of the hydrological and meteorological context https://doi.org/10.5194/hess-2021-594 Preprint. is necessary in pursuit of more reliable annual estimates, which may include seasonal and episodic-related variations. In this scenario, alternative methods such as groundwater level monitoring, Darcy's law calculations, electric resistivity tomography characterization, among other methods, may be instrumental in the design of proper sampling strategies and to capture seasonal 290 and episodic variations (e.g., Folch et al., 2020;Palacios et al., 2019).

Climate change
Climate change and its associated social and environmental impacts have become one of the most pressing scientific challenges for the 21 st century. This requires the acquisition of a holistic and integrative knowledge of systems and processes for modelling and predicting future scenarios. Fundamental research relating environmental key variables (e.g., temperature, sea-level rise, 295 precipitation) with social processes (e.g., land demand, coastal overpopulation, groundwater squeeze) becomes crucial to this aim. Research on SGD is no exception to this trend and, in recent decades, several studies have evaluated the key factors contributing to groundwater flows discharging into the ocean. Some examples of SGD research linked to climate change include; understanding sea-level and/or tidal controls on SGD (Gonneea et al., 2013;Wilson et al., 2015); the influence of land-use changes (Rufí-Salís et al., 2019); the relationship to increasing seawater intrusion (Werner et al., 2013); and the fate 300 and evolution of nutrients in groundwater (Beusen et al., 2013;Van Meter et al., 2018;Tait et al., 2014).
The precipitation-recharge relationship is one of the key parameters influencing the discharge of groundwater. Indeed, the amount of precipitation and the frequency and/or distribution of rainfall events, together with the hydrogeological characteristics of the receiving aquifers, strongly affects both the quantity and chemical quality of terrestrial groundwater discharge to the ocean (Kundzewicz and Döll, 2009;Stigter et al., 2014). Climate models indicate substantial spatial variation 305 in future changes to average precipitation. Whilst the increased specific humidity and transport of water vapor from tropic regions is likely to increase the amount of precipitation in high latitude land masses, precipitation in subtropical and semi-arid regions, like the Mediterranean Sea, is expected to decrease. Simultaneously, EPE in these regions are likely to increase in intensity and frequency (IPCC, 2021). Consequently, the yearly recharge of groundwater associated with precipitation in the Mediterranean region may diminish, also reducing the annual discharge of groundwater. In that scenario, EPEs may become 310 a major driving force of SGD, having a dominant significance in the annual fluxes of solutes to the coastal ocean.
The potential relevance of EPEs on SGD in future conditions can be qualitatively evaluated for Maresme County by considering the period from October 2019 to April 2020, when 3 precipitation events of >75 mm occurred (Fig. 2). Since the historical recurrence of EPE in the area is around 13 months (based on the meteorological data from 2015 to 2020), this 7month period can be considered as a future-like year with increased recurrence of EPE. Assuming that each one of the EPE 315 produces an increase in SGD comparable to the event monitored in this study, the relative contribution of SGD during EPEs would represent 30% (IQR: 15 -70%) and 22% (15 -40%) of the annual SGD issued by the terrestrial and marine fraction, respectively. Similarly, nutrient fluxes associated with EPE for this period would represent 34% (IQR: 15 -70%) for DIN and 30% (IQR: 15 -60%) for DIP and DSi of the yearly nutrient inputs supplied by SGD. Notice that due to the assumptions made in the determination of groundwater and nutrient fluxes (e.g., steady state, endmember selection; see appendix A), the estimates 320 are seemingly conservative, especially considering that the monitored EPE is minor relative to other EPEs occurring in 2019 and 2020 (Fig. 2). These estimates emphasize the need for integrating episodic events, such as EPE, in future climate change scenarios, in order to properly constrain the fluxes of groundwater and solutes to the coastal ocean driven by SGD.

Conclusions
Extreme precipitation events are potential drivers of submarine groundwater discharge and driven nutrients to the coastal 325 ocean. The lack of studies assessing the impact of these episodic events mask the implications that they may have for coastal geochemical cycles, coastal ecosystems, nutrient budgets, and hydrological cycle estimates. We have assessed the fluxes of SGD induced by an EPE in an ephemeral stream-dominated basin of the western Mediterranean Sea. SGD induced by the EPE increased by one order of magnitude and represented up to 13 and 8% of the total annual discharge of groundwater of terrestrial and marine SGD, respectively. Similarly, fluxes of nutrients driven by SGD during an EPE represented 11 -13% of the annual 330 total SGD, and up to ~30% during abnormally rainy seasons. This study highlights the relevance of these extreme events on the discharge of groundwater and solutes to the coastal ocean, noting their implications for annual SGD estimates and the possible consequences on coastal biogeochemistry cycles. The results of this study contribute to the understanding of the evolution of SGD with respect to future climate change scenarios, presenting an opportunity for streamlining future research in order to help managers and policy makers better estimate SGD and its related consequences. 335

A SGD and nutrient fluxes calculations
In this appendix, we develop the methodology used for determining the SGD and associated nutrient fluxes to the coastal ocean. This includes the definition of the conceptual model and the discussion of the model assumptions used in the calculations. 340

A.1 Radium mass balance
The magnitude of SGD and its associated nutrient fluxes are quantified in this study by using Ra isotopes, which is one of the most commonly applied techniques (Garcia- Orellana et al., 2021;Taniguchi et al., 2019). Whilst single isotopes can be used for quantifying SGD driven by single pathways, the combination of different isotopes is instrumental in discriminating SGD in sites with multiple pathways (Alorda-Kleinglass et al., 2019;Charette, 2007;Rodellas et al., 2017;Tamborski et al., 2017b). 345 In this study, Ra isotopes are used for discriminating and quantifying both the fluxes of terrestrial and marine SGD. Whilst terrestrial SGD represents a net input of water to the ocean, marine SGD comprises disparate discharge processes solely involving the circulation of seawater through permeable sediments or the coastal aquifer (i.e., porewater exchange, shoreface circulation of seawater, seasonal exchange of seawater; Garcia- Orellana et al., 2021;Michael et al., 2011). A steady-state mass balance of short-lived 224 Ra (T1/2 = 3.66 d) and long-lived 228 Ra (T1/2 = 5.75 y) isotopes was constructed as follows: 350 The two terms on the right-hand side of Eq. ( Table A1.

A.2.1 Sources and sinks of Ra
The proposed model for quantifying the Ra flux to the sea assumes that terrestrial and marine groundwater discharge are the only sources of Ra at the study site (Eq. A1). Diffusive fluxes of Ra from sediments were considered negligible, due to the presence of coarse-grained sands with low specific surface area (Luek and Beck, 2014) and are assumed to represent low (10%) inputs compared to the total Ra inputs (Beck et al., 2007;Garcia-Orellana et al., 2014, 2021. Ra inputs from surface 370 water were also discarded for the sampling conducted in March 2020 (BF) due to the total absence of runoff during the sampling period. In October 2019, 4 days before the first sampling conducted at the study site, runoff occurred in direct response to an EPE (~90 mm). However, considering the flushing time of Ra isotopes in the coastal system (see S1.2.2), the Ra delivered by this punctual runoff may have decreased by >90% for the first sampling (P1) and by >99% for the second sampling (P2), due to decay and mixing with offshore waters. It is thus assumed that runoff for these samplings represents a 375 minor source of Ra at the study site relative to that from SGD. Atmospheric deposition was discarded as a major source of Ra since its contribution in small-scale study sites is often <<1% (Garcia- Orellana et al., 2021). Production of Ra from dissolved Th was implicitly included by reporting the activities of Ra isotopes as 'excess' Ra activities (activities non-supported by their progenitors). The decay of Ra and the exchange with offshore waters were considered as the major sinks of Ra. The decay was assessed via the Ra inventories at in the study site and the offshore exchange, by evaluating the flushing time of Ra ( 2 ). 380

A.2.2 Radium flushing time
The flushing time of Ra ( 2 ) is a parameter that describes the transport of Ra in surface water bodies due to advection and dispersion processes (Monsen et al., 2002). In Ra mass balances, this parameter is fundamental for evaluating the exchange of Ra between coastal and offshore waters. In this work, rather than evaluating Ra flushing times, we used 224 Ra/ 228 Ra of coastal and offshore waters (1,000 m from coastline) to determine the water apparent age ( 3 ) (Moore, 2000). The water apparent age 385 is a good proxy for temporal scales of advective and mixing processes occurring at the study site. Coastal waters in Maresme County presented 224 Ra/ 228 Ra activity ratios ranging from 1.6 to 2.9 times higher than those of offshore waters, which led to seawater apparent ages of 2.4±0.9, 5.6±1.9, and 5.0±1.7 days for the first, second, and third sampling, respectively. The lower seawater residence time of the first sampling is coherent with the oceanographic conditions (e.g., higher winds, waves and currents that enhanced advection and exchange with offshore waters) linked to the extreme precipitation event occurring 4 390 days before (see Fig. 2).

A.2.3 Steady state conditions
Steady state conditions (i.e., tracer inventories do not vary with time; /( · ) = 0) are often assumed in Ra mass balances (e.g., Alorda-Kleinglass et al., 2019;Beck et al., 2008;Rodellas et al., 2017). This assumption implies that Ra inputs and outputs are balanced for a time period equivalent to the tracer residence time in the system . In Maresme 395 County, the tracer residence time ranged from 1.6 to 2.6 days for 224 Ra and from 2.4 to 5.6 days for 228 Ra. The tracer residence time can be estimated by dividing the radium inventory in surface waters by the sum of all losses (i.e., radioactive decay and exchange with offshore waters) . The assumption of steady state may therefore not be valid due to the significant difference between Ra activities from the first and second samplings (P1 and P2; Fig. 3), which were carried out only 4 days apart. However, assuming steady state conditions instead of a continuous decrease of activities in coastal waters 400 (− / ) (the pattern that was observed in the EPE from 2019; Fig. 3), results in conservative estimates of SGD induced by EPE.

A.2.4 Endmember selection
Due to the large spatial variability of Ra isotopes in the groundwater activity at the experimental site of Argentona, constraining the Ra activity of the SGD endmember for both the terrestrial and marine components is particularly difficult. To overcome 405 this limitation, we used the activity ratio of 224 Ra/ 228 Ra measured in the potential groundwater endmembers and the coastal ocean, which can help in identifying the most likely endmembers . Endmembers were selected according to the following conditions: (1) the selection of the terrestrial Ra endmembers ( '()* ) was constrained to groundwater samples with low salinities (Sal < 5); and (2) the 224 Ra/ 228 Ra activity ratios of both endmembers must satisfy the where ./%!!# / ./%!!+ is the ratio between the total fluxes of 224 Ra and 228 Ra to the coastal ocean for each of the three samplings. A terrestrial and marine SGD was determined for each of any possible combinations between terrestrial and marine Ra endmembers that satisfied the above-mentioned conditions, and we reported the final SGD fluxes as the median value and the interquartile range. Conservative fluxes of nutrients were computed by multiplying the minimum concentrations of 415 nutrients within the terrestrial and marine endmembers (discriminated by conditions 1 and 2), and the median (± interquartile range) of each SGD component (terrestrial and marine).

B Darcy's law calculations
The relative significance of EPE in annual SGD estimates derived from the Ra mass balance was compared with Darcy flux estimates ( = − 4 · ). For these calculations, hydraulic conductivity ( 4 [m s -1 ]) was assumed to be in the order of 10 -3 m s -420 1 (characteristic of clean sands) in the range of local studies (personal communication T. Goyetche and L. Del Val). The hydraulic gradient was determined as the difference between mean sea level (MSL) data and groundwater level data acquired from a CTD diver deployed in a piezometric well at the experimental site of the Argentona ephemeral stream (N3-15, 80 m from the shoreline). Absolute Darcy flux results (Fig. B1) should only be taken as indicative. In fact, the relative variation of Darcy flux during EPE can be used as a proxy for groundwater discharge. Results indicate that the EPE from October 2019, 425 December 2019, and January 2020 represented 2, 4, and 6%, respectively, of the annual groundwater discharge. The relative significance of EPE derived from these calculations is slightly lower than those obtained from the Ra mass balance. This discrepancy can be associated with the different discharge processes that each method captures. Whilst Ra mass balance enables the quantification of processes with different spatiotemporal scales and different compositions of groundwater (e.g., terrestrial groundwater discharge, porewater exchange), Darcy's law only captures the discharge of meteoric or brackish 430 groundwater due to the hydraulic gradient at the shallowest aquifer.