Submarine groundwater discharge (SGD) is widely recognized as a key pathway of groundwater-seawater exchange, linking terrestrial aquifers to coastal ecosystems and influencing nutrient dynamics, water quality, and biogeochemistry cycling (Mee, 2012; Amato et al., 2016; Bone et al., 2006; Geng and Michael, 2021; Moore et al., 2002; Guild et al., 2025) Despite major advances over the past decades, a fundamental challenge remains insufficiently addressed: how processes operating across multiple spatial and temporal scales collectively control the magnitude, timing, and ecological impacts of SGD (Figure 1). Although individual drivers, such as tides, waves, sea-level anomalies, density gradients, and aquifer geologic heterogeneity, have been studied extensively, most investigations isolate a subset of processes due to methodological or computational constraints. These reductionist approaches leave important gaps limit the applicability of current SGD models and conceptual frameworks to real coastlines, where multiple drivers co-occur and interact nonlinearly. For example, large-scale density-driven flow models often neglect wave- and tide-driven circulation, even though wave setup and tidal oscillations can generate dynamic pressure gradients that restructure the freshwater-saltwater interface across event to seasonal timescales and substantially enhance SGD (De Sieyes et al., 2011; Michael et al., 2016; Yu et al., 2017; Taniguchi et al., 2019; Jiang et al., 2021; Beebe et al., 2022). Likewise, studies that focus solely on wave pumping or tidal oscillations may overlook longer-term sea-level anomalies and seasonal recharge variability because of limited spatial and temporal domains, even though these processes modulate the background hydraulic gradients governing SGD (Tur-Piedra et al., 2024; Keshariya et al., 2025). Growing evidence indicates that omitting key multiscale drivers can lead to substantial predictive errors (Xin et al., 2010; Kretschmer et al., 2023). Models that neglect wave-pumping- or tide-driven flows have been shown to underestimate nearshore circulation cells and SGD fluxes by an order of magnitude, while studies that ignore seasonal variability often misattribute observed water-quality trends because the timing of SGD pulses is mismatched with ecological responses (Sawyer et al., 2013; Jiang et al., 2021; Lin et al., 2024). Similarly, field observations from heterogenous coastlines have revealed that small-scale sediment variability can generate highly localized discharge hotspots that are not captured by coarse-resolution regional models (e.g., Stieglitz et al., 2008). These examples illustrate a broader issue: current understanding of SGD is limited not by a lack of knowledge of individual processes, but by a lack of integration across scales.

This perspective argues that resolving this multiscale disconnect is essential for improving predictive understanding of SGD, especially as climate and anthropogenic pressures reshape coastal zones. By conceptualizing SGD as a nested system governed by processes spanning spatial scales, from pore-scale flow pathways and sediment heterogeneity to coastline-scale hydrodynamics, and temporal scales, from wave and tidal fluctuations to seasonal and long-term climatic drivers, we can more accurately determine how SGD shapes coastal ecosystem structure, function, and resilience. This Perspective does not aim to provide a comprehensive solution to the long-standing challenges of SGD characterization, nor to propose a single unifying methodology. Instead, it seeks to synthesize existing knowledge across spatial and temporal scales, identify structural disconnects in current approaches, and articulate realistic pathways for incremental progress under persistent data limitations and uncertainty.

Submarine groundwater discharge (SGD) originates at the smallest spatial scales, where the sediment and rock at microscales govern how water, solutes, and microorganisms migrate through the subsurface (Evans and Lizarralde, 2003; Viso et al., 2010). At this fundamental scale, pore geometry, characterized by grain size distribution, porosity, and pore connectivity and tortuosity, governs the hydraulic conductivity and capillarity that determine the medium’s effective transport properties (Berg, 2014; Ghanbarian et al., 2013). Subtle variations in grain packing and the presence of micro-fractures or micropores can significantly influence how density-driven flow evolves, particularly when freshwater and seawater interact. Density contrasts generate gravitational instabilities that produce fingering patterns, localized convective cells, and sharp mixing interfaces where biogeochemical reactions are amplified (Li et al., 2019; Meng et al., 2020; Pu et al., 2020). Microbial communities also exert a strong influence on the effective transport properties of porous media. Through biofilm growth, extracellular polymeric substance (EPS) production, and microbially mediated mineral precipitation and dissolution, microorganisms actively modify pore geometry and connectivity. These alterations can reduce or enhance permeability, redistribute porosity, and change the reactive surface area available for sorption and redox reactions, thereby reshaping flow and solute transport pathways over time (Taylor and Jaffé, 1990; Thullner et al., 2002). Grain-surface reactions, including sorption-desorption, carbonate dissolution, metal redox cycling, and organic matter degradation, further regulate the mobility and transformation of nutrients, carbon species, and contaminants within sediment pores (Goldberg et al., 2007; Huang and Weber, 1997; McMahon et al., 2011). Although the importance of these microscale processes has long been acknowledged, only recently have advanced techniques such as microfluidics and X-ray imaging, provided direct, quantitative observations of pore-scale flow and reaction dynamics under controlled conditions. Yet a persistent challenge remains: translating these fine-resolution observations into scalable parameters that can reliably inform mesoscale and field-scale modeling.

As pore-scale processes integrate over larger spatial domains, sediment heterogeneity and subsurface architecture exert dominant control over the magnitude, pathways, and spatial distribution of SGD (Michael and Voss, 2008; Michael et al., 2016; Geng et al., 2020; Kreyns et al., 2020; Heiss et al., 2020; Yu and Michael, 2022). Coastal aquifers are rarely homogeneous; instead, they comprise stratified, discontinuous, and anisotropic sedimentary and volcanic units shaped by complex depositional histories and subsequent diagenetic, weathering, and structural modifications (Todd and Mays, 2005; Cantelon et al., 2022; Semeniuk, 1981). Variations in grain size, facies transitions, and abrupt permeability contrasts create preferential flow paths as well as diffusion-limited zones where solutes may experience prolonged residence and enhanced reaction (Long et al., 1982). Localized lenses of high- or low-porosity material interbedded fine layers, and permeability anisotropy further modulate hydraulic gradients and determine where exchange fluxes become enhanced and diminished (Gelhar et al., 1992; Sudicky, 1986; Zhang et al., 2024). In volcanic island settings, heterogeneity is amplified by the inherent complexity of basaltic architectures. Highly permeable lava flows, clinker zones, and interflow rubble layers are often interlayered with low-permeability ash beds, weathered saprolite, and paleosols (Kiernan et al., 2003; Kreyns et al., 2020). Secondary porosity formed by fractures, cooling joints, voids, and lava tubes creates a network of conduits and barriers that can either move groundwater quickly over long distances or isolate it within separate compartments (Berkowitz, 2002). These structure elements strongly influence whether SGD emerges as focused point-source discharge, manifested as submarine springs, seepage faces, or as diffusive discharge distributed across broad shelf areas with long and variable residence times (Michael and Voss, 2008). In sedimentary coastal systems, additional heterogeneity arises from buried estuarine deposits, deltaic deposits, tidal flat sequences, and cross-bedded units that impart directional anisotropy to groundwater flow (Russoniello et al., 2013; Semeniuk, 1981). Biogenic modifications, including burrow networks, root channels, and bioturbated layers, further generate fine-scale preferential pathways that enhance mixing while also creating micro-zones of reduced permeability (Hose and Stumpp, 2019). Over geologic timescales, aquifer architecture is further reshaped by shoreline migration, sediment compaction, sea-level change, and tectonic deformation, all of which modify hydraulic connectivity and reorganize the spatial configuration of SGD pathways (Minderhoud et al., 2025; Church and Slaymaker, 1989).

Advances in characterization techniques are greatly enhancing our capacity to resolve and interpret this structural complexity. Marine geophysics including controlled source electromagnetics, multi-channel seismic reflection and acoustic imaging, electrical resistivity imaging, distributed temperature sensing, fiber-optic techniques, tracer tests, and Bayesian or machine-learning-enhanced inverse modeling are allowing researchers to identify SGD across larger spatial scales and map seafloor heterogeneity and pore fluid salinity with greater accuracy and depth (Andréis and MacGregor, 2008; Tur-Piedra et al., 2024; Taniguchi et al., 2019). Yet despite these advances, the functional implications of structural and offshore aquifer heterogeneity remain underexplored, largely because structural geophysical data lack resolution to allow integration with process-based measurements, and because small-scale heterogeneity often produces hydrologic and biogeochemical effects that only become apparent at much larger scales. This disconnect limits our ability to understand how heterogeneity reshapes groundwater-seawater exchange. Heterogeneity not only redistributes groundwater flow but also modulates biogeochemical transformations by influencing residence time distributions, redox zonation, and mixing intensity (Santos et al., 2021). These changes regulate nutrient retention, contaminant attenuation, and ecological connectivity between terrestrial and marine environments (Cai et al., 2011). Recognizing SGD as a multiscale, evolving subsurface system, linking pore-scale processes with the stratigraphic, geomorphological, and structural frameworks of coastal aquifers, is therefore essential for improving predictive capabilities. Only by explicitly embracing this spatial multiscaling can we capture the emergent patterns of SGD and their ecological and geochemical consequences.

SGD is continuously reshaped by hydrodynamic forcing that operates across a wide spectrum of temporal and spatial scales, producing exchange patterns that cannot be understood using a single-scale perspective. At the shortest timescales, seconds to minutes, wave swash motions and wave setup impose rapidly fluctuating cross-shore hydraulic gradients that force high-frequency seawater-groundwater exchange across the beach face, reorganizing pore-scale flow paths and creating alternating phases of infiltration and exfiltration (Geng et al., 2017; Geng et al., 2020a; Delisle et al., 2025; Geng et al., 2025); these high-frequency oscillations modulate mixing within the shallow aquifers, enhance dispersive exchange across the sediment-water interface (Olorunsaye and Heiss, 2024; Zheng et al., 2024), and generate transient SGD pulses that respond directly to wave energy (Yu et al., 2022). At hourly timescales, tidal oscillations impose longer-period pressure gradients that propagate landward and seaward (Geng and Boufadel, 2017; Huang et al., 2025), driving oscillatory flow, establishing tidally driven circulation cells within the subterranean estuary (Wang et al., 2025), and pumping solutes through beach aquifers (Geng et al., 2020b). These tidal pressure variations generate flushing cycles that extend beyond the wave-affected zone and modulate both the magnitude and timing of SGD (McKenzie et al., 2021; Hingst et al., 2024). At seasonal to interannual scales, variability in precipitation, aquifer recharge, regional groundwater levels, and coastal water levels modifies the land-sea hydraulic gradients that control the magnitude and alongshore distribution of SGD (Tur-Piedra et al., 2024; Dang et al., 2025; Lopez et al., 2025). These intermediate-scale processes can amplify or dampen the effects of shorter-term tidal and wave-driven forcing, generating multifrequency interactions and nonlinear responses that challenge steady-state or single-scale conceptualizations of coastal groundwater flow. Over decadal to centennial timescales, sea-level rise, land subsidence, and other geomorphic adjustments shift freshwater-saltwater interfaces, alter aquifer geometry and storage, and reorganize the spatial configuration of SGD pathways (Geng and Michael, 2021; Nicholls et al., 2021; Minderhoud et al., 2025). These long-term boundary changes establish the broader hydrogeologic framework within which higher-frequency processes operate, generating hierarchical cross-scale interactions that propagate from pore-scale mixing zones to kilometer-scale coastal discharge patterns.

Progress in resolving these multiscale hydrodynamic controls increasingly depends on integrated observational and modeling methodologies (Wang et al., 2022; Jin et al., 2024). Numerical approaches now span from pore-scale reactive transport simulations to coastal-scale density-dependent flow models, each capturing different dimensions of the hydrodynamic spectrum (Geng, et al., 2020c; Geng et al., 2021; Wang et al., 2024). Complementary geophysical, geochemical, tracer-based, and fiber-optic methods provide spatial and temporal constraints needed to validate, refine, and upscale these models (Folch et al., 2020; Furlanetto et al., 2024; Zhao et al., 2022). Machine-learning and data-assimilation frameworks offer new opportunities to extract hydrodynamic patterns from large datasets and to translate small-scale process understanding into actionable parameterizations for larger-scale modeling (McKenzie et al., 2023; Cao et al., 2024; Dai et al., 2025). Long-term monitoring networks equipped with autonomous sensors, coupled with time-series and spectral analyses in the frequency domain, have been used to reveal temporal dynamics, including rapid responses to storms, spring-neap tidal cycles, and climate extremes (Geng and Boufadel, 2017; Nordio et al., 2023; Williams et al., 2024). Such multiscale methodological integration is critical for resolving how interactions across different timescales give rise to emergent SGD behavior.

The multiscale controls described in Sections 2 and 3 reveal a central challenge for SGD research: integrating fine-scale subsurface architecture with hydrodynamic drivers acting across broad temporal and spatial scales. Future progress will require theoretical and computational frameworks that explicitly couple pore-scale processes with field-scale transport while acknowledging persistent data limitations. Although pore-scale imaging and characterization have advanced rapidly, a key unresolved challenge is identifying which fine-scale structural and biogeochemical features exert first-order control on effective field-scale transport parameters, and how these controls depend on hydrodynamic conditions. Addressing this challenge requires distinguishing between measurable fine-scale attributes and those that are transferable and predictive under realistic wave-, tide-, and recharge-driven forcing, rather than assuming that increased resolution alone will yield improved field-scale understanding. Data-driven approaches are most effective when applied to specific tasks, such as inferring effective transport parameters or quantifying uncertainty, rather than attempting full predictive replacement of physics-based models. Operative strategies include machine-learning-assisted upscaling that uses a limited set of pore-scale structural descriptors (e.g., connectivity metrics or characteristic length scales) in combination with high-resolution process-based simulations to estimate effective transport parameters; physics-informed data-driven models that constrain learning using mass and momentum conservation while allowing parameters to vary with subsurface heterogeneity and dynamic hydrodynamic forcing; and Bayesian data-assimilation frameworks that integrate geophysical imaging and hydrogeologic observations to iteratively update spatially distributed parameters, explicitly quantify uncertainty, and resolve scale-dependent spatial variability that cannot be captured by deterministic upscaling alone. A second priority is improving representation of multiscale temporal forcing in SGD studies. Future work should focus on identifying dominant modes of interaction among wave-driven swash, tidal oscillations, seasonal recharge, and long-term sea-level trends. Addressing these challenges will require robust upscaling approaches that connect high-frequency coastal forcing with intermediate-scale hydroclimate variability and long-term evolution of coastal boundary conditions. Example studies include phase-averaged and net-inflow approaches, which respectively upscale wave effects into an effective mean head gradient (i.e., wave setup) or represent wave-induced seawater infiltration as a net flux boundary condition (Robinson et al., 2014; Geng and Boufadel, 2015). Finally, deep and confined aquifers represent a critical but underexplored component of the SGD spectrum. While deep SGD may contribute substantial nutrient and contaminant fluxes, its quantification is limited by sparse data and detection challenges. Progress in this area will likely depend on targeted integration of geophysical imaging, geochemical tracers, and autonomous offshore observations, rather than comprehensive mapping, to identify preferential pathways and bound their potential contribution.

Advances will also depend on improved mapping and interpretation of terrestrial aquifer structure, as subsurface architecture fundamentally constrains where and how SGD emerges. While inadequate subsurface information remains a pervasive limitation in most coastal settings, this constraint underscores, rather than diminishes, the need for integrated, multiscale approaches. Because no single observational technique can resolve the full complexity of subsurface pathways, progress will depend on selectively combining complementary methods, including hydrogeologic characterization, thermal and acoustic imaging, tracer-based surveys, offshore geophysics, and emerging fiber-optic approaches, to delineate dominant discharge zones and bound flux estimates under uncertainty. Long-term, high-resolution monitoring at a limited number of strategically chosen sites, combined with targeted tracer injections and coordinated field experiments, provides a realistic pathway for resolving temporal variability and system responses that episodic sampling cannot capture. Analytical tools, including spectral, statistical, and machine-learning approaches, should be viewed not as substitute for data scarcity, but as means to extract process-relevant information from sparse, noisy, and multiscale observations. Predicting how SGD will respond to climate change further amplifies these challenges. Sea-level rise, altered precipitation patterns, intensified storms, and changes in groundwater withdrawal will modify aquifer structure and hydrodynamic gradients. Progress will therefore require downscaling large-scale climatic and oceanographic drivers to the spatial and temporal scales at which SGD operates, supported by coupled, density-dependent groundwater-nearshore hydrodynamic models capable of representing for anticipating changes in nutrient and contaminant delivery, carbon cycling, ecosystem thresholds, and coastal resilience. Together, these directions emphasize that predictive capability in SGD will emerge not from data volume, but from integrative frameworks designed to operate under persistent data limitations.