When groundwater–surface water (gw–sw) interactions are simulated, two architectural families compete: sequential (loosely coupled) workflows that first compute recharge with a soil-zone model and then impose it on a separate groundwater model, and fully coupled integrated surface–subsurface models that solve land-surface, variably saturated subsurface, and channel flow as one continuum. A 2013 comparison of both families at Havelock, Quebec (Guay et al., Hydrological Processes 27(16):2258–2270) found materially different water budgets — 8.9% higher recharge, opposite sign of net aquifer–rock flux under full coupling — despite similar total discharge, and posed unresolved issues that structure this problem: determine the regimes and scales (hydrogeologic setting, feedback intensity, discretization, boundary-condition treatment, heterogeneity) where sequential workflows are adequate versus where full coupling is required; and establish how classical diagnostic partitions of streamflow into baseflow and event contributions (and the companion quantity, recharge) should be defined and extracted from integrated models, in which surface and subsurface waters exchange continuously and no internal separation exists. Later literature advanced both components — particle-tracking-based evaluation of classical baseflow filters (Partington et al. 2012), a multi-model intercomparison benchmark (Maxwell et al. 2014), coupling-strategy reviews (Haque et al. 2021), and 2023 demonstrations that classical recharge definitions fail in variably saturated models (Gong et al.; Waldowski et al.) — but resolves neither: no quantitative adequacy criteria for coupling-strategy choice exist, and the definitional problem for baseflow/event/recharge remains actively contested, with reviews through 2026 still listing both among open challenges.

Background

Groundwater–surface water (gw–sw) interaction modeling supports water-resources assessment, stream ecology, and contamination studies, and two architectural families are used. The first is sequential (also called loosely coupled or one-way): a soil-zone model computes recharge — the flux of water reaching the water table — which is then imposed as a boundary input on a separate three-dimensional groundwater flow model. A representative workflow, used at a small study catchment in Havelock, Quebec by Guay et al. (2013), combined HELP (Hydrologic Evaluation of Landfill Performance, a quasi-two-dimensional infiltration and water-balance model) with FEFLOW (a finite-element groundwater flow code), calibrated against measured water levels. The second family consists of integrated surface–subsurface hydrological models (ISSHMs), which solve land-surface runoff, channel flow, and variably saturated subsurface flow — governed by Richards' equation, the partial differential equation for water movement in unsaturated porous media — as a single physically coupled continuum; CATHY (CATchment HYdrology), applied to the entire Havelock flow domain by Guay et al., is one example, alongside codes such as HydroGeoSphere, ParFlow, and MIKE SHE. The comparison produced two headline findings: integrated and sequential water budgets differ in physically meaningful ways (recharge 233 mm/y under full coupling, 8.9% higher than sequential; net upward flow from the fractured aquifer versus a predicted net downward flow; annual discharge differing by 9.6%), and the sequential approach cannot represent surface–subsurface feedbacks, for example exfiltration of groundwater near gaining stream reaches (reaches that receive groundwater inflow and thus lose water from the aquifer to the stream).

The comparison also exposed a diagnostic tension that motivates part of this problem. Hydrograph separation — decomposing a streamflow time series into baseflow, the slowly varying groundwater-fed component, and quick event contributions from storms — is a standard tool of classical hydrology, usually implemented with empirical automated filters. The sequential workflow delivers such a decomposition as a by-product of its architecture (at Havelock: surface runoff 231 mm/y, subsurface runoff 36 mm/y, baseflow 189 mm/y), but the partition reflects model architecture rather than a physical accounting of water. An integrated model, in contrast, tracks water continuously across compartments and produces a single outlet discharge with no internal separation; recovering component contributions requires post-processing, for example particle tracking (numerically following water parcels through the simulated flow field to label their origin and pathway). Related definitional questions recur for recharge itself: in an integrated model the water table is internal to the domain, so 'recharge' is not a model input or term and must be reconstructed by post-processing, with subtleties arising from storage dynamics in the capillary fringe, the zone above the water table that remains nearly saturated.

Subsequent work sharpened both issues but did not close them. Partington et al. (2012) used an integrated model with particle tracking to generate a physically based baseflow time series and showed that classical automated baseflow separation methods disagree with it — the first systematic confrontation of filter-based separation with integrated-model ground truth. Liggett et al. (2014) showed that solute dispersion further complicates tracer-based separations in integrated models. Maxwell et al. (2014) established a multi-model benchmark intercomparison for integrated models, but compared coupled codes against each other rather than mapping where sequential workflows suffice. Reviews of coupling strategies (Haque et al. 2021; Ntona et al. 2022) offer only qualitative, case-specific guidance, and 2023 work demonstrated concretely that classical recharge definitions 'do not yield meaningful results' when implemented in variably saturated models (Gong et al.) and must be estimated by ambiguous post-processing (Waldowski et al.). Reviews through 2025–2026 continue to list coupling-strategy selection and interface validation among open challenges, now often proposing machine-learning hybrids rather than claiming the regime question is settled.

Problem Statement

Determine the regimes and scales under which sequential (loosely coupled) groundwater–surface water modeling workflows are adequate and those under which fully coupled integrated surface–subsurface formulations are required — where the axes of the regime determination include catchment scale, hydrogeologic setting, intensity and dynamics of surface–subsurface exchange (such as groundwater discharge to gaining reaches and water-table fluctuation), spatial and vertical discretization, boundary-condition treatment (including lower-boundary choices such as an imposed constant head at the model base), and parameter heterogeneity. As an integral part of the same objective, establish how the classical diagnostic partition of streamflow into baseflow and event contributions — and the companion quantity, groundwater recharge — should be defined, quantified, and extracted in fully integrated models, in which surface and subsurface waters exchange continuously and no explicit component separation exists, validating any proposed definitions against physics-based ground truth.

References

  1. Guay, C., Nastev, M., Paniconi, C., Sulis, M. (2013). Comparison of two modeling approaches for groundwater–surface water interactions. Hydrological Processes 27(16):2258–2270 (online 2012). DOI: 10.1002/hyp.9323. https://doi.org/10.1002/hyp.9323
  2. Partington, D., Brunner, P., Simmons, C.T., Werner, A.D., Therrien, R., Maier, H.R., Dandy, G.C. (2012). Evaluation of outputs from automated baseflow separation methods against simulated baseflow from a physically based, surface water–groundwater flow model. Journal of Hydrology 458–459:28–39. DOI: 10.1016/j.jhydrol.2012.06.029. https://doi.org/10.1016/j.jhydrol.2012.06.029
  3. Liggett, J.E., Werner, A.D., Smerdon, B.D., Partington, D., Simmons, C.T. (2014). Fully integrated modeling of surface–subsurface solute transport and the effect of dispersion in tracer hydrograph separation. Water Resources Research 50(10):7750–7765. DOI: 10.1002/2013WR015040. https://doi.org/10.1002/2013WR015040
  4. Maxwell, R.M., Putti, M., Meyerhoff, S., Delfs, J.-O., Ferguson, I.M., Ivanov, V., Kim, J., Kolditz, O., Kollet, S.J., Kumar, M., Lopez, S., Niu, J., Paniconi, C., Park, Y.-J., Phanikumar, M.S., Shen, C., Sudicky, E.A., Sulis, M. (2014). Surface–subsurface model intercomparison: A first set of benchmark results to diagnose integrated hydrology and feedbacks. Water Resources Research 50(2):1531–1549. DOI: 10.1002/2013WR013725. https://doi.org/10.1002/2013WR013725
  5. Haque, A., Salama, A., Lo, K., Wu, P. (2021). Surface and Groundwater Interactions: A Review of Coupling Strategies in Detailed Domain Models. Hydrology 8(1):35. DOI: 10.3390/hydrology8010035. https://doi.org/10.3390/hydrology8010035
  6. Ntona, M.M., Busico, G., Mastrocicco, M., Kazakis, N. (2022). Modeling groundwater and surface water interaction: An overview of current status and future challenges. Science of the Total Environment 846:157355. DOI: 10.1016/j.scitotenv.2022.157355. https://doi.org/10.1016/j.scitotenv.2022.157355
  7. Gong, C., Cook, P.G., Therrien, R., Wang, W., Brunner, P. (2023). On Groundwater Recharge in Variably Saturated Subsurface Flow Models. Water Resources Research 59(9):e2023WR034920. DOI: 10.1029/2023WR034920. https://doi.org/10.1029/2023WR034920
  8. Waldowski, B., Sánchez-León, E., Cirpka, O.A., Brandhorst, N., Hendricks Franssen, H.-J., Neuweiler, I. (2023). Estimating Groundwater Recharge in Fully Integrated pde-Based Hydrological Models. Water Resources Research 59(3):e2022WR032430. DOI: 10.1029/2022WR032430. https://doi.org/10.1029/2022WR032430
  9. Khurshid, N., Kumar, R., Vishwakarma, D.K., Kumar, S., Pandit, B.A., Khan, I., Pukhta, M.S., Yadav, K.K. (2025). Interaction Modeling of Surface Water and Groundwater: An Evaluation of Current and Future Issues. Water Conservation Science and Engineering 10(1):37. DOI: 10.1007/s41101-025-00360-5. https://doi.org/10.1007/s41101-025-00360-5
  10. Nanda, L., Panday, D.P., Srivastava, P.K., Mahapatra, D.M., Bhan, U., Kumar, M. (2026). Evolving Approaches in Groundwater-Surface Water Interaction Studies: From Numerical Models to Machine Learning. Current Pollution Reports 12(1):4. DOI: 10.1007/s40726-026-00395-9. https://doi.org/10.1007/s40726-026-00395-9