Measuring Groundwater–Surface Water Exchange

① Stream gaging & differential gaging

Continuous stage and discharge records at one or more points. Differential gaging — measuring Q at the top and bottom of a reach and computing the difference — gives the net gain or loss to/from the aquifer.

• Provides: stream stage (m) at a station, converted via a site-specific rating curve to discharge Q (cfs or m³/s). Differencing two stations gives the net reach-scale gain or loss (cfs gained or lost over the reach length); not the location of exchange within the reach.
• Scale: reach (10s of m to 10s of km).
• Strengths: direct, integrative, well-established (USGS NWIS); good for identifying reach-scale "hot spots" of gain or loss.
• Limitations: resolution is set by the propagated discharge-measurement uncertainty — differencing two USGS "good" ratings (each ±5% of Q per Sauer & Meyer 1992) leaves a ~±7% band, so a reach gain/loss smaller than that band is statistically indistinguishable from zero (not wildly uncertain — just unresolvable); also biased by ungaged tributaries and diversions, macrophytes, shifting margins, high sediment loads, and unstable streambeds (Ma et al. 2024).
• Best in: perennial and intermittent streams during stable low-flow periods; relatively homogeneous reaches.
• Poor for: tidally-influenced reaches (tides dominate the signal); high or flood flows — the absolute resolution threshold scales with Q while exchange typically does not, ratings are often extrapolated above the highest direct measurement, and scour/morphology changes invalidate the rating mid-event; braided / scour-dominated channels.
• Best practice: dry-season synoptic; multiple stations; combine with seepage runs.

② Seepage meter

A bottomless drum pressed into the streambed, connected to a collapsible bag. Captures all upward (gaining) or downward (losing) flow through the streambed area beneath the drum over a measurement interval. The only tool that directly measures GW exchange across a submerged sediment boundary (Ma et al. 2024).

• Provides: the vertical water flux through the streambed beneath the drum footprint (m/d or cm/d), with sign indicating direction (gaining + ; losing −), integrated over the drum area (~0.1 m²) and the deployment interval (minutes to days).
• Scale: point (≈ 0.1 m²).
• Strengths: direct flux measurement; inexpensive; resolves fine spatial and temporal variability.
• Limitations: bag inefficiency and friction biases (must be calibrated — Rosenberry & Menheer 2006); meter housing can induce interstitial flow into the pan; hyporheic exchange can be hard to separate from net GW discharge.
• Best in: still or low-velocity water — lakes, wetlands, slow streams; perennial and intermittent SW.
• Poor for: high-current or wave environments (currents distort or fold the bag and shift hydraulic head); coarse cobble or armored streambeds; frozen ground. Tidal/marine: only with rigid-housing or continuous-recording variants that isolate the bag from pressure swings.
• Best practice: deploy in arrays; replicate; calibrate each unit.

③ Heat as a tracer (streambed probes, FO-DTS, airborne TIR)

Three related techniques that share a common physics. Streambed probes: vertical arrays of thermistors at multiple depths; the diel and seasonal surface-water temperature signal propagates down differently depending on vertical flux direction and magnitude (1-D analytical or numerical inverse solution). Fiber-optic distributed temperature sensing (FO-DTS): a cable along the streambed records temperature at meter-scale resolution along an entire reach, identifying point and diffuse discharge zones from temperature-contrast and buffering signatures. Airborne thermal infrared (TIR): aerial or drone-mounted sensors map the surface-temperature field of the river to find GW discharge plumes. Per Ma et al. (2024), these are heat-as-tracer methods, not geophysical methods.

• Provides: Streambed probes — vertical Darcy flux (m/d) at the point, from a 1-D heat-transport inversion of the temperature time-series. FO-DTS — meter-resolution temperature distribution along the cable, identifying the location of GW-discharge zones (cold-summer / warm-winter signatures); flux is qualitative. Airborne TIR — basin-scale map of surface water-temperature anomalies that locate GW plumes; not a quantitative flux.
• Scale: point (probes) to short reach (FO-DTS) to basin (TIR).
• Strengths: non-invasive once installed; high-frequency, high-density data; captures both gaining and losing conditions; deployment cost can be tuned to study budget (Ma et al. 2024).
• Limitations: needs a GW–SW temperature gradient — small differences increase imprecision and extreme fluxes are undetectable; inverse solution sensitive to thermal-property assumptions; FO-DTS and TIR locate exchange but rarely quantify the flux; TIR does not penetrate the water column.
• Best in: systems with strong seasonal or diel temperature contrast; works in moving and still water; perennial and intermittent streams.
• Poor for: cool gaining systems with little year-round T contrast; thermally stratified poorly mixed pools (biases FO-DTS and TIR); high-energy scouring reaches where probe arrays don't survive. Tidal: usable, but tidal heat advection must be modeled separately.
• Tools: VFLUX (R/Matlab), 1DTempPro (USGS).

④ Piezometer / minipiezometer + Darcy

Drive a small-diameter piezometer below the streambed to measure the head under the stream. Compare to stream stage → vertical hydraulic gradient (VHG). Combined with streambed K (slug test), gives a Darcy flux.

• Provides: hydraulic head (m) below the streambed at the screen depth. Differenced with stream stage → vertical hydraulic gradient (VHG, dimensionless); paired with streambed K (from a slug or grain-size test) → vertical Darcy flux (m/d) at the piezometer location.
• Scale: point.
• Strengths: direct head data; mature technique; quick to install; straightforward to analyze; can leverage existing well networks.
• Limitations: point-K is hard to upscale; only valid when streambed K is reasonably uniform under the piezometer; lab permeameter tests disturb sediment structure and miss preferential pathways.
• Best in: perennial and intermittent streams with installable (sand/silt/fine gravel) beds; works across moving and still water; tidal sites are workable if heads are recorded continuously and corrected for tidal loading.
• Poor for: armored cobble or highly compacted beds (installation); frozen ground or permafrost; very high-flow channels where scour or floodwater damage equipment.

⑤ Pumping/aquifer test interpreted for depletion

A constant-rate pumping test analyzed with stream-aquifer drawdown solutions (Hunt 1999, Hantush 1965). Observation wells between the well and the stream reveal stream influence as a divergence from Theis drawdown.

• Provides: drawdown time-series at observation wells between the pumping well and the stream. Fitted to stream-aquifer drawdown solutions → estimates of transmissivity T, storativity S, and streambed conductance λ — the parameters that drive analytical predictions of stream depletion Q_s(t)/Q under any pumping schedule (page 02).
• Scale: reach (~100s of m to km).
• Strengths: estimates the depletion parameters that matter for prediction (T, S, λ) in a single test; observation wells reveal real connection state.
• Limitations: expensive; requires sufficient pumping duration for the stream signal to emerge above background noise; non-unique inversions (Lough & Hunt 2006).
• Best in: low-flow / dry-season periods when stream stage is stable; wells close enough to the stream that drawdown propagates within the test duration.
• Poor for: tidally-influenced reaches (stream stage not constant); periods of active recharge or fluctuating stream stage; intermittent streams during dry conditions when there is no surface water to deplete.

⑥ Geophysics (ERT, SP, IP, EM, GPR, seismic)

Electrical resistivity (ERT), self-potential (SP), induced polarization (IP), electromagnetic induction (EM), ground-penetrating radar (GPR), and seismic surveys reveal the geometry of saturated zones, clay lenses, streambed thickness, and stream-aquifer connection state. Often the only practical way to image the GW–SW interface where intrusive instruments can't be installed or where the question is the spatial extent of the connected zone.

• Provides: spatially-distributed bulk physical properties of the subsurface — electrical resistivity (ERT, EM), dielectric contrast (GPR), seismic velocity (seismic), or natural electric potential (SP). Interpreted to map saturated/unsaturated zones, streambed-layer thickness, depth to bedrock, salinity contrasts, and disconnection zones. Generally yields geometry and spatial extent, not flux magnitude; SP can additionally indicate streaming-potential signals associated with seepage.
• Scale: reach to basin (vertical resolution cm to 100s of m).
• Strengths: minimally invasive; spatially referenced; images the GW–SW interface beyond the channel (under banks and alluvium); identifies disconnection zones; well-suited to monitoring dynamic hydrological processes (Ma et al. 2024).
• Limitations: exchange rate is hard to quantify from geophysics alone (gives geometry, not flux); inversion non-unique; signals are jointly sensitive to geology and pore-water properties; requires geophysics expertise and calibration against intrusive methods.
• Best in: all SW settings — a primary choice for tidal, marine, and brackish environments and for sites with inland saline GW or contamination, where electrical contrast is strong; protected or hard-to-access reaches.
• Poor for: highly conductive SW that masks subsurface contrast (limits resistivity / EM); GPR in clay-rich or saline media (signal attenuates rapidly); applications that need a quick quantitative flux without intrusive follow-up.

⑦ Groundwater ET measurement

Eddy-covariance flux towers, sap-flow sensors, or analysis of nighttime water-table fluctuations (White method) quantify phreatophyte use of shallow groundwater. Critical for the "captured ET" term in the depletion budget (page 03).

• Provides: evapotranspiration rate (mm/d). Eddy covariance — latent-heat flux (W/m²) converted to ET over the tower's fetch footprint. Sap flow — per-stem transpiration scaled to stand-level ET. White method — daily ET (mm/d) from the diel water-table fluctuation amplitude. Partitioning the total ET into GW-supplied vs. SW-supplied components requires additional analysis.
• Scale: patch (sap flow, White method) to landscape (eddy covariance).
• Strengths: closes the water budget by quantifying the captured-ET term; strong constraint on numerical models.
• Limitations: instrumentation cost; strict siting and fetch requirements; partitioning SW-ET from GW-ET is hard; weather conditions affect eddy-covariance closure.
• Best in: riparian and phreatophyte stands with shallow water tables; not flow-regime-dependent (works across perennial, intermittent, and disconnected reaches).
• Poor for: deep-water-table settings where ET is not GW-supported; small isolated GDEs below the EC tower's footprint; winter / dormant-season at high latitudes.

⑧ Tracers (isotopes, dye, age)

Stable isotopes (²H, ¹⁸O, ³H, ¹⁴C, ²²²Rn, Ra, SF₆), major ions, fluorescent dyes, and injected solute tracers distinguish stream water from groundwater of different ages and travel paths. Useful for confirming connectivity, identifying discrete discharge zones, and unmixing sources in complex reaches.

• Provides: depends on the tracer. Environmental tracers (²H, ¹⁸O, ²²²Rn, major ions) — concentrations in stream and GW → end-member mixing fractions (% GW vs. % SW) along a reach. Age tracers (³H, ¹⁴C, CFCs, SF₆) — apparent GW age → mean residence time and inferred recharge rates. Injected solute tracers — breakthrough curves at downstream samplers → reach-integrated transient storage, dispersion, and GW inflow rate. Dye — travel time, dispersion, connectivity confirmation.
• Scale: reach to basin; environmental tracers span sample-point to ~120 km reach lengths.
• Strengths: fingerprints sources; reach-scale spatial averaging (injected); long-term flow-path information (environmental tracers using ¹⁴C, ³H); complements physical flux methods.
• Limitations: lab cost; equifinality in mixing models; injected tracers usually require steady flow; biogeochemical reactions can alter tracer chemistry; environmental-tracer end-member concentrations are often uncertain in space and time; concentrations near the analytical detection limit yield unresolvable (not inflated) flux estimates, and LOD-substitution conventions can bias mass balances.
• Best in: systems with strong isotopic or chemical contrast between GW and SW; reach-scale source identification at low to moderate flows.
• Poor for: highly dynamic flows (steady-state assumption fails for injected tracers and many environmental-tracer interpretations); tidal reaches where fresh-marine end-members vary in space and time; systems with strong biogeochemical alteration of the tracer.

Sample across time

• Wet vs. dry season — exchange direction can reverse.
• Diel cycle — ET drawdown and stream-stage diel signals reveal connection state.
• Pre- and post-irrigation season — to capture pumping effects.
• Wet-year and dry-year deployments to capture climate variability.