Conservation law
mass
Residual equation
$$R = Q_{\mathrm{GW\rightarrow SW}} - Q_{\mathrm{SW\rightarrow GW}} - Q_{\mathrm{GW,net}}$$
where $Q_{\mathrm{GW\rightarrow SW}}$ is gross groundwater discharge to surface water, $Q_{\mathrm{SW\rightarrow GW}}$ is gross surface-water leakage to groundwater, and $Q_{\mathrm{GW,net}}$ is the independently reported net groundwater contribution to surface water. All terms have units of water volume per unit time ($L^3T^{-1}$), or equivalent depth per unit time ($LT^{-1}$).
Tolerance and its denominator
The relative residual is evaluated against the gross groundwater–surface water exchange and must remain below the suite's 5% engineering tolerance. Gross exchange is used as the denominator because large opposing groundwater–surface water fluxes may produce a small net exchange. Normalizing by the net flux would become unstable when gaining and losing fluxes nearly cancel. An absolute tolerance is used when gross exchange approaches zero.
How would an unphysical model fail this?
A model can close its overall water budget while reporting internally inconsistent groundwater–surface water exchange components. For example, it may report 100 units of groundwater discharge to surface water and 30 units of surface-water leakage to groundwater, but report a net groundwater contribution of 100 rather than 70.
Such a model could still satisfy the catchment-scale water budget if the discrepancy is compensated by another storage or boundary term. A conventional global budget-closure test could therefore pass it, whereas this probe would fail because
$$ 100-30-100=-30. $$
Other deliberately broken models can reverse the sign convention or omit one direction of exchange while preserving the global water balance. The exchange-consistency residual catches these failures.
How is the case generated?
Generate a small synthetic coupled groundwater–surface water system using a fixed random seed. The case contains both gaining and losing surface-water reaches so that groundwater-to-surface-water and surface-water-to-groundwater fluxes are simultaneously nonzero.
The forcing and boundary conditions are chosen to produce appreciable exchange in both directions while keeping the system numerically simple. The model is asked to report gross groundwater-to-surface-water exchange, gross surface-water-to-groundwater exchange, and net groundwater contribution to surface water over the same spatial domain and time window.
No observational data or prescribed "correct" exchange magnitude is required. The probe tests only the mass-accounting identity among the three model-reported quantities.
Will you build it?
Yes — assign it to me
Conservation law
mass
Residual equation
where$Q_{\mathrm{GW\rightarrow SW}}$ is gross groundwater discharge to surface water, $Q_{\mathrm{SW\rightarrow GW}}$ is gross surface-water leakage to groundwater, and $Q_{\mathrm{GW,net}}$ is the independently reported net groundwater contribution to surface water. All terms have units of water volume per unit time ($L^3T^{-1}$ ), or equivalent depth per unit time ($LT^{-1}$ ).
Tolerance and its denominator
The relative residual is evaluated against the gross groundwater–surface water exchange and must remain below the suite's 5% engineering tolerance. Gross exchange is used as the denominator because large opposing groundwater–surface water fluxes may produce a small net exchange. Normalizing by the net flux would become unstable when gaining and losing fluxes nearly cancel. An absolute tolerance is used when gross exchange approaches zero.
How would an unphysical model fail this?
A model can close its overall water budget while reporting internally inconsistent groundwater–surface water exchange components. For example, it may report 100 units of groundwater discharge to surface water and 30 units of surface-water leakage to groundwater, but report a net groundwater contribution of 100 rather than 70.
Such a model could still satisfy the catchment-scale water budget if the discrepancy is compensated by another storage or boundary term. A conventional global budget-closure test could therefore pass it, whereas this probe would fail because
Other deliberately broken models can reverse the sign convention or omit one direction of exchange while preserving the global water balance. The exchange-consistency residual catches these failures.
How is the case generated?
Generate a small synthetic coupled groundwater–surface water system using a fixed random seed. The case contains both gaining and losing surface-water reaches so that groundwater-to-surface-water and surface-water-to-groundwater fluxes are simultaneously nonzero.
The forcing and boundary conditions are chosen to produce appreciable exchange in both directions while keeping the system numerically simple. The model is asked to report gross groundwater-to-surface-water exchange, gross surface-water-to-groundwater exchange, and net groundwater contribution to surface water over the same spatial domain and time window.
No observational data or prescribed "correct" exchange magnitude is required. The probe tests only the mass-accounting identity among the three model-reported quantities.
Will you build it?
Yes — assign it to me