 |
| Titel |
Improving prediction of hydraulic conductivity by constraining capillary bundle models to a maximum pore size |
| VerfasserIn |
Sascha Iden, Andre Peters, Wolfgang Durner |
| Konferenz |
EGU General Assembly 2017
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 19 (2017) |
| Datensatznummer |
250146345
|
| Publikation (Nr.) |
 EGU/EGU2017-10367.pdf |
|
|
|
|
|
| Zusammenfassung |
| Soil hydraulic properties are required to solve the Richards equation, the most widely applied
model for variably-saturated flow. While the experimental determination of the water
retention curve does not pose significant challenges, the measurement of unsaturated
hydraulic conductivity is time consuming and costly. The prediction of the unsaturated
hydraulic conductivity curve from the soil water retention curve by pore-bundle models is a
cost-effective and widely applied technique. A well-known problem of conductivity
prediction for retention functions with wide pore-size distributions is the sharp drop in
conductivity close to water saturation. This problematic behavior is well known for the van
Genuchten model if the shape parameter n assumes values smaller than about 1.3. So far, the
workaround for this artefact has been to introduce an explicit air-entry value into the capillary
saturation function. However, this correction leads to a retention function which is not
continuously differentiable and thus a discontinuous water capacity function. We
present an improved parametrization of the hydraulic properties which uses the
original capillary saturation function and introduces a maximum pore radius only
in the pore-bundle model. Closed-form equations for the hydraulic conductivity
function were derived for the unimodal and multimodal retention functions of van
Genuchten and have been tested by sensitivity analysis and applied in curve fitting
and inverse modeling of multistep outflow experiments. The resulting hydraulic
conductivity function is smooth, increases monotonically close to saturation, and
eliminates the sharp drop in conductivity close to saturation. Furthermore, the new
model retains the smoothness and continuous differentiability of the water retention
curve. We conclude that the resulting soil hydraulic functions are physically more
reasonable than the ones predicted by previous approaches, and are thus ideally suited
for numerical simulations with the Richards equation or multiphase flow models. |
|
|
|
|
|
|