 |
| Titel |
What is geological entropy and why measure it? A parsimonious approach for predicting transport behaviour in heterogeneous aquifers |
| VerfasserIn |
Marco Bianchi, Daniele Pedretti |
| Konferenz |
EGU General Assembly 2017
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 19 (2017) |
| Datensatznummer |
250138787
|
| Publikation (Nr.) |
 EGU/EGU2017-1909.pdf |
|
|
|
|
|
| Zusammenfassung |
| We present an approach to predict non-Fickian transport behaviour in alluvial aquifers from
knowledge of physical heterogeneity. This parsimonious approach is based on only two
measurable parameters describing the global variability and the structure of the hydraulic
conductivity (K) field: the variance of the ln(K) values (σY 2), and a newly developed index
of geological entropy (HR), based on the concept of Shannon information entropy. Both σY 2
and HR can be obtained from data collected during conventional hydrogeological
investigations and from the analysis of a representative model of the spatial distribution of K
classes (e.g. hydrofacies) over the domain of interest. The new index HR integrates multiple
characteristics of the K field, including the presence of well-connected features,
into a unique metric that quantifies the degrees of spatial disorder in the K field
structure. Stochastic simulations of tracer tests in synthetic K fields based on realistic
distributions of hydrofacies in alluvial aquifers are conducted to identify empirical
relations between HR, σY 2, and the first three central temporal moments of the
resulting breakthrough curves (BTCs). Results indicate that the first and second
moments tend to increase with spatial disorder (i.e, HR increasing). Conversely, high
values of the third moment (i.e. skewness), which indicate significant post-peak
tailing in the BTCs and non-Fickian transport behaviour, are observed in more
orderly structures (i.e, HR decreasing), or for very high σY 2 values. We show that
simple closed-form empirical expressions can be derived to describe the bivariate
dependency between the skewness of the BTC and corresponding pairs of HR
and σY 2. This dependency shows clear correlation for a broad range of structures
and Kvariability levels. Therefore, it provides an effective and broadly applicable
approach to explain and predict non-Fickian transport in real aquifers, such as those at
the well-known MADE site and at the Lawrence Livermore National Laboratory. |
|
|
|
|
|
|