 |
| Titel |
Geostatistical description of lithofacies distribution in the aquifer system of Cremona, Italy. |
| VerfasserIn |
L. Guadagnini, M. Riva, M. Salmaso, F. Saraceni, S. Straface, A. Guadagnini |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250022932
|
|
|
|
|
|
| Zusammenfassung |
| We develop two alternative conceptual models to describe the heterogeneous spatial
distribution of geomaterials within the groundwater system in the proximity of the city of
Cremona, Italy. The key hydrogeologic feature of the region is the occurrence of the Springs
Belt which develops across the entire Lombardia region and provides a major source of fresh
water for agricultural needs. During recent years the natural springs of the Cremona aquifer
have been increasingly threatened by over-abstraction and contamination by agricultural
fertilizers. The area investigated includes the main natural springs in the region, and is located
between the Adda and the Serio rivers, covering a surface of approximately 785 km2. The
groundwater system is constituted by two main productive aquifers, which are separated by a
locally discontinuous aquitard. The vertical variability of geomaterials distribution inferred
from available well logs suggests that the system is relatively heterogeneous on the given
observation scale. Lithofacies distribution within each identified aquifer is estimated upon
considering two alternative conceptual models: (a) a composite medium scheme, and (b) a
multiple-continua approach. In the former scenario, the system is conceptualized as
composed by disjoint blocks of different materials, the boundaries of which can
be uncertain. The latter approach assumes that the porous medium is composed
by a set of overlapping continua, whose relative fraction at a given location can
be uncertain. We start by classifying available sedimentological information and
group the various identified lithotypes into five separate clusters. An extension
of the indicator-based approach of Guadagnini et al. [2004] is then developed in
order to provide a geostatistical chracterization of lithotypes distribution when the
system is described as a composite medium. A multi-continua description is achieved
by means of multiple indicator Kriging techniques. With the aid of formal model
selection criteria we associate different tested indicator variogram models (and
associated parameters) with a weight, or posterior probability, representing their
relative degrees of likelihood. Interestingly, we noted that some of the weights are not
small enough to justify adopting one variogram model at the exclusion of all the
others. This suggests that predicting the aquifer behavior to future forcings and
assessing the corresponding predictive uncertainty may best be accomplished using
a multimodel approach such as maximum likelihood Bayesian model averaging
(MLBMA).
References
Guadagnini, L., A. Guadagnini, and D.M. Tartakovsky (2004), Probabilistic
Reconstruction of geologic facies, J. of Hydrol., 294, 57-67. |
|
|
|
|
|
|