|
Titel |
Stochastic Modeling of the Environmental Impacts of the Mingtang Tunneling
Project |
VerfasserIn |
Xiaojun Li, Yandong Li, Ching-Fu Chang, Ziyang Chen, Benjamin Zhi Wen Tan, Jon Sege, Changhong Wang, Yoram Rubin |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250140152
|
Publikation (Nr.) |
EGU/EGU2017-3501.pdf |
|
|
|
Zusammenfassung |
This paper investigates the environmental impacts of a major tunneling project in China. Of
particular interest is the drawdown of the water table, due to its potential impacts on
ecosystem health and on agricultural activity. Due to scarcity of data, the study pursues a
Bayesian stochastic approach, which is built around a numerical model. We adopted the
Bayesian approach with the goal of deriving the posterior distributions of the dependent
variables conditional on local data. The choice of the Bayesian approach for this study is
somewhat non-trivial because of the scarcity of in-situ measurements. The thought guiding
this selection is that prior distributions for the model input variables are valuable tools even if
that all inputs are available, the Bayesian approach could provide a good starting point for
further updates as and if additional data becomes available. To construct effective
priors, a systematic approach was developed and implemented for constructing
informative priors based on other, well-documented sites which bear geological
and hydrological similarity to the target site (the Mingtang tunneling project). The
approach is built around two classes of similarity criteria: a physically-based set of
criteria and an additional set covering epistemic criteria. The prior construction
strategy was implemented for the hydraulic conductivity of various types of rocks
at the site (Granite and Gneiss) and for modeling the geometry and conductivity
of the fault zones. Additional elements of our strategy include (1) modeling the
water table through bounding surfaces representing upper and lower limits, and
(2) modeling the effective conductivity as a random variable (varying between
realizations, not in space). The approach was tested successfully against its ability
to predict the tunnel infiltration fluxes and against observations of drying soils. |
|
|
|
|
|