Geology Reference
In-Depth Information
from a sparse one by adding additional measurement points. This algorithm
may not be useful for optimizing a dense network either by deleting the
measurement points or by shifting them. Bras and Rodriguez-Iturbe (1976)
applied multivariate estimation theory to obtain the aerial mean precipitation
of an event over a fixed area considering the spatial uncertainty and correlation
of process, correlation in measurement errors and non-homogeneous sampling
costs.
One of the challenges of monitoring network design in a fractured rock
setting is the heterogeneity of the rocks. Nativ et al. (1999) summarize the
activities and problems associated with the monitoring of contaminated
groundwater in porous, low permeability fractured chalk in the Negev Desert,
Israel. Mayer et al. (1994) have recently presented a related methodology for
designing a monitoring network for plume detection that is based on finding
an optimal design that considers three objectives: minimizing the number of
monitoring wells, maximizing the probability of detecting a contaminant
leak and minimizing the expected area of contamination at the time of
detection. They develop examples based on a two-dimensional areal flow
model and examine the factors influencing trade-offs between the three design
objectives. The intention in their work is to illustrate a decision model for
evaluating monitoring network designs at waste management facilities that
overlie fractured bedrock.
There is a recent work also on Optimization techniques applied in solving
groundwater flow management problems (Ahlfeld et al., 2002). Ahlfeld et
al., in their work, used simple and advanced methods of linear optimization
techniques and includes also sensitivity and range analysis. MODOFC
software uses optimization techniques to find the best combination of pumping
rates and well locations to achieve criteria specified by the user. The response
of the groundwater system is modelled using the groundwater flow simulator.
Design criteria can be specified that encompass construction and operation
costs, pumping rates and volumes, water level elevations and groundwater
flow directions. However, the use of hydraulic head surface is the simplest
and quickest means of flow analysis and provide the most accurate flow
representation.
According to Bogardi et al. (1985), the observation network design itself
may involve the following interrelated factors: (1) observation effort (cost,
time, instrumentation, etc.); (2) relative importance of the various parameters;
(3) the different geostatistical properties of the parameters; and (4) estimation
accuracy or error criteria for the various parameters. The key precondition
for using geostatistics for observation network design is that the variograms
be available for the design parameters. Clearly if no observation network
exists, no variogram based on sample information over the entire area can
be available. For any network design model, one must have some information
on the overall areal variation of the parameter to be observed even if no
“hard” data set, that is, sample information on such variation exists.
