Engineering projects involving hydrogeology are faced with uncertainties be
cause the earth is heterogeneous, and typical data sets are fragmented and
disparate. In theory, predictions provided by computer simulations using ca
librated models constrained by geological boundaries provide answers to sup
port management decisions, and geostatistical methods quantify safety margi
ns. In practice, current methods are limited by the data types and models t
hat can be included, computational demands, or simplifying assumptions. Dat
a Fusion Modeling (DFM) removes many of the limitations and is capable of p
roviding data integration and model calibration with quantified uncertainty
for a variety of hydrological, geological, and geophysical data types and
models. The benefits of DFM for waste management, water supply, and geotech
nical applications are savings in time and cost through the ability to prod
uce visual models that fill in missing data and predictive numerical models
to aid management optimization. DFM has the ability to update field-scale
models in real time using PC or workstation systems and is ideally suited f
or parallel processing implementation. DFM is a spatial state estimation an
d system identification methodology that uses three sources of information:
measured data, physical laws, and statistical models for uncertainty in sp
atial heterogeneities. What is new in DFM is the solution of the causality
problem in the data assimilation Kalman filter methods to achieve computati
onal practicality. The Kalman filter is generalized by introducing informat
ion filter methods due to Bierman coupled with a Markov random field repres
entation for spatial variation. A Bayesian penalty function is implemented
with Gauss-Newton methods. This leads to a computational problem similar to
numerical simulation of the partial differential equations (PDEs) of groun
dwater. In fact, extensions of PDE solver ideas to break down computations
over space form the computational heart of DFM. State estimates and uncerta
inties can be computed for heterogeneous hydraulic conductivity fields in m
ultiple geological layers from the usually sparse hydraulic conductivity da
ta and the often more plentiful head data. Further, a system identification
theory has been derived based on statistical likelihood principles. A maxi
mum likelihood theory is provided to estimate statistical parameters such a
s Markov model parameters that determine the geostatistical variogram. Fiel
d-scale application of DFM at the DOE Savannah River Site is presented and
compared with manual calibration. DFM calibration runs converge in less tha
n 1 h on a Pentium Pro PC for a 3D model with more than 15,000 nodes. Run t
ime is approximately linear with the number of nodes. Furthermore, conditio
nal simulation is used to quantify the statistical variability in model pre
dictions such as contaminant breakthrough curves. Published by Elsevier Sci
ence B.V.