Data Fusion Modeling for Groundwater Systems using Generalized Kalman Filtering

Engineering projects involving groundwater systems are faced with uncertainties because the earth is heterogeneous and data sets are fragmented. Current methods providing support for management decisions are limited by the data types, models, computations, or simplifications. Data fusion modeling (DFM) removes many of the limitations and provides predictive modeling to help close the management control loop. DFM is a spatial and temporal state estimation and system identification methodology that uses three sources of information: measured data, physical laws, and statistical models for uncertainty in spatial heterogeneities and for temporal variation in driving terms. Kalman filtering methods are generalized by introducing information filtering methods due to Bierman coupled with (1) a Markov random field representation for spatial variation and (2) the representer method for transient dynamics from physical oceanography. DFM provides benefits for waste management, water supply, and geotechnical applications.