DAGPY: A HIGHLY EFFICIENT OPEN-SOURCE PYTHON PACKAGE FOR DATA ASSIMILATION IN PRACTICAL GROUNDWATER MODELING VIA THE ENSEMBLE KALMAN FILTER

The Ensemble Kalman Filter (EnKF) has been a revolutionary technique with major applications for the assimilation of observations and reduction of uncertainty in real-time prediction of dynamical systems. More recently, extensive research has centered around the suitability of the EnKF to tackle complex parameter estimation problems in groundwater applications, with a focus on synthetic systems that are insightful and highlight the potential of the technique. Most implementations reported in the literature, however, are not freely available or accessible to the community, limiting the broad use of this powerful tool. With this in mind, we developed an open-source, community platform for data assimilation in groundwater modeling: DAGpy. This package features a robust implementation of the EnKF for the estimation of permeability and porosity fields in both synthetic and real applications by assimilating hydraulic head and concentration observations. DAGpy (i) supports different versions of the EnKF, (ii) is modular and object-oriented, (iii) can be run in series and in parallel for the forecasting and updating steps, (iv) represents different subsurface flow and transport processes implemented in MODFLOW via the Python interface Floppy, (v) supports standard I/O data formats, and (vi) provides a myriad of post-processing tools for result analysis and visualization. Here we illustrate the potential of DAGpy with two case studies. First, we present an analysis of a tracer test in a highly connected alluvial aquifer. And second, we use DAGpy to reconstruct heterogeneous hydraulic conductivity and porosity fields by assimilating jointly head and concentration observations.