Blind source separation for groundwater pressure analysis based on nonnegative matrix factorization

The identification of the physical sources causing spatial and temporal fluctuations of aquifer water levels is a challenging, yet a very important hydrogeological task. The fluctuations can be caused by variations in natural and anthropogenic sources such as pumping, recharge, barometric pressures, etc. The source identification can be crucial for conceptualization of the hydrogeological conditions and characterization of aquifer properties. We propose a new computational framework for model-free inverse analysis of pressure transients based on Nonnegative Matrix Factorization (NMF) method for Blind Source Separation (BSS) coupled with k-means clustering algorithm, which we call NMFk. NMFk is capable of identifying a set of unique sources from a set of experimentally measured mixed signals, without any information about the sources, their transients, and the physical mechanisms and properties controlling the signal propagation through the subsurface flow medium. Our analysis only requires information about pressure transients at a number of observation points, m, where m≥r, and r is the number of unknown unique sources causing the observed fluctuations. We apply this new analysis on a data set from the Los Alamos National Laboratory site. We demonstrate that the sources identified by NMFk have real physical origins: barometric pressure and water-supply pumping effects. We also estimate the barometric pressure efficiency of the monitoring wells. The possible applications of the NMFk algorithm are not limited to hydrogeology problems; NMFk can be applied to any problem where temporal system behavior is observed at multiple locations and an unknown number of physical sources are causing these fluctuations.

[1]  Christian Jutten,et al.  Space or time adaptive signal processing by neural network models , 1987 .

[2]  S. P. Neuman,et al.  Three‐dimensional numerical inversion of pneumatic cross‐hole tests in unsaturated fractured tuff: 2. Equivalent parameters, high‐resolution stochastic imaging and scale effects , 2001 .

[3]  A Revil,et al.  A Potential‐Based Inversion of Unconfined Steady‐State Hydraulic Tomography , 2009, Ground water.

[4]  K. Karasaki,et al.  Estimation of reservoir properties using transient pressure data: An asymptotic approach , 2000 .

[5]  Frédéric Frappart,et al.  An independent component analysis filtering approach for estimating continental hydrology in the GRACE gravity data , 2011 .

[6]  Todd C. Rasmussen,et al.  Identifying and Removing Barometric Pressure Effects in Confined and Unconfined Aquifers , 1997 .

[7]  H. Sebastian Seung,et al.  Unsupervised Learning by Convex and Conic Coding , 1996, NIPS.

[8]  Xin Liu,et al.  Document clustering based on non-negative matrix factorization , 2003, SIGIR.

[9]  Walter A. Illman,et al.  Capturing aquifer heterogeneity: Comparison of approaches through controlled sandbox experiments , 2011 .

[10]  Asoke K. Nandi,et al.  Blind Source Separation , 1999 .

[11]  Christopher J. Keylock,et al.  A resampling method for generating synthetic hydrological time series with preservation of cross‐correlative structure and higher‐order properties , 2012 .

[12]  Andres Alcolea,et al.  Inverse problem in hydrogeology , 2005 .

[13]  Christian Jutten,et al.  Blind separation of sources, part I: An adaptive algorithm based on neuromimetic architecture , 1991, Signal Process..

[14]  P. Paatero,et al.  Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values† , 1994 .

[15]  T. Yeh,et al.  Stochastic inversion of pneumatic cross-hole tests and barometric pressure fluctuations in heterogeneous unsaturated formations , 2008 .

[16]  David T. W. Jones,et al.  Signatures of mutational processes in human cancer , 2013, Nature.

[17]  Alan Julian Izenman,et al.  Modern Multivariate Statistical Techniques , 2008 .

[18]  T. Yeh,et al.  Hydraulic tomography: Development of a new aquifer test method , 2000 .

[19]  T. Rasmussen,et al.  Removal of Barometric Pressure Effects and Earth Tides from Observed Water Levels , 2005, Ground water.

[20]  Jürgen Kusche,et al.  Separation of global time-variable gravity signals into maximally independent components , 2012, Journal of Geodesy.

[21]  Hiromitsu Saegusa,et al.  Hydraulic tomography in fractured granite: Mizunami Underground Research site, Japan , 2009 .

[22]  Velimir V Vesselinov An alternative conceptual model of groundwater flow and transport in saturated zone beneath the Pajarito Plateau , 2005 .

[23]  M. Stratton,et al.  Deciphering Signatures of Mutational Processes Operative in Human Cancer , 2013, Cell reports.

[24]  D. E. Broxton,et al.  Geologic Framework of a Groundwater System on the Margin of a Rift Basin, Pajarito Plateau, North‐Central New Mexico , 2005 .

[25]  Martin A. Fischler,et al.  The Representation and Matching of Pictorial Structures , 1973, IEEE Transactions on Computers.

[26]  Upmanu Lall,et al.  Modeling multivariable hydrological series: Principal component analysis or independent component analysis? , 2007 .

[27]  Andrzej Cichocki,et al.  A New Learning Algorithm for Blind Signal Separation , 1995, NIPS.

[28]  B. Mirus,et al.  Advanced Methods for Modeling Water- Levels and Estimating Drawdowns with SeriesSEE, an Excel Add-In , 2014 .

[29]  Ziauddin Muhammed Kamran. Blind source separation using higher-order statistics. , 2000 .

[30]  Gene H. Golub,et al.  An analysis of the total least squares problem , 1980, Milestones in Matrix Computation.

[31]  H. Sebastian Seung,et al.  Learning the parts of objects by non-negative matrix factorization , 1999, Nature.

[32]  Andrzej Cichocki,et al.  Nonnegative Matrix and Tensor Factorization T , 2007 .

[33]  Jürgen Kusche,et al.  Separation of deterministic signals using independent component analysis (ICA) , 2012, Studia Geophysica et Geodaetica.

[34]  H. Sebastian Seung,et al.  Algorithms for Non-negative Matrix Factorization , 2000, NIPS.

[35]  C. V. Theis The relation between the lowering of the Piezometric surface and the rate and duration of discharge of a well using ground‐water storage , 1935 .

[36]  Erkki Oja,et al.  Independent component analysis: algorithms and applications , 2000, Neural Networks.

[37]  Walter A. Illman,et al.  Three‐dimensional transient hydraulic tomography in a highly heterogeneous glaciofluvial aquifer‐aquitard system , 2011 .

[38]  Josh H. McDermott The cocktail party problem , 2009, Current Biology.

[39]  M. Watkins,et al.  GRACE Measurements of Mass Variability in the Earth System , 2004, Science.

[40]  Velimir V. Vesselinov,et al.  Three‐dimensional numerical inversion of pneumatic cross‐hole tests in unsaturated fractured tuff: 1. Methodology and borehole effects , 2001 .

[41]  J. A. Hartigan,et al.  A k-means clustering algorithm , 1979 .

[42]  S. Srivastava,et al.  Characterization of guanidiniocarbonyl pyrroles in water by pH-dependent UV Raman spectroscopy and component analysis. , 2008, Physical chemistry chemical physics : PCCP.

[43]  Frédéric Frappart,et al.  Denoising Satellite Gravity Signals by Independent Component Analysis , 2010, IEEE Geoscience and Remote Sensing Letters.

[44]  P. Rousseeuw Silhouettes: a graphical aid to the interpretation and validation of cluster analysis , 1987 .

[45]  Velimir V Vesselinov,et al.  Identification of Pumping Influences in Long‐Term Water Level Fluctuations , 2011, Ground water.

[46]  John H. Maindonald,et al.  Modern Multivariate Statistical Techniques: Regression, Classification and Manifold Learning , 2009 .