Long short-term memory embedded nudging schemes for nonlinear data assimilation of geophysical flows

Reduced rank nonlinear filters are increasingly utilized in data assimilation of geophysical flows, but often require a set of ensemble forward simulations to estimate forecast covariance. On the other hand, predictor-corrector type nudging approaches are still attractive due to their simplicity of implementation when more complex methods need to be avoided. However, optimal estimate of nudging gain matrix might be cumbersome. In this paper, we put forth a fully nonintrusive recurrent neural network approach based on a long short-term memory (LSTM) embedding architecture to estimate the nudging term, which plays a role not only to force the state trajectories to the observations but also acts as a stabilizer. Furthermore, our approach relies on the power of archival data and the trained model can be retrained effectively due to power of transfer learning in any neural network applications. In order to verify the feasibility of the proposed approach, we perform twin experiments using Lorenz 96 system. Our results demonstrate that the proposed LSTM nudging approach yields more accurate estimates than both extended Kalman filter (EKF) and ensemble Kalman filter (EnKF) when only sparse observations are available. With the availability of emerging AI-friendly and modular hardware technologies and heterogeneous computing platforms, we articulate that our simplistic nudging framework turns out to be computationally more efficient than either the EKF or EnKF approaches.

[1]  J. M. Lewis,et al.  Dynamic Data Assimilation: A Least Squares Approach , 2006 .

[2]  Peter Jan,et al.  Particle Filtering in Geophysical Systems , 2009 .

[3]  K. Emanuel,et al.  Optimal Sites for Supplementary Weather Observations: Simulation with a Small Model , 1998 .

[4]  Redouane Lguensat,et al.  The Analog Data Assimilation , 2017 .

[5]  Lee A. Feldkamp,et al.  Neurocontrol of nonlinear dynamical systems with Kalman filter trained recurrent networks , 1994, IEEE Trans. Neural Networks.

[6]  Christopher K. Wikle,et al.  Atmospheric Modeling, Data Assimilation, and Predictability , 2005, Technometrics.

[7]  D. Stauffer,et al.  Use of Four-Dimensional Data Assimilation in a Limited-Area Mesoscale Model. Part I: Experiments with Synoptic-Scale Data , 1990 .

[8]  L. Biferale,et al.  Synchronization to Big Data: Nudging the Navier-Stokes Equations for Data Assimilation of Turbulent Flows , 2019, Physical Review X.

[9]  J. Blum,et al.  Back and forth nudging algorithm for data assimilation problems , 2005 .

[10]  Ionel M. Navon,et al.  A reduced‐order approach to four‐dimensional variational data assimilation using proper orthogonal decomposition , 2007 .

[11]  D. Lohse,et al.  Scaling and dissipation in the GOY shell model , 1994, chao-dyn/9409001.

[12]  Henry D. I. Abarbanel,et al.  Predicting the Future: Completing Models of Observed Complex Systems , 2013 .

[13]  E. Lorenz Predictability of Weather and Climate: Predictability – a problem partly solved , 2006 .

[14]  Fumin Zhang,et al.  An LSTM based Kalman Filter for Spatio-temporal Ocean Currents Assimilation , 2019, WUWNet.

[15]  B. R. Noack,et al.  Optimal nonlinear eddy viscosity in Galerkin models of turbulent flows , 2014, Journal of Fluid Mechanics.

[16]  G. Evensen,et al.  Analysis Scheme in the Ensemble Kalman Filter , 1998 .

[17]  Mehdi Ghommem,et al.  pyROM: A computational framework for reduced order modeling , 2019, J. Comput. Sci..

[18]  Sepp Hochreiter,et al.  The Vanishing Gradient Problem During Learning Recurrent Neural Nets and Problem Solutions , 1998, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[19]  Istvan Szunyogh,et al.  A Local Ensemble Kalman Filter for Atmospheric Data Assimilation , 2002 .

[20]  Jaideep Pathak,et al.  Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data. , 2017, Chaos.

[21]  Adrian Sandu,et al.  POD/DEIM reduced-order strategies for efficient four dimensional variational data assimilation , 2014, J. Comput. Phys..

[22]  A. Piacentini,et al.  Determination of optimal nudging coefficients , 2003 .

[23]  W. Zeng,et al.  Comparison of the use of a physical-based model with data assimilation and machine learning methods for simulating soil water dynamics , 2020 .

[24]  Alexander Gluhovsky,et al.  The structure of energy conserving low-order models , 1999 .

[25]  H. Storch,et al.  Optimal Spectral Nudging for Global Dynamic Downscaling , 2017 .

[26]  Petros Koumoutsakos,et al.  Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks , 2018, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[27]  A. Mazzino,et al.  Inferring flow parameters and turbulent configuration with physics-informed data assimilation and spectral nudging , 2018, Physical Review Fluids.

[28]  Marc Bocquet,et al.  Bayesian inference of dynamics from partial and noisy observations using data assimilation and machine learning , 2020, ArXiv.

[29]  T. N. Krishnamurti,et al.  Physical initialization for numerical weather prediction over the tropics , 1991 .

[30]  R. S. Bell,et al.  The Meteorological Office analysis correction data assimilation scheme , 1991 .

[31]  Anuj Karpatne,et al.  Physics-Guided Machine Learning for Scientific Discovery: An Application in Simulating Lake Temperature Profiles , 2020, Trans. Data Sci..

[32]  Jeffrey L. Anderson A Non-Gaussian Ensemble Filter Update for Data Assimilation , 2010 .

[33]  Ionel M. Navon,et al.  Data Assimilation for Geophysical Fluids , 2009 .

[34]  Jürgen Schmidhuber,et al.  Kalman filters improve LSTM network performance in problems unsolvable by traditional recurrent nets , 2003, Neural Networks.

[35]  Milija Zupanski,et al.  Comparison of sequential data assimilation methods for the Kuramoto–Sivashinsky equation , 2009 .

[36]  Georgiy L. Stenchikov,et al.  Spectral nudging to eliminate the effects of domain position and geometry in regional climate model simulations , 2004 .

[37]  Pejman Shoeibi Omrani,et al.  Deep Learning and Data Assimilation for Real-Time Production Prediction in Natural Gas Wells , 2018, ArXiv.

[38]  Francisco J. Gonzalez,et al.  Deep convolutional recurrent autoencoders for learning low-dimensional feature dynamics of fluid systems , 2018, ArXiv.

[39]  A.H. Haddad,et al.  Applied optimal estimation , 1976, Proceedings of the IEEE.

[40]  Ivan Oseledets,et al.  Predicting dynamical system evolution with residual neural networks , 2019, Keldysh Institute Preprints.

[41]  Edriss S. Titi,et al.  Continuous Data Assimilation Using General Interpolant Observables , 2013, J. Nonlinear Sci..

[42]  C. C. Pain,et al.  Rapid spatio-temporal flood prediction and uncertainty quantification using a deep learning method , 2019, Journal of Hydrology.

[43]  J. Blum,et al.  A nudging-based data assimilation method: the Back and Forth Nudging (BFN) algorithm , 2008 .

[44]  G. Evensen Data Assimilation: The Ensemble Kalman Filter , 2006 .

[45]  Christopher C. Pain,et al.  Optimal reduced space for Variational Data Assimilation , 2019, J. Comput. Phys..

[46]  Theodore B. Trafalis,et al.  Machine Learning Methods for Data Assimilation , 2010 .

[47]  Jeffrey L. Anderson,et al.  Scalable Implementations of Ensemble Filter Algorithms for Data Assimilation , 2007 .

[48]  Jaideep Pathak,et al.  Backpropagation algorithms and Reservoir Computing in Recurrent Neural Networks for the forecasting of complex spatiotemporal dynamics , 2019, Neural Networks.

[49]  Wenjie Zhang,et al.  Data-driven reduced order model with temporal convolutional neural network , 2020 .

[50]  Kookjin Lee,et al.  Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders , 2018, J. Comput. Phys..

[51]  Henry D. I. Abarbanel,et al.  Machine Learning: Deepest Learning as Statistical Data Assimilation Problems , 2017, Neural Computation.

[52]  Marc Bocquet,et al.  Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: a case study with the Lorenz 96 model , 2019, J. Comput. Sci..

[53]  Ionel M. Navon,et al.  An Optimal Nudging Data Assimilation Scheme Using Parameter Estimation , 1992 .

[54]  Eric Blayo,et al.  A reduced-order strategy for 4D-Var data assimilation , 2005, 0709.2825.

[55]  Juan Du,et al.  Parameterised non-intrusive reduced order methods for ensemble Kalman filter data assimilation , 2018, Computers & Fluids.

[56]  L. Agostini Exploration and prediction of fluid dynamical systems using auto-encoder technology , 2020 .

[57]  J. L. Roux An Introduction to the Kalman Filter , 2003 .

[58]  Jaideep Pathak,et al.  Model-Free Prediction of Large Spatiotemporally Chaotic Systems from Data: A Reservoir Computing Approach. , 2018, Physical review letters.

[59]  Alexander Gluhovsky,et al.  Effective low-order models for atmospheric dynamics and time series analysis. , 2016, Chaos.

[60]  D. Simon Optimal State Estimation: Kalman, H Infinity, and Nonlinear Approaches , 2006 .

[61]  Dit-Yan Yeung,et al.  Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting , 2015, NIPS.

[62]  H. Storch,et al.  A Spectral Nudging Technique for Dynamical Downscaling Purposes , 2000 .

[63]  Omer San,et al.  Data-driven recovery of hidden physics in reduced order modeling of fluid flows , 2020, Physics of Fluids.

[64]  G. Evensen Sequential data assimilation with a nonlinear quasi‐geostrophic model using Monte Carlo methods to forecast error statistics , 1994 .

[65]  Petros Koumoutsakos,et al.  Machine Learning for Fluid Mechanics , 2019, Annual Review of Fluid Mechanics.

[66]  Ionel M. Navon,et al.  Efficiency of a POD-based reduced second-order adjoint model in 4 D-Var data assimilation , 2006 .

[67]  S. Lakshmivarahan,et al.  Nudging Methods: A Critical Overview , 2013 .

[68]  Michael W. Mahoney,et al.  Physics-informed Autoencoders for Lyapunov-stable Fluid Flow Prediction , 2019, ArXiv.

[69]  A. Carrassi,et al.  Data assimilation by delay‐coordinate nudging , 2015, 1510.07884.

[70]  H. Abarbanel,et al.  Accurate state and parameter estimation in nonlinear systems with sparse observations , 2014 .

[71]  Ditlevsen,et al.  Cascades and statistical equilibrium in shell models of turbulence. , 1996, Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics.

[72]  Dongbin Xiu,et al.  On generalized residue network for deep learning of unknown dynamical systems , 2020, J. Comput. Phys..

[73]  A. Segers,et al.  Machine learning for observation bias correction with application to dust storm data assimilation , 2019, Atmospheric Chemistry and Physics.

[74]  Peter R. Oke,et al.  A deterministic formulation of the ensemble Kalman filter : an alternative to ensemble square root filters , 2008 .

[75]  Julien Pettré,et al.  Data-Driven Crowd Simulation with Generative Adversarial Networks , 2019, CASA.

[76]  Louis J. Durlofsky,et al.  A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems , 2019, J. Comput. Phys..

[77]  A. Mohan,et al.  Compressed Convolutional LSTM: An Efficient Deep Learning framework to Model High Fidelity 3D Turbulence , 2019, 1903.00033.

[78]  Petros Koumoutsakos,et al.  Data-assisted reduced-order modeling of extreme events in complex dynamical systems , 2018, PloS one.

[79]  Henry D. I. Abarbanel,et al.  Predicting the Future , 2013 .

[80]  W. Bastiaanssen,et al.  Local calibration of remotely sensed rainfall from the TRMM satellite for different periods and spatial scales in the Indus Basin , 2012 .

[81]  P. Houtekamer,et al.  A Sequential Ensemble Kalman Filter for Atmospheric Data Assimilation , 2001 .

[82]  C. C. Pain,et al.  Data-driven modelling of nonlinear spatio-temporal fluid flows using a deep convolutional generative adversarial network , 2020, Computer Methods in Applied Mechanics and Engineering.

[83]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[84]  Traian Iliescu,et al.  Continuous data assimilation reduced order models of fluid flow , 2019, Computer Methods in Applied Mechanics and Engineering.

[85]  Henry D. I. Abarbanel,et al.  Estimating the state of a geophysical system with sparse observations: time delay methods to achieve accurate initial states for prediction , 2016 .

[86]  John Derber,et al.  A Global Oceanic Data Assimilation System , 1989 .

[87]  B. Hunt,et al.  A comparative study of 4D-VAR and a 4D Ensemble Kalman Filter: perfect model simulations with Lorenz-96 , 2007 .

[88]  Prasanna Balaprakash,et al.  Reduced-order modeling of advection-dominated systems with recurrent neural networks and convolutional autoencoders , 2020, Physics of Fluids.

[89]  George E. Karniadakis,et al.  Hidden physics models: Machine learning of nonlinear partial differential equations , 2017, J. Comput. Phys..

[90]  Raluca Radu,et al.  Spectral nudging in a spectral regional climate model , 2008 .

[91]  Hans von Storch,et al.  Dynamical downscaling: Assessment of model system dependent retained and added variability for two different regional climate models , 2008 .

[92]  J. C. Quinn,et al.  The Number of Required Observations in Data Assimilation for a Shallow-Water Flow , 2013 .

[93]  Richard A. Anthes,et al.  Data Assimilation and Initialization of Hurricane Prediction Models , 1974 .

[94]  A. Stuart,et al.  Sampling the posterior: An approach to non-Gaussian data assimilation , 2007 .

[95]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[96]  David R. Stauffer,et al.  Use of Four-Dimensional Data Assimilation in a Limited-Area Mesoscale Model Part II: Effects of Data Assimilation within the Planetary Boundary Layer , 1991 .

[97]  Yike Guo,et al.  Model error correction in data assimilation by integrating neural networks , 2019, Big Data Min. Anal..

[98]  Jaideep Pathak,et al.  A Machine Learning‐Based Global Atmospheric Forecast Model , 2020 .

[99]  John D. Horel,et al.  Sensitivity of a Spectrally Filtered and Nudged Limited-Area Model to Outer Model Options , 1996 .

[100]  Marc Bocquet,et al.  Data assimilation as a learning tool to infer ordinary differential equation representations of dynamical models , 2019, Nonlinear Processes in Geophysics.

[101]  T. Petroliagis,et al.  Error growth and estimates of predictability from the ECMWF forecasting system , 1995 .

[102]  Weixuan Li,et al.  Trimmed Ensemble Kalman Filter for Nonlinear and Non-Gaussian Data Assimilation Problems , 2018, 1808.05465.