Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications

Description

This book contains the most recent progress in data assimilation in meteorology, oceanography and hydrology including land surface. It spans both theoretical and applicative aspects with various methodologies such as variational, Kalman filter, ensemble, Monte Carlo and artificial intelligence methods. Besides data assimilation, other important topics are also covered including adaptive observations, sensitivity analysis, parameter estimation and AI applications. The book is useful to individual researchers as well as graduate students for a reference in the field of data assimilation.  Preface Contents Contributors Data Assimilation for Chaotic Dynamics 1 Introduction 2 Chaos in Atmospheric and Oceanic Flows 2.1 Measuring Sensitivity to Initial Conditions 3 Data Assimilation in Chaotic Systems—How the Dynamics Impacts the Way We Assimilate Data 3.1 Linear Dynamics: The Effect of Chaos on the Kalman Filter and Smoother 3.2 Nonlinear Dynamics: The Effect of Chaos on the Ensemble Kalman Filter 4 Data Assimilation for Chaotic Systems—How Chaos Becomes an Opportunity 4.1 Targeting Observations Using the Unstable Subspace 4.2 Assimilation in the Unstable Subspace 5 Forward Looking 5.1 AUS in a Non-Gaussian Filter? 5.2 Data Assimilation and Random Attractors 6 Summary and Conclusion References Multifidelity Data Assimilation for Physical Systems 1 Introduction 1.1 Notation 1.2 The Data Assimilation Problem 1.3 Multifidelity Models 2 Control Variates 2.1 Ensemble Control Variates 3 Multifidelity Filtering 3.1 Multifidelity Kalman Filter 3.2 Multifidelity Ensemble Kalman Filter 3.3 Other `Multi-x' Data Assimilation Algorithms 4 Multifidelity Observations 5 Numerical Experiments 6 Discussion References Filtering with One-Step-Ahead Smoothing for Efficient Data Assimilation 1 Introduction 2 Problem Formulation 3 One-Step-Ahead Smoothing (OSAS) Formulation of Bayesian Filtering 3.1 The Generic Algorithm 3.2 State-Space Transform 4 OSAS-Like Filtering for Small-Dimensional Systems 4.1 The OSAS-Based Kalman Filter (KF-OSAS) 4.2 The OSAS-Based Particle Filter (PF-OSAS) 5 OSAS-Like Filtering for Large-Dimensional Systems 5.1 The OSAS-Based Ensemble Kalman Filter (EnKF-OSAS) 5.2 State-Parameters Estimation with OSAS-Based Ensemble Filtering 6 Summary References Sparsity-Based Kalman Filters for Data Assimilation 1 Introduction 2 The Sparsity of Error Covariance 3 Sparse-UKF 3.1 Sparse Matrix Algebra 3.2 UKF 3.3 Sparse-UKF 3.4 Lorenz-96 Model 4 Progressive-EKF 4.1 Basic Ideas 4.2 Progressive-EKF 4.3 Examples 5 Conclusions References Perturbations by the Ensemble Transform 1 Introduction 2 Ensemble Perturbations and Ensemble Transform 3 Perturbations in LETKF in NWP Models 3.1 Cases of SPEEDY-LETKF 3.2 Case of NHM-LETKF 4 Perturbations by Ensemble Transform in a Cloud Resolving Model 4.1 2 km NHM-LETKF 4.2 Cycle Experiments and Verification 5 Summary and Concluding Remarks References Stochastic Representations for Model Uncertainty in the Ensemble Data Assimilation System 1 Introduction 2 Methodology 2.1 Local Ensemble Transform Kalman Filter (LETKF) 2.2 Numerical Weather Prediction (NWP) Model 2.3 Stochastic Perturbation Hybrid Tendencies (SPHT) Scheme 3 Experimental Designs 4 Results 5 Summary References Second-Order Methods in Variational Data Assimilation 1 Introduction 2 Variational Data Assimilation 3 Computing the Hessian 4 Parameter Estimation 5 Sensitivity Analysis 6 Sensitivity with Respect to Observations 7 Application for a Sea Thermodynamics Model 8 Conclusions References Statistical Parameter Estimation for Observation Error Modelling: Application to Meteor Radars 1 Introduction 1.1 Definitions, Sources, and Characteristics 1.2 Error Correlation 1.3 Operational Treatment 1.4 Outlook 2 Diagnosing Observation Error Including Error of Representation 2.1 Innovation Based Estimation Methods 2.2 Ensemble Methods 2.3 Representation Error 2.4 Sensitivity Diagnostics 2.5 Other Methods 2.6 Current Operational Practice 2.7 Inter-Channel Radiance Assimilation 2.8 Spatial Correlations 3 Practical Application: Application to Meteor Radar Assimilation 3.1 Meteor Radar Observations 3.2 Assimilation System 3.3 A First Look at Error Estimates 3.4 Differences By Station 3.5 Experiments with Inflated Ensemble Variance and Inflated Observation Error Variance 3.6 Observation Impact 3.7 Root-Mean-Squared Error (RMSE) 3.8 Temporal Correlation 4 Discussion and Conclusions References Observability Gramian and Its Role in the Placement of Observations in Dynamic Data Assimilation 1 Introduction 2 Notations and Statement of Problem 2.1 Model 2.2 Observations 2.3 Innovation/Forecast Error 2.4 Cost Functional 2.5 Statement of Problem 3 Dynamics of Evolution of Forward Sensitivities 4 Relation Between Adjoint Sensitivity and Initial Control Error: Linear Case 5 Relation Between Adjoint Sensitivity and Initial Control Error: Non Linear Case 6 Air-Sea Interaction Example 7 Conclusions References Placement of Observations for Variational Data Assimilation: Application to Burgers’ Equation and Seiche Phenomenon 1 Introduction 2 Burgers’ Equation 3 Data Assimilation Experiment with Burgers’ Equation 4 Seiche Dynamics 4.1 Data Assimilation for Seiche 5 Conclusions References Analysis, Lateral Boundary, and Observation Impacts in a Limited Area Model 1 Forecast Sensivity to Observation Impact 2 COAMPS and NAVDAS 2.1 COAMPS Atmospheric Model 2.2 NAVDAS 2.3 COAMPS FSOI and Lateral Boundary Impacts 2.4 Forecast Domain 3 Forecast Error Reduction 4 Model Space Impacts 5 Observation Impacts 5.1 Radiosonde Verification 6 Summary References Assimilation of In-Situ Observations 1 Introduction 2 Radiosonde Observations 3 Surface Observations 4 Aircraft-Based Observations 5 Summary and Discussion Appendix 1: Definitions of Acronyms Appendix 2: Station Metadata Considerations References GNSS-RO Sounding in the Troposphere and Stratosphere 1 Fundamentals of the Radio Occultation Measurement 2 Typical Use of GNSS-RO in NWP 2.1 GNSS-RO Processing 3 Assimilation Methods and Error Statistic Assumptions 3.1 Forward Operators: H(x) 3.2 Error Statistic Assumptions 4 GNSS-RO Impact in NWP Systems 5 Future Directions for the Observation and Methods References Impact of Assimilating the Special Radiosonde Observations on COAMPS Arctic Forecasts During the Year of Polar Prediction 1 Introduction 2 Synoptic Features 3 Experimental Design 4 Discussion of Results 5 Summary and Conclusion References Images Assimilation: An Ocean Perspective 1 Introduction 2 Images Source and Processing: The Ocean Example 3 Methods for Image Assimilation and Their Limitations 3.1 Indirect Assimilation of Image 3.2 Direct Assimilation of Images 4 The Cost Function 5 Conclusion References Sensitivity Analysis in Ocean Acoustic Propagation 1 Introduction 2 The Model 3 Sensitivity Analysis 4 Numerical Experiments 5 Discussion and Summary Appendix: Equation of Sound Speed with Its Tangent Linear and Adjoint References Difficulty with Sea Surface Height Assimilation When Relying on an Unrepresentative Climatology 1 Introduction 2 Numerical Model: NCOM-4DVAR 2.1 Forward Model 2.2 Data Assimilation Configuration 3 Observations: SSHA and Glider Temperature and Salinity 3.1 SSHA Observations 3.2 Glider Data 3.3 Co-location of Assimilated Data 4 Experiment Setup and Results 4.1 Fit to Assimilated Data 4.2 Fit to Unassimilated Data 4.3 Direct Comparison of Experiment Results Against Glider Jade Temperature and Salinity 4.4 Comparison Against Recovered SSH 4.5 Comparison Against MODAS Profiles 5 Discussion 6 Summary References Theoretical and Practical Aspects of Strongly Coupled Aerosol-Atmosphere Data Assimilation 1 Introduction 1.1 Background on Coupled Data Assimilation System 1.2 Theoretical Description of Coupled Data Assimilation System 1.3 Single Observation Experiment 2 Current Status on Aerosol-Atmosphere Coupled Data Assimilation 2.1 Operational Centers and Research Community 2.2 Global Versus Regional Applications 3 Aerosol Observation and Forward Operator 3.1 Retrievals Versus Direct Measurements 3.2 AOD Observation Operator 3.3 AOD Error and Bias Estimation 4 Challenges 4.1 Choice of Control Variables 4.2 Background Error Covariance 4.3 Non-Gaussianity and Non-Linearity 4.4 Insufficient Data for Independent Verification 5 Experiments and Results 5.1 Case Study 5.2 Overview of the RAMS-MLEF System 5.3 Application of the RAMS-MLEF System 5.4 Synthetic Geostationary Satellite Imagery. 5.5 Model Response to Adjustments from Data Assimilation. 6 Summary and Future Directions References Improving Near-Surface Weather Forecasts with Strongly Coupled Land–Atmosphere Data Assimilation 1 Introduction 2 The Relationship Between Soil Moisture and Near-Surface Atmospheric Conditions 3 Strongly Coupled Versus Weakly Coupled Land–Atmosphere Data Assimilation 3.1 Characteristics of Background Error Covariance of Soil Moisture and Atmospheric States in Strongly Coupled Land–Atmosphere Data Assimilation 3.2 Soil Moisture Data Assimilation: Weakly Versus Strongly Coupled Data Assimilation 4 Enhanced Near-Surface Weather Forecasts Using Strongly Coupled Land–Atmosphere Data Assimilation 5 Discussion and Concluding Remarks 5.1 Summary and Discussion 5.2 Concluding Remarks References Ensemble Kalman Filter Experiments at 112-km and 28-km Resolution for the Record-Breaking Rainfall Event in Japan in July 2018 1 Introduction 2 NICAM-LETKF System and Experimental Settings 3 Results 3.1 Analysis 3.2 Ensemble Forecast 3.3 Downscaled Forecast 4 Summary and Concluding Remarks References Convective-Scale Data Assimilation and Precipitation Prediction with a Local Ensemble Transform Kalman Filter Radar Assimilation System Over Complex Terrain: A Thorough Investigation with the Heavy Rainfall in Taiwan on 16 June 2008 1 Introduction 2 WRF-Local Ensemble Transform Kalman Filter Radar Assimilation System (WLRAS) 3 Impact of Assimilating the Radar Data on Short-Term Precipitation Prediction 4 Limitations in WLRAS 5 Improving Heavy Rainfall Prediction in Taiwan with WLRAS 6 Summary and Future Work References Interpretation of Forecast Sensitivity Observation Impact in Data Denial Experiments 1 Introduction 2 Experiment Design 3 Northern Hemisphere Summer Forecast Verification Results 4 Northern Hemisphere Summer Forecast Sensitivity Observation Impact 5 Northern Hemisphere Summer 24-H Moist Total Energy Error Norms 6 Summary and Discussion Appendix: Definitions of Acronyms References Modelling the Background Error Covariance Matrix: Applicability Over the Maritime Continent 1 Introduction 2 Estimation of the Background Errors 2.1 Innovations Method 2.2 ‘National Meteorological Center’ (NMC) Method 2.3 Ensemble Methods 3 Modelling the Background Error Covariance Matrix 3.1 Homogeneity 3.2 Isotropy 3.3 Time Stationarity 3.4 Multivariate Balance Relationships 4 Comparison of Methods and Validity of Assumptions 4.1 Modelled Background Error Covariance Matrix Using the “Lagged” NMC Method 4.2 Raw Background Error Covariance Matrix Using the “Lagged” NMC Method 4.3 Raw Background Error Covariance Matrix Using the Ensemble Method 4.4 Modelled Background Error Covariance Matrix Using the Ensemble Method 5 Concluding Remarks References Operational Assimilation of Radar Data from the European EUMETNET Programme OPERA in the Météo-France Convective-Scale Model AROME 1 Introduction 2 OPERA Radar Data 2.1 Description of the Data 2.2 Pre-processing of the Data 2.3 Monitoring 3 Assimilation in AROME-France 3.1 AROME-France Assimilation System 3.2 Performances of the Assimilation in the Observation Space 3.3 Mean Scores 3.4 Case Study 4 Conclusion References The 2020 Global Operational NWP Data Assimilation System at Météo-France 1 Introduction 2 Main Features of the Global NWP Model 3 Description of the Data Assimilation System 3.1 High Resolution 4D-Var System 3.2 Ensemble Data Assimilation System 3.3 The Observation Usage 4 Behavior of the Current Operational Configuration 5 Conclusions and Planned Evolutions References An Overview of KMA’s Operational NWP Data Assimilation Systems 1 Introduction 2 UM Data Assimilation 2.1 History 2.2 Data Assimilation for the Global Model (GDAPS-UM) 2.3 Data Assimilation in Local Models 2.4 Ensemble Prediction Systems (EPS) 2.5 Observational Data Used in the Assimilation 2.6 Hybrid-4DVAR and VarBC 3 KIM Data Assimilation 3.1 History 3.2 Data Assimilation for the Global Model (GDAPS-KIM) 4 KLAPS (Very Short Range Forecast Model and Data Assimilation) 5 Conclusion References Index