Dusanka Zupanski And Scott Denning Colorado State University Fort Collins, CO 80523-1375 CMDL Workshop on Modeling and Data Analysis of Atmospheric CO.

Slides:



Advertisements
Similar presentations
Data-Assimilation Research Centre
Advertisements

Introduction to Data Assimilation NCEO Data-assimilation training days 5-7 July 2010 Peter Jan van Leeuwen Data Assimilation Research Center (DARC) University.
Modeling of Data. Basic Bayes theorem Bayes theorem relates the conditional probabilities of two events A, and B: A might be a hypothesis and B might.
Introduction to Data Assimilation Peter Jan van Leeuwen IMAU.
Ensemble data assimilation (EnsDA) activities of the GOES-R project Progress report Dusanka Zupanski CIRA/CSU GOES-R meeting 8 September 2004 Dusanka Zupanski,
P2.1 ENSEMBLE DATA ASSIMILATION: EXPERIMENTS USING NASA’S GEOS COLUMN PRECIPITATION MODEL D. Zupanski 1, A. Y. Hou 2, S. Zhang 2, M. Zupanski 1, C. D.
Dusanka Zupanski CIRA/Colorado State University Fort Collins, Colorado Model Error and Parameter Estimation Joint NCAR/MMM CSU/CIRA Data Assimilation Workshop.
Critical issues of ensemble data assimilation in application to GOES-R risk reduction program D. Zupanski 1, M. Zupanski 1, M. DeMaria 2, and L. Grasso.
1B.17 ASSESSING THE IMPACT OF OBSERVATIONS AND MODEL ERRORS IN THE ENSEMBLE DATA ASSIMILATION FRAMEWORK D. Zupanski 1, A. Y. Hou 2, S. Zhang 2, M. Zupanski.
Dusanka Zupanski CIRA/Colorado State University Fort Collins, Colorado A General Ensemble-Based Approach to Data Assimilation Model Error and Parameter.
Dusanka Zupanski CIRA/Colorado State University Fort Collins, Colorado Ensemble Kalman Filter Guest Lecture at AT 753: Atmospheric Water Cycle 21 April.
P1.8 QUANTIFYING AND REDUCING UNCERTAINTY BY EMPLOYING MODEL ERROR ESTIMATION METHODS Dusanka Zupanski Cooperative Institute for Research in the Atmosphere.
Model error estimation employing ensemble data assimilation Dusanka Zupanski and Milija Zupanski CIRA/Colorado State University, Fort Collins, CO, U.S.A.
Dusanka Zupanski CIRA/Colorado State University Fort Collins, Colorado Ensemble Data Assimilation and Prediction: Applications to Environmental Science.
MODEL ERROR ESTIMATION Cooperative Institute for Research in the Atmosphere Research Benefits to NOAA:This is a novel research approach, providing an optimal.
Hou/JTST Exploring new pathways in precipitation assimilation Arthur Hou and Sara Zhang NASA Goddard Space Flight Center Symposium on the 50 th.
1 アンサンブルカルマンフィルターによ る大気海洋結合モデルへのデータ同化 On-line estimation of observation error covariance for ensemble-based filters Genta Ueno The Institute of Statistical.
Ibrahim Hoteit KAUST, CSIM, May 2010 Should we be using Data Assimilation to Combine Seismic Imaging and Reservoir Modeling? Earth Sciences and Engineering.
Parameter Estimation and Data Assimilation Techniques for Land Surface Modeling Qingyun Duan Lawrence Livermore National Laboratory Livermore, California.
Carbon Flux Bias Estimation at Regional Scale using Coupled MLEF-PCTM model Ravindra Lokupitiya Department of Atmospheric Science Colorado State University.
Advanced data assimilation methods- EKF and EnKF Hong Li and Eugenia Kalnay University of Maryland July 2006.
Ensemble Kalman Filter Methods
Introduction to Numerical Weather Prediction and Ensemble Weather Forecasting Tom Hamill NOAA-CIRES Climate Diagnostics Center Boulder, Colorado USA.
Maximum Liklihood Ensemble Filter (MLEF) Dusanka Zupanski, Kevin Robert Gurney, Scott Denning, Milia Zupanski, Ravi Lokupitiya June, 2005 TransCom Meeting,
A comparison of hybrid ensemble transform Kalman filter(ETKF)-3DVAR and ensemble square root filter (EnSRF) analysis schemes Xuguang Wang NOAA/ESRL/PSD,
Lecture II-2: Probability Review
Ensemble Data Assimilation and Uncertainty Quantification Jeffrey Anderson, Alicia Karspeck, Tim Hoar, Nancy Collins, Kevin Raeder, Steve Yeager National.
3D/4D-Var Methods Liang Xu (NRL) JCSDA Summer Colloquium on Satellite DA 1 3D-Var/4D-Var Solution Methods Liang Xu Naval Research Laboratory, Monterey,
1 ESTIMATING THE STATE OF LARGE SPATIOTEMPORALLY CHAOTIC SYSTEMS: WEATHER FORECASTING, ETC. Edward Ott University of Maryland Main Reference: E. OTT, B.
CSDA Conference, Limassol, 2005 University of Medicine and Pharmacy “Gr. T. Popa” Iasi Department of Mathematics and Informatics Gabriel Dimitriu University.
Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology MeteoSwiss High-resolution data assimilation in COSMO: Status and.
Assimilation of HF Radar Data into Coastal Wave Models NERC-funded PhD work also supervised by Clive W Anderson (University of Sheffield) Judith Wolf (Proudman.
1 GSI/ETKF Regional Hybrid Data Assimilation with MMM Hybrid Testbed Arthur P. Mizzi NCAR/MMM 2011 GSI Workshop June 29 – July 1, 2011.
DoD Center for Geosciences/Atmospheric Research at Colorado State University WSMR November 19-20, 2003 ARMY Research Lab and CIRA/CSU Collaboration on.
2004 SIAM Annual Meeting Minisymposium on Data Assimilation and Predictability for Atmospheric and Oceanographic Modeling July 15, 2004, Portland, Oregon.
A unifying framework for hybrid data-assimilation schemes Peter Jan van Leeuwen Data Assimilation Research Center (DARC) National Centre for Earth Observation.
ECE 8443 – Pattern Recognition LECTURE 10: HETEROSCEDASTIC LINEAR DISCRIMINANT ANALYSIS AND INDEPENDENT COMPONENT ANALYSIS Objectives: Generalization of.
Applications of optimal control and EnKF to Flow Simulation and Modeling Florida State University, February, 2005, Tallahassee, Florida The Maximum.
MODEL ERROR ESTIMATION EMPLOYING DATA ASSIMILATION METHODOLOGIES Dusanka Zupanski Cooperative Institute for Research in the Atmosphere Colorado State University.
MODEL ERROR ESTIMATION IN ENSEMBLE DATA ASSIMILATION FRAMEWORK Dusanka Zupanski Cooperative Institute for Research in the Atmosphere Colorado State University.
Maximum Likelihood Estimation and Simplified Kalman Filter tecniques for real time Data Assimilation.
Research Vignette: The TransCom3 Time-Dependent Global CO 2 Flux Inversion … and More David F. Baker NCAR 12 July 2007 David F. Baker NCAR 12 July 2007.
INVERSE MODELING OF ATMOSPHERIC COMPOSITION DATA Daniel J. Jacob See my web site under “educational materials” for lectures on inverse modeling atmospheric.
Data assimilation, short-term forecast, and forecasting error
# # # # An Application of Maximum Likelihood Ensemble Filter (MLEF) to Carbon Problems Ravindra Lokupitiya 1, Scott Denning 1, Dusanka Zupanski 2, Kevin.
Data assimilation and forecasting the weather (!) Eugenia Kalnay and many friends University of Maryland.
DATA ASSIMILATION AND MODEL ERROR ESTIMATION Dusanka Zupanski Cooperative Institute for Research in the Atmosphere Colorado State University Fort Collins,
INVERSE MODELING TECHNIQUES Daniel J. Jacob. GENERAL APPROACH FOR COMPLEX SYSTEM ANALYSIS Construct mathematical “forward” model describing system As.
Errors, Uncertainties in Data Assimilation François-Xavier LE DIMET Université Joseph Fourier+INRIA Projet IDOPT, Grenoble, France.
Quality of model and Error Analysis in Variational Data Assimilation François-Xavier LE DIMET Victor SHUTYAEV Université Joseph Fourier+INRIA Projet IDOPT,
NCAF Manchester July 2000 Graham Hesketh Information Engineering Group Rolls-Royce Strategic Research Centre.
Local Predictability of the Performance of an Ensemble Forecast System Liz Satterfield and Istvan Szunyogh Texas A&M University, College Station, TX Third.
Introducing Error Co-variances in the ARM Variational Analysis Minghua Zhang (Stony Brook University/SUNY) and Shaocheng Xie (Lawrence Livermore National.
WCRP Extremes Workshop Sept 2010 Detecting human influence on extreme daily temperature at regional scales Photo: F. Zwiers (Long-tailed Jaeger)
Prepared by Dusanka Zupanski and …… Maximum Likelihood Ensemble Filter: application to carbon problems.
, Karina Apodaca, and Man Zhang Warn-on-Forecast and High-Impact Weather Workshop, February 6-7, 2013, National Weather Center, Norman, OK Utility of GOES-R.
Ensemble data assimilation and model error estimation algorithm Developed by Milija Zupanski and Dusanka Zupanski CIRA/CSU Dusanka Zupanski, CIRA/CSU
A Random Subgrouping Scheme for Ensemble Kalman Filters Yun Liu Dept. of Atmospheric and Oceanic Science, University of Maryland Atmospheric and oceanic.
Geogg124: Data assimilation P. Lewis. What is Data Assimilation? Optimal merging of models and data Models Expression of current understanding about process.
Ensemble forecasting/data assimilation and model error estimation algorithm Prepared by Dusanka Zupanski and Milija Zupanski CIRA/CSU Denning group meeting.
École Doctorale des Sciences de l'Environnement d’Île-de-France Année Universitaire Modélisation Numérique de l’Écoulement Atmosphérique et Assimilation.
École Doctorale des Sciences de l'Environnement d’ Î le-de-France Année Modélisation Numérique de l’Écoulement Atmosphérique et Assimilation.
The Ensemble Kalman filter
Scale-dependent localization: test with quasi-geostrophic models
LOGNORMAL DATA ASSIMILATION: THEORY AND APPLICATIONS
Data Assimilation Theory CTCD Data Assimilation Workshop Nov 2005
Department of Civil and Environmental Engineering
Information content in ensemble data assimilation
Probabilistic Robotics
Sarah Dance DARC/University of Reading
Presentation transcript:

Dusanka Zupanski And Scott Denning Colorado State University Fort Collins, CO CMDL Workshop on Modeling and Data Analysis of Atmospheric CO 2 Observations in North America September 2004 ftp://ftp.cira.colostate.edu/Zupanski/presentations ftp://ftp.cira.colostate.edu/Zupanski/manuscripts Critical issues of ensemble data assimilation in application to carbon cycle studies

 Introduction: EnsDA approaches  Non-linear processes  Model error and parameter estimation  Uncertainty estimates  Correlated observations  Non-Gaussian PDFs  Conclusions and future work Dusanka Zupanski, CIRA/CSU OUTLINE:

Probabilistic approach to data assimilation and forecasting or Ensemble Data Assimilation (EnsDA) Dusanka Zupanski, CIRA/CSU Provides the following: (1) Optimal solution or state estimate (e. g., optimal CO 2 analysis) (2) Optimal estimates of model error and empirical parameters (3) Uncertainty of the analysis (a component of the analysis error covariance P a ) (4) Uncertainty of the estimated model error and parameters (components of the analysis error covariance P a ) (5) Estimate of forecast uncertainty (the forecast error covariance P f )

DATA ASSIMILATION (ESTIMATION THEORY) Discrete stochastic-dynamic model Dusanka Zupanski, CIRA/CSU Discrete stochastic observation model w k-1 – model error (stochastic forcing) M – non-linear dynamic (NWP) model G – model (matrix) reflecting the state dependence of model error  k – measurement + representativeness error M  D H– non-linear observation operator ( M  D )

(1) State estimate (optimal solution): (2) Estimate of the uncertainty of the solution: ENSEMBLE KALMAN FILTER or EnsDA APPROACH In EnsDA solution is defined in ensemble subspace (reduced rank problem) ! KALMAN FILTER APPROACH MAXIMUM LIKELIHOOD ESTIMATE (VARIATIONAL APPROACH ): MINIMUM VARIANCE ESTIMATE (KALMAN FILTER APPROACH ): DATA ASSIMILATION EQUATIONS:

Ensemble Data Assimilation (EnsDA) Dusanka Zupanski, CIRA/CSU (1) Maximum likelihood approach (involves an iterative minimization of a functional)  x mode (MLEF, Zupanski 2004) (2) Minimum variance approach (calculates ensemble mean)  x mean x mode x mean x PDF(x) x mode = x mean x PDF(x) Non-GaussianGaussian

Critical issues: Non-linear processes Dusanka Zupanski, CIRA/CSU - Use only non-linear models (tangent-linear, adjoint models are not needed) - Iterative minimization is beneficial for non-linear processes Example: KdVB model (M. Zupanski, 2004)

Critical issues: Model error and parameter estimation Dusanka Zupanski, CIRA/CSU - Estimate and correct all major sources of uncertainty: initial conditions, model error, boundary conditions, empirical parameters - Unified algorithm: EnsDA+state augmentation approach (Zupanski and Zupanski, 2004) Example: KdVB model

10 obs101 obs EnsDA experiments with KdVB model (PARAMETER estimation impact)

EnsDA experiments with KdVB model (BIAS estimation impact) Dusanka Zupanski, CIRA/CSU

Critical issues: Uncertainty estimates Dusanka Zupanski, CIRA/CSU - Analysis error covariance P a (analysis uncertainty) - Forecast error covariance P f (forecast uncertainty) - Both defined in ensemble sub-space KdVB model example:

Critical issues: Correlated observations Dusanka Zupanski, CIRA/CSU Problem: Numerous observations (~ ) are being projected onto a small ensemble sub-space (~ ) ! Loss of observed information! Remedies:  Process observations one by one (Anderson 2001, Bishop et al. 2001; Hamill et al. 2001). Or  Process observations successively over relatively small local areas (LEKF, Ott et al. 2004). Assumption in both approaches: Observations being processed separately are uncorrelated (independent)! This may not be justified for dense satellite observations.

Critical issues: Correlated observations Dusanka Zupanski, CIRA/CSU How does the observed information impact the uncertainty estimate of the optimal solution (analysis error covariance P a ) ? - square root of analysis error covariance (N state x N ens ) - square root of forecast error covariance (N state x N ens ) - impact of observations on the optimal solution (N ens x N ens ) The eigenvalue spectrum of (I+A) -1/2 may help understand the impact of observations, and perhaps find a better solution for correlated observations.

RAMS model example Dusanka Zupanski, CIRA/CSU A safe approach to prevent loss of observed information, assuming independent observations: N obs  N ens. If eigenvalues of (I+A) -1/2 spread over the entire interval [0,1], ensemble size (N ens ) is appropriate for a given observation number (N obs ).

CSU shallow-water model on geodesic grid When system can learn from its past, less information from observations is needed ! Smooth start (in cycle 1) can improve the performance of EnsDA Milija Zupanski, CIRA/CSU Analysis error smaller than obs error (Results from M. Zupanski et al.)

Non-Gaussian PDFs Non-linear Atmospheric- Hydrology- Carbon state variables and observations are likely to have non-Gaussian PDFs.  MLEF, as a maximum likelihood estimate, is a suitable tool for examining the impact of different PDFs.  Develop a non-Gaussian PDF framework (M. Zupanski) - allow for non-Gaussian observation errors - apply the Bayes theorem for multiple events Milija Zupanski, CIRA/CSU

CSU EnsDA algorithm is currently being examined in application to NASA’s GEOS column model in collaboration with: -A. Hou and S. Zhang (NASA/GMAO) -C. Kummerow (CSU/Atmos. Sci.) NASA’s GEOS column model example R 1/2 =  R 1/2 = 2  Prescribed observation errors directly impact innovation statistics. Since the observation error covariance R is the only input required by the system, it could be tuned! Dusanka Zupanski, CIRA/CSU In case we solved all critical issues, one problem remains: How to define observation error covariance matrix R, if it is not known?

 EnsDA approaches are very promising since they can provide not only optimal estimate of the state, but also the uncertainty of the optimal estimate.  The experience gained so far indicates that the EnsDA approach is suitable for addressing critical issues of data assimilation in Carbon cycle studies.  Model error and parameter estimation are necessary ingredients of a data assimilation algorithm.  Problems involved in Carbon data assimilation require a state-of- the art approach. We anticipate findings from different scientific disciplines (e. g., atmospheric science, ecology, hydrology) to be of mutual benefits.  It is especially important to gain experience with complex coupled models (e. g., RAMS-SiB-CASA), correlated (satellite) observations, and non-Gaussian PDFs in the future. Dusanka Zupanski, CIRA/CSU

Dusanka Zupanski, CIRA/CSU