GEOS-CHEM adjoint development using TAF

Slides:



Advertisements
Similar presentations
Sensitivity Studies (1): Motivation Theoretical background. Sensitivity of the Lorenz model. Thomas Jung ECMWF, Reading, UK
Advertisements

EKF, UKF TexPoint fonts used in EMF.
Assimilation Algorithms: Tangent Linear and Adjoint models Yannick Trémolet ECMWF Data Assimilation Training Course March 2006.
A Discrete Adjoint-Based Approach for Optimization Problems on 3D Unstructured Meshes Dimitri J. Mavriplis Department of Mechanical Engineering University.
The Inverse Regional Ocean Modeling System:
Introduction to Adjoint Methods in Meteorology
Setup of 4D VAR inverse modelling system for atmospheric CH 4 using the TM5 adjoint model Peter Bergamaschi Climate Change Unit Institute for Environment.
Engineering Optimization – Concepts and Applications Engineering Optimization Concepts and Applications Fred van Keulen Matthijs Langelaar CLA H21.1
Aspects of Conditional Simulation and estimation of hydraulic conductivity in coastal aquifers" Luit Jan Slooten.
Lecture 21 Continuous Problems Fréchet Derivatives.
Stochastic Differentiation Lecture 3 Leonidas Sakalauskas Institute of Mathematics and Informatics Vilnius, Lithuania EURO Working Group on Continuous.
October, Scripps Institution of Oceanography An Alternative Method to Building Adjoints Julia Levin Rutgers University Andrew Bennett “Inverse Modeling.
Exploiting Satellite Observations of Tropospheric Trace Gases Ross N. Hoffman, Thomas Nehrkorn, Mark Cerniglia Atmospheric and Environmental Research,
Improving estimates of CO 2 fluxes through a CO-CO 2 adjoint inversion Monika Kopacz, Daniel J. Jacob, Parvadha Suntharalingam April 12, rd GEOS-Chem.
A Variational Carbon Data Assimilation System (CDAS-4DVar)
Advanced data assimilation methods with evolving forecast error covariance Four-dimensional variational analysis (4D-Var) Shu-Chih Yang (with EK)
Intercontinental Source Attribution of Ozone Pollution at Western U.S. Sites Using an Adjoint Method Lin Zhang, Daniel J. Jacob, Monika Kopacz Harvard.
Retrieval Theory Mar 23, 2008 Vijay Natraj. The Inverse Modeling Problem Optimize values of an ensemble of variables (state vector x ) using observations:
Advanced data assimilation methods- EKF and EnKF Hong Li and Eugenia Kalnay University of Maryland July 2006.
Lecture 10: Support Vector Machines
GEOS-Chem Adjoint: construction and long-term maintainability A. Sandu, K. Singh: Virginia Tech D. Henze: Caltech K. Bowman: JPL.
Constraining the N 2 O budget using isotopologues and isotopomers Y. L. Yung, S. P. Kim, M. C. Liang and R. L. Shia Divisions of Geological and Planetary.
Maximum Liklihood Ensemble Filter (MLEF) Dusanka Zupanski, Kevin Robert Gurney, Scott Denning, Milia Zupanski, Ravi Lokupitiya June, 2005 TransCom Meeting,
1 Using satellite data in joint CO 2 -CO inversion to improve CO 2 flux estimates Helen Wang 1,4, Daniel J. Jacob 1, Monika Kopacz 1, Dylan B.A. Jones.
Daven K. Henze with Amir Hakami and John H. Seinfeld Caltech, Chemical Engineering Support from: NSF, EPA, TeraGrid and JPL Supercomp., W. & S. Davidow.
Applications of Adjoint Methods for Aerodynamic Shape Optimization Arron Melvin Adviser: Luigi Martinelli Princeton University FAA/NASA Joint University.
A Brief Overview of Methods for Computing Derivatives Wenbin Yu Department of Mechanical & Aerospace Engineering Utah State University, Logan, UT.
Automatic Generation of Numerical Codes Jože Korelc University of Ljubljana, Slovenia
Intercomparison methods for satellite sensors: application to tropospheric ozone and CO measurements from Aura Daniel J. Jacob, Lin Zhang, Monika Kopacz.
Constraining the lightning-NO x (LiNOx) source using TES O 3 observations N. Bousserez, R.V. Martin, K.W. Bowman, D.K. Henze 5th International GEOS–Chem.
Results from the Carbon Cycle Data Assimilation System (CCDAS) 3 FastOpt 4 2 Marko Scholze 1, Peter Rayner 2, Wolfgang Knorr 1 Heinrich Widmann 3, Thomas.
FastOpt A road map to an adjoint analysis of the summer 2007 low in Arctic sea-ice area‏ Frank Kauker, Thomas Kaminski, Michael Karcher, Ralf Giering,
Interfaces EO data with Atmospheric and Land Surface Model: Progress report Liang Feng, Paul Palmer.
Overview of Techniques for Deriving Emission Inventories from Satellite Observations Frascati, November 2009 Bas Mijling Ronald van der A.
Analysis of TraceP Observations Using a 4D-Var Technique
Exploiting observed CO:CO 2 correlations in Asian outflow to invert simultaneously for emissions of CO and CO 2 Observed correlations between trace gases.
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.
Development of an EnKF to estimate CO 2 fluxes from realistic distributions of X CO2 Liang Feng, Paul Palmer
INVERSE MODELING OF ATMOSPHERIC COMPOSITION DATA Daniel J. Jacob See my web site under “educational materials” for lectures on inverse modeling atmospheric.
Non-Linear Parameter Optimisation of a Terrestrial Biosphere Model Using Atmospheric CO 2 Observation - CCDAS Marko Scholze 1, Peter Rayner 2, Wolfgang.
Emissions of CO from Asia during TRACE-P Paul Palmer, Daniel Jacob, Dylan Jones, Colette Heald David Streets, Glen Sachse, Hanwant Singh
FastOpt Quantitative Design of Observational Networks M. Scholze, R. Giering, T. Kaminski, E. Koffi P. Rayner, and M. Voßbeck Future GHG observation WS,
INVERSE MODELING TECHNIQUES Daniel J. Jacob. GENERAL APPROACH FOR COMPLEX SYSTEM ANALYSIS Construct mathematical “forward” model describing system As.
1 Assimilation of EPS tropospheric ozone for air quality forecast B. Sportisse, M. Bocquet, V. Mallet, I. Herlin, JP. Berroir, H. Boisgontier ESA EUMETSAT.
Quality of model and Error Analysis in Variational Data Assimilation François-Xavier LE DIMET Victor SHUTYAEV Université Joseph Fourier+INRIA Projet IDOPT,
Error correlation between CO 2 and CO as a constraint for CO 2 flux inversion using satellite data from different instrument configurations Helen Wang.
FastOpt CAMELS A prototype Global Carbon Cycle Data Assimilation System (CCDAS) Wolfgang Knorr 1, Marko Scholze 2, Peter Rayner 3,Thomas Kaminski 4, Ralf.
Inverse Modeling of Surface Carbon Fluxes Please read Peters et al (2007) and Explore the CarbonTracker website.
RESULTS: CO constraints from an adjoint inversion REFERENCES Streets et al. [2003] JGR doi: /2003GB Heald et al. [2003a] JGR doi: /2002JD
Anders Nielsen Technical University of Denmark, DTU-Aqua Mark Maunder Inter-American Tropical Tuna Commission An Introduction.
1 Radiative Transfer Models and their Adjoints Paul van Delst.
Adjoint models: Theory ATM 569 Fovell Fall 2015 (See course notes, Chapter 15) 1.
École Doctorale des Sciences de l'Environnement d’ Î le-de-France Année Modélisation Numérique de l’Écoulement Atmosphérique et Assimilation.
Mesoscale Assimilation of Rain-Affected Observations Clark Amerault National Research Council Postdoctoral Associate - Naval Research Laboratory, Monterey,
Comparison of adjoint and analytical approaches for solving atmospheric chemistry inverse problems Monika Kopacz 1, Daniel J. Jacob 1, Daven Henze 2, Colette.
Estimating emissions from observed atmospheric concentrations: A primer on top-down inverse methods Daniel J. Jacob.
Potential of Observations from the Tropospheric Emission Spectrometer to Constrain Continental Sources of Carbon Monoxide D. B. A. Jones, P. I. Palmer,
Carbon Cycle Data Assimilation with a Variational Approach (“4-D Var”) David Baker CGD/TSS with Scott Doney, Dave Schimel, Britt Stephens, and Roger Dargaville.
The Carbon Cycle Data Assimilation System (CCDAS)
Gleb Panteleev (IARC) Max Yaremchuk, (NRL), Dmitri Nechaev (USM)
Adjoint modeling and applications
Spring AGU, Montreal May 20, 2004
Chapter 6 Calibration and Application Process
Monika Kopacz, Daniel Jacob, Jenny Fisher, Meghan Purdy
Assimilating Tropospheric Emission Spectrometer profiles in GEOS-Chem
HARVARD GROUP TRACE-P ACTIVITIES (a planned papers)
Constraints on Asian Carbon Fluxes using TRACE-P CO2/CO Correlations
GCHP adjoint tasks Model design
Verification and Validation Using Code-Based Sensitivity Techniques
Derivatives and Gradients
Presentation transcript:

GEOS-CHEM adjoint development using TAF Monika Kopacz, Daniel Jacob, Dylan Jones (UT), Parvadha Suntharalingam, Paul Palmer April 5, 2005

What is an adjoint model? X Y adjoint model Inverse map x y K K* For linear operator K, mapping from space X to space Y. Adjoint operator KT For nonlinear model K represents linearization of the original model

Why the need for adjoint model? constraints on emissions with high resolution can consider nonlinear processes computationally efficient for sensitivity studies

Adjoint of GEOS-CHEM 3 approaches to obtaining derivative code: analytic derive adjoint operator analytically and discretize discrete Lagrange function define L = J +lT (K x – yobs)  li = li+1 + (Ki+1 xi+1 – yobs,i+1) direct differentiation of the code: Automatic Differentiation Transformation of Algorithm in Fortran (TAF) source to source compiler Recipes for Adjoint Code Construction, Giering and Kaminski, 1998

What is TAF? Source to source compiler developed by Ralf Giering and Thomas Kaminski; a commercial, license-based product supplied by Fast Opt® Forward differentiation, Tangent Linear Model: TAF dynamic_loop.f dynamic_loop_tl.f Reverse differentiation, Adjoint Model: TAF dynamic_loop_ad.f dynamic_loop.f TAF dynamic_loop(x, fc) ad_dynamic_loop(x, adx, fc, adfc) where x can represent emissions field fc can represent a scalar ‘cost’ variable then adx stores dfc/dx adfc stores dfc/dfc = 1 http://fastopt.com

First step Tangent Linear Model challenges benefits Semi-automatically derived, stand-alone model Tangent Linear Model challenges benefits ? general TAF-GC interfacing ? resolving flow nonlinearities ? Fortran 90 support ? manual changes verification: comparison with green’s function approach Check linearity of the code Resolve TAF-GC compatibility

Next step Adjoint Model challenges benefits Semi-automatically derived, stand-alone model Adjoint Model challenges benefits ? all TLM Fortran issues ? storage/recomputation for reverse mode ? non-trivial model evaluation ? manual changes ? Interfacing with optimization package verification: comparison with previous CO inversions, (Heald et al. 2003, Palmer et al. 2003) efficient sensitivity studies J  inversions for large state vector and large data sets J  data assimilation

Minimizing the gradient 0 ° 2 1 3 x2 x1 x3 x0 Adjoint Transport GEOS- CHEM Measurement Sampling Estimated Fluxes Modeled Concentrations simulated Measurements “True” Assumed Errors Weighted Residuals Cost function J Flux Update a priori fluxes Adjoint Fluxes = J Minimum of cost function J Courtesy of David Baker

Preliminary results Finite Difference Tangent Linear Model 3D field Scalar perturbation

SEMI-AUTOMATIC CAPABILITY TO GENERATE TLM/ADM End product ADJOINT MODEL TANGENT LINEAR MODEL OPTIMIZATION PACKAGE SEMI-AUTOMATIC CAPABILITY TO GENERATE TLM/ADM

Automatic differentiation Z = X * SIN (Y**2) differential transpose ADY = ADY + ADZ*X *COS(Y**2)*2*Y ADX = ADX + ADZ*SIN(Y**2) ADZ = 0.0 from Recipes for Adjoint Code Construction by R. Giering and T. Kaminski, 1998

Analytic adjoint operator for advection Consider