Jan Mandel/Blog/2010 Dec 2011 Jan

From CCM
Jump to: navigation, search

To the current blog page and the archive

<previous next>

These are mostly notes from our regular meetings. See also

Contents

December 6, 2010: Data Assimilation Seminar

Exascale computing

IMA Workshop last week Numerical Solutions of Partial Differential Equations: Fast Solution Techniques in IMA Thematic Year on Simulating Our Complex World: Modeling, Computation and Analysis, video:

Further inks:

Cloud computing

December 15-16: Wildfires session at AGU Fall meeting

Our presentations

Cell simulations of long-term forest renewal and interaction with wildfires

  • NH34A-01 A Conceptual Framework for Fire Ecology in a Changing Climate: Z. Gedalof
  • NH34A-03 Stand-replacing patches within a ‘mixed severity’ fire regime: quantitative characterization using recent fires in a long- established natural fire area: B. Collins, S. Stephens
  • NH33B-02 A forest-fire model with natural fire resistance Yoder, M R, Turcotte, D L, Rundle, J B, Glasscoe, M T, Donnellan, A

Combustion model and WFDS

December 18: Facebook servers

December 22: Wavelets

December 23: WRF-Fire

December 25: Spectral EnKF

December 30: EU wildfire projects

I have compiled a list of links to various EU projects sites. So far, I could not find any publicly accessible data, software, or manuals. Some have password-protected download links, others have only an email contact, which I did not pursue.

Fire Paradox

EU Fire Lab

MesoNH/ForeFire

Mesoscale atmosphere model Meso-nh, coupled with a tracer-based fire model.

January 24: Data assimilation seminar: Dry run of the AMS presentations

January 25: AMS annual meeting

Our papers

Computational intelligence methods and their applications to environmental science

Assimilation of observations into models: Advanced methods

  • 3:30 PM J12.1 TOWARD THE ASSIMILATION OF IMAGES, Francois-Xavier Le Dimet, INRIA, Grenoble, France
    • Satellite images now used qualitatively not quantitative way, very large data set cannot be directly use operationally, information not data, atmosphere "integral" of radiative properties of the atmosphere, information is borne by discontinuities; experiment: rotating tank 14m "Coriolis rotating platform" 1. pseudobservation - estimate velocities from images use as regular observation; 2 direct assimilation of images; temporal coherency of a sequence of images by law of conservation of brightness (seems much like what is done for radar images), solving conservation law by variational minimization plus Tichonov regularization; multiscale field optimization; image approach: add equation to make physical sense; Curvelet transform; extract structure by applying threshold to gradient of concentration.
  • 3:45 PM J12.2 On the Gaussian approach to adaptive covariance inflation, Takemasa Miyoshi, University of Maryland, College Park, MD
    • multiplicative inflation, additive, Anderson's Bayesian approach (2007, 2009), estimates for the inflation from sampling error estimate; localization - at each gridpoint separately; adaptive inflation accounts for model errors and limited ensemble size, 100%+ occasionally OK in limited region. Fixed inflation: spread too small over the denssely observed areas http://code.google.com/p/miyoshi
  • 4:00 PM J12.3, Localization and correlation in ensemble Kalman filters, Jeffrey Anderson, NCAR, Boulder, CO
    • impact of a single observation, regress increments onto each component of the state; localization is to find the multiplication of each regression coefficient; here, as a methodology to correct sampling error; need additional assumption on the prior; natural assumption about correlation of state and observation; minimize the multiplicative coefficient to minimize the RMSE; gives localization as a function of ensemble size N and sample correlation. Compares RMSE. "Spread is closer to the RMS error but still defficient". Ensembles size 20.
  • 4:15 PM J12.4 A kernel-density based ensemble filter applicable to high-dimensional systems, Thomas M. Hamill, NOAA / ESRL, Boulder, CO; and J. S. Whitaker
    • ref Anderson 2010 MWR pdf when prior is obviously nongaussian; cycled enkf can create nongaussian states all collapse except one; also for large ensembles Lawson and Hansen MWF 2003; from Anderson 2003: split update in 2 steps, compute increment then regress into state; "observation prior update"; more expensive, only when nongaussian by Kolmogorov-Smirnov test; nongaussianity detected only rarely; overall perturbation is superposition of old decaying structures and new ones, randomizes perturbation structures
  • 4:30 PM J12.5 What should an “outer loop” for ensemble data assimilation look like? Craig H. Bishop, NRL, Monterey, CA; and D. Hodyss
    • theoretically justifiable ensemble outer look approximates true nongaussian but less particles; fixed-truth error distribution; likelihood density; distribution of forecasts given guesses of truth rather than distribution of truth; would need K*64 members; jla: "outer loop" is a concept related 4DVAR
  • 4:45 PM J12.6 Bias correction using analysis increments within an ensemble Kalman filter data assimilation, Ji-Sun Kang, University of Maryland, College Park, MD; and E. Kalnay and T. Miyoshi
    • first perform analysis without bias correction, average forecast-analysis for every month - we have the estimated bias as a monthly mean, now perform another analysis with the bias correction; same with bias correction forecast-truth (or was it analysis-truth?) works better
  • 5:00 PM J12.7 Spectral and morphing ensemble Kalman filters, Jan Mandel, University of Colorado, Denver, CO; and J. D. Beezley and L. Cobb

Predictability

January 30: WRF-Fire paper

Finishing the WRF-Fire model paper for GMD

January 31: Data assimilation seminar

Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox