Jan Mandel/Blog/2011 Feb Mar

From CCM
Jump to: navigation, search

To the current blog page and the archive

<previous next>


February 2: Graduate chat: A Researcher’s Perspective on Proposals and Grants

February 5: WRF-Fire paper for GMD

Jan Mandel, Jonathan D. Beezley, and Adam K. Kochanski, Coupled atmosphere-wildland fire modeling with WRF-Fire, UCD CCM Report 295, February 2011. Geoscientific Model Development (GMD), submitted. Also available as arXiv:1102.1343.

February 7: Data assimilation seminar: IEEE Prague paper

  • Jon leads
  • For 6th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS'2011), September 15-17, 2011, Prague, Czech Republic
    • Abstract submission Feb 27, 2011, From Author Instructions: Prospective authors are invited to submit an extended abstract of 3-4 pages (in English only). The manuscript should define the scope of the paper, emphasize on new advances, theories and/or applications and include an analysis of results and findings. It must be structured such that the program committee will be able to understand the originality and the value of the work. It is not clear if the final camera-ready paper is going to be the same.
    • Camera ready paper submission Jun 15, 2011
    • Proceedings: IEEE
  • Reference: Jan Mandel, Jonathan D. Beezley, and Loren Cobb, Spectral and morphing ensemble Kalman filters, abstract presentation files draft paper directory (will not continue, this paper and the next one supersede the draft)

February 8: Jon's matlab osimorph classes

February 9: Revisit registration and morphing

February 14: CCM Colloquium: Olivier Desjardins

February 18: My colloquium at Applied Math, CU Boulder

February 21: Data assimilation seminar: Nonparametric and wavelet methods

  • Nonparametric statistics (Loren)
  • Wavelet estimation of covariance (Jon) photos

February 24: WRF-Fire: Fuel height

February 27: Wavelet EnKF paper submitted

February 28: Data assimilation seminar: Probability measures - Basic concepts and types of convergence of random variables

March 4: Abstract for ISF 2011

  • The 31st Annual International Symposium on Forecasting, June 26-29, 2011, Prague, The University of Economics
  • submitted abstract

March 5: Bedrich's optical flow bibliography search

Bedrich found some new references: Li-2010-RFM Brox-2004-HAO Faisal-2007-HAO Papin-2000-TCH Zhou-2001-TNM and a list of links at http://wwwcsif.cs.ucdavis.edu/~homerick/WeatherCloud%20Tracking.html

March 7: Data assimilation seminar: Loren Cobb: Ergodic theory

March 8: Proofs of the GMD paper

March 9: WRF-Fire: GMDD paper published

March 13: Finishing the Harmanli paper

March 14: Data assimilation seminar: Jon Beezley: Kalman filter in Hilbert space and white noise

March 16: WRF-Fire

March 20: WRF-Fire: merging improved heat flux code

Before at 10s: Fgrnhfx-master-hill.png After at 10s: Fgrnhfx-vktest2-hill.png

Before at 5min: Fgrnhfx-master-hill-5min.png After at 5min: Fgrnhfx-vktest2-hill-5min.png

fuel_left_method=1 Fuel frac method 1 50sec.png fuel_left_method=2 Fuel frac method 2 50sec.png

fixed: Fuel frac method 2 50sec 5f6709b1638308600b.png but flux has hotspots: Fgrnhfx method 2 50sec 5f6709b1638308600b.png

probably because method 2 is not monotonous in tign. Max difference in fuel fraction between method 1 and 2 cca 11%.

  • merged into master (fast forward), but do not use method 2 yet
  • Is method 1 is monotonous in tign? It would be enough to show that tign interpolation is monotonous.
  • Next: Could there be spurious jumps in flux if the (tign,lfn) -> flux mapping is not smooth?
  • This all needs to be resolved before implementing canopy fire, because spurious heat flux hotspots will cause spurious canopy ignition.
  • More problems in fuel calculation found: fuel_left was 0.25 on domain boudary after first time step, causing heat released from 75% of the fuel in the first time step along domain boundary. Reason: should be divided to average only where computed. fuel_left can increase due to approximation errors, causing large negative heat flux spikes. Also, fire_area not initialized on boundary when computed, relying on default initialization in the caller instead. Fixed Sun, 27 Mar 2011. Compare with the "After at 10s" above and note different scale, 105 not 106.

At 10s: Fgrnhfx-hill-10s-Mar27-2011.png and in the same color scale: Fgrnhfx-hill-10s-Mar27-2011-caxis.png

March 22: "WRF-Fire Applied in Bulgaria" paper appeared

  • Nina Dobrinkova, Georgi Jordanov, and Jan Mandel, WRF-Fire Applied in Bulgaria, Numerical Methods and Applications, I. Dimov, S. Dimova, and N. Kolkovska, eds., Lecture Notes in Computer Science 6046, Springer, Berlin/Heidelberg, 2011, pp. 133-140. Proceedings of 7th International Conference on Numerical Methods and Applications - NM&A'10, August 20 - 24, 2010, Borovets, Bulgaria. doi:10.1007/978-3-642-18466-6_15. Preprint arXiv:1007.5347, July 2010.

March 24: WRF-Fire: Perimeter ignition

  • code
  • Input ignition times over whole domain from file input_tign_g, replay fire until fire_perimeter_time, atmosphere follows, then fire simulation takes over
  • The ignition times serve also as as the initial level set function
  • Test problem: cd WRFV3/test/em_fire/hill, see file README.txt

Prescribed ignition time: Tign prescribed.png Replay ends at 20 sec: Tign grnhfx 20s.png

Simulation to 40s: Tign grnhfx 40s.png Simulation to 300s: Tign grnhfx 300s.png

March 28: Data assimilation seminar: Bryan Smith: Statistics on Riemannian Manifolds

See also

Personal tools