A data-enabled wildfire smoke model

Donna Calhoun, Jodi Mead, Patricia Azike, Sandra Babyale

Date :

3rd Biennial Meeting of the SIAM Pacific Northwest Section, Society for Industrial and Applied Mathematics, Vancouver, WA (USA) (Mini-symposium speaker)
May 20 - May 22, 2022


We report on progress towards a data enabled wildfire smoke model. The model equations are conservative transport equations for smoke, using background wind data obtained from reanalysis data sources. The forward model will be implemented in ForestClaw (D. Calhoun and C. Burstedde), an parallel, finite volume solver for an adaptively refined mesh hierarchy of logically Cartesian meshes. The numerical scheme, based on the f-wave variation of the wave propagation algorithm (R. J. LeVeque, Univ. of Washington) is the underlying solver for logically Cartesian grid patches. Data assimilation will be carried out using the representer method (A. F. Bennett, 1992). In this approach, the model itself is a weak constraint, and a generalized inversion is then a best fit to both the dynamics and the data.

Slides : siam-pnw-2022.pdf

Watch : Video

Conference/Talk site