DDDAS: Data Dynamic Simulation for Disaster Management

From openwfm
Revision as of 19:25, 8 March 2010 by Erikj (talk | contribs)
Jump to navigation Jump to search

NSF Award No:0325314, 0324989, 0324988, 0324876, 0324910

Nuggett fig01.gif

Project Title:
ITR/NGS: Collaborative Research: DDDAS: Data Dynamic Simulation for Disaster Management

Investigators:

Jan Mandel, Anatolii Puhalski, Craig Johns, Leopoldo P. Franca, Craig C. Douglas, Janice L. Coen, Anthony Vodacek, Robert Kremens, Guan Qin

Institution:
University of Colorado at Denver, University of Kentucky, National Center for Atmospheric Research, Rochester Institute of Technology, Texas A&M University

Website:

math.ucdenver.edu/~jmandel/fires

Description of Graphic Image:
Simulated sparse measurements at the locations marked by triangles are assimilated into a highly nonlinear model of temperature at the front of an advancing fire. The red line is the truth and the green points are an ensemble of simulations. The standard Ensemble Kalman Filter (EnKF) method matches the data points well, but it does not approximate the truth away from data points (b). In several assimilation steps, this would result in a breakdown of the filter. A new stabilized method provides good match for the whole solution and a stable filtering process (c).

Project Description and Outcome

Ideas:The goal of this project is to provide a data driven real-time atmosphere-wildfire model with data acquired from weather data streams, sensors on location, and airborne images. The project is developing new data driven assimilation methods for highly nonlinear problems. The model consists of an ensemble of simulation. The data assimilation methods modify the model from data that arrives while the model is running.

Tools:A data driven massively parallel software framework was developed to link data assimilation algorithms, data acquisition, and an ensemble of simulations.


<tbody>
Nuggett fig02.jpg

Description of Graphic Image:
Left to right and top to bottom: The Reference solution represents the truth. Data assimilation by a standard ENKF algorithm results in an unstable solution because of the nonlinear behavior of wildfire. Stabilization gives the regularized solution ENKF+reg. Without data assimilation, the solution would develop as in the Comparison; the data assimilation shifts the model towards the truth. The model state is a probability distribution, visualized in the two ENKF figures as the superposition of transparent temperature profiles of ensemble members.