CIRA » Home » Research » Project » NOAA Ensemble Data Assimilation Research for Hurricane Forecasting
Project Title: Ensemble Data Assimilation Research for Hurricane Forecasting
Principal Investigators: Dr Milija Zupanski (CSU)
Postdoctoral Scientists: Man Zhang (CSU)
Sponsor: NOAA
Project duration: 1 July 2010 - 30 June 2014
Funded amount: $200,000
Goals and objectives:
1. Install an experimental EnKF system on the NOAA R&D computer
2. Run a sample of approximately 10 different storm cases cycled over several days
3. Perform diagnostics of the evolved covariances between ensemble members
4. Attempt to simulate these relationships with the SDBE recursive filter system, and
5. Compare forecasts of track and intensity from both the EnKF system and 3-D Var system with SDBEs.
Summary:
NCEP’s plans for regional data assimilation of the hurricane vortex and surrounding environment are to augment the current 3-D Var system with Situation-Dependent Background Errors (SDBEs). One of the critical unsolved problems for SDBEs is the balance between wind and thermodynamic increments to the background as dictated by the background error covariances. In the NCEP 3-D Var, anisotropic SDBEs are simulated through spatial recursive filters linked to the background atmospheric state; in this way, flow-dependency can be incorporated into the 3-D Var framework. However, there is currently limited knowledge and research needed about the best formulation of the balance constraints for increments as determined by error covariances.
This research problem can be approached by investigating the balances determined by ensemble-based data assimilation, specifically the current class of Ensemble Kalman Filter (EnKF) schemes. We will employ an EnKF algorithm developed at Colorado State University, named the Maximum Likelihood Ensemble Filter (MLEF), for improving the formulation of balance constraints in SBDEs.
