Begin Main Content

Presenter: Jeff Whitaker

Seminar Title: Results from a GFS-Based Ensemble Data Assimilation System for Reanalysis

An ensemble-based data assimilation system has been developed for the NCEP Global Forecast System model. The development has been a collaborative effort between researchers at NCEP, ESRL, the University of Maryland, NRL/Monterey and Colorado State. Although the system performs as well as the GSI for standard variables (such as 500 hPa height), and better for others (such as precipitable water), we have found that the benefits of flow-dependent background-error covariances really become apparent when the system is applied to historical reanalysis. Based on these results, NOAA is using this system to produce a 100+ year atmospheric reanalysis using only surface pressure data. Analyses have now been completed for 1908-1958, and the period 1892-present will be available for download in 2009. Statistics summarizing the accuracy of these analyses for upper-air fields will be presented, as well as high-resolution case studies for selected high-profile cases, such as the September 1938 New England Hurricane, the devastating 11/11/11 U.S. cold-air outbreak, and the infamous Tri-State tornado outbreak of March 1925. A novel aspect of this reanalysis system is that it provides estimates of analysis uncertainty, which are particularly important for long-term climate analyses. Some examples of the utility of these uncertainty estimates will also be presented.

return to Seminars