The transition process from Research to Operations in the Hurricane Forecast Improvement Project (HFIP)

The Hurricane Forecast Improvement Program (HFIP) has goals to reduce the error in track and intensity forecast guidance from numerical model systems by 20% in five years (by 2014) and 50% in 10 years. Additional goals include skillful 7 day forecasts and a greatly increased ability to forecast rapid intensification and decay. To meet these goals by the deadlines the HFIP has organized a large component of the hurricane communityto focus on various aspects of development of the numerical and statistical model forecast guidance systems expected to lead to these improvements. This includes development of advanced data assimilation systems and ensemble systems at high resolution for both regional and global models followed by statistical post processing. The transition to operations process includes several components: a real-time demonstration system run from August through October each hurricane season where various new technology is tested and demonstrated, a quasi-operational part of the real-time system where some of the components are presented to operational forecasters as part of their forecast process, and where various teams are assigned to specifically bring to operations technology that has been proven in previous years. These teams include members from across HFIP and represent both research and operational organizations. There are on the order of 60 scientists from both research and operations involved in the program, many funded directly by HFIP. All members of the project are involved each year in the budgeting process.

In addition to organizing the community to focus on short term development and transitioning that development to operations, the project has developed its own super computer system in Boulder where the real-time system is run in the summer and development for HFIP is conducted during the rest of the year. Currently this computer consists of 23,000 cores spread over three different machines.

This talk will outline how HFIP is organized to accomplish rapid development and transition to operations and note some recent successful examples.