Michael Herrera
National Research Council at Naval Research Laboratory, Washington, DC, USA
Istvan Szunyogh;
Adam Brainard
Texas A&M University, College Station, TX, USA
Dave Kuhl;
Karl Hoppel
Naval Research Laboratory, Washington, DC, USA
Craig Bishop;
Teddy Holt;
Qingyun Zhao
Naval Research Laboratory, Monterey, California, USA
The 8th EnKF Data Assimilation Workshop
Regionally Enhanced Global (REG) Data Assimilation (DA)
Talk:
Herrera_Montreal_2018v2.pdf
A new data assimilation approach is presented which produces a global analysis by using forecast information provided by a global model and a limited area model in multiple limited area domains. The limited area forecast information is introduced into the data assimilation process by blending short-term forecasts from the global and regional model and using the blended forecast to compute the innovations, the differences between the observations and their predicted values. The approach is tested by an implementation on the US Navy's 4-dimensional variational global data assimilation system, and global and limited area numerical weather prediction models, however it is not dependent on the DA approach. Since REG DA acts as a black box, it can be easily implemented with other DA approaches such as a hybrid 4D-Var system or one of the EnKF implementations. The results of month-long global forecast experiments indicate statistically significant forecast improvements due to REG DA, with the largest improvements observed for Hurricane Sandy and frontal passages over the central plains of North America.
A new data assimilation approach is presented which produces a global analysis by using forecast information provided by a global model and a limited area model in multiple limited area domains. The limited area forecast information is introduced into the data assimilation process by blending short-term forecasts from the global and regional model and using the blended forecast to compute the innovations, the differences between the observations and their predicted values. The approach is tested by an implementation on the US Navy's 4-dimensional variational global data assimilation system, and global and limited area numerical weather prediction models, however it is not dependent on the DA approach. Since REG DA acts as a black box, it can be easily implemented with other DA approaches such as a hybrid 4D-Var system or one of the EnKF implementations. The results of month-long global forecast experiments indicate statistically significant forecast improvements due to REG DA, with the largest improvements observed for Hurricane Sandy and frontal passages over the central plains of North America.
Thanks to McGill's Atmospheric and Oceanic Sciences department for hosting this web-site.