Thursday, 31 August 2023
Boundary Waters (Hyatt Regency Minneapolis)
Radar is an important instrument of observation for improving the accuracy of short-range NWP model forecasts. Korea Institute of Atmospheric Prediction Systems (KIAPS) developed the global NWP system that was made operational at the Korea Meteorological Administration (KMA) in April 2020. The NWP model –named the Korean Integrated Model (KIM) - is a non-hydrostatic model based on a cubed-sphere grid, and the global data assimilation (DA) system is based on a hybrid-4DEnVar. Currently, the development of a high-resolution regional model based on is underway, we are developing a convection-permitting DA system for a radar observation assimilation.
The difference in time and space scale between radar observation and numerical models is very large. Most operational short-range NWP systems perform data assimilation with a background field of several km resolutions at an hourly analysis cycle, but radar observation data are produced at a spatial resolution of 250m every five minutes. Therefore, it is necessary to optimize the space and time scale of the observation data in order to minimiz,ㅡon error in the data assimilation.
In this study, we tried to optimize the spatial distribution of radar velocity observation by combining super-obbing and thinning methods. In the KIM data assimilation system, the thinning method is applied to most observations, but it is not suitable for high-density radar observation. Therefore, super-obbing was additionally applied to expand the space representation of observation. When applying super-obbing, a plan to consider the characteristics of radial velocity was applied. The value of super-obbing average was calculated in consideration of the direction to the observation site in consideration of the characteristics of the radial velocity. In addition, a method of maintaining a constant super-obbing grid size regardless of distance was applied around the radar observation site. After expanding the spatial scale of observation to an appropriate scale through super-obbing, thinning was additionally applied to efficiently utilize observation data in data assimilation. In the study, the super-obbing grid formation method and sensitivity experiments results in data assimilation will be introduced.
The difference in time and space scale between radar observation and numerical models is very large. Most operational short-range NWP systems perform data assimilation with a background field of several km resolutions at an hourly analysis cycle, but radar observation data are produced at a spatial resolution of 250m every five minutes. Therefore, it is necessary to optimize the space and time scale of the observation data in order to minimiz,ㅡon error in the data assimilation.
In this study, we tried to optimize the spatial distribution of radar velocity observation by combining super-obbing and thinning methods. In the KIM data assimilation system, the thinning method is applied to most observations, but it is not suitable for high-density radar observation. Therefore, super-obbing was additionally applied to expand the space representation of observation. When applying super-obbing, a plan to consider the characteristics of radial velocity was applied. The value of super-obbing average was calculated in consideration of the direction to the observation site in consideration of the characteristics of the radial velocity. In addition, a method of maintaining a constant super-obbing grid size regardless of distance was applied around the radar observation site. After expanding the spatial scale of observation to an appropriate scale through super-obbing, thinning was additionally applied to efficiently utilize observation data in data assimilation. In the study, the super-obbing grid formation method and sensitivity experiments results in data assimilation will be introduced.

