Sho Yokota;
Hiromu Seko;
Masaru Kunii;
Hiroshi Yamauchi;
Eiichi Sato
Meteorological Research Institute, Japan Meteorological Agency
The 8th EnKF Data Assimilation Workshop
State-dependent Additive Covariance Inflation for Radar Reflectivity Assimilation
Talk:
20180508_Yokota_z_ptb.pptx
Direct assimilation of radar reflectivity can correct not only the hydrometeor but also the atmospheric state (e.g., wind, temperature, water vapor) based on the forecast error covariance including inter-variable correlation. This correlation can be given climatologically in three-dimensional variational method (3D-Var) or calculated by the linear model in four-dimensional variational method (4D-Var). However, reasonable estimation of the correlation has not been sufficiently accomplished. In ensemble Kalman filter (EnKF), the correlation can be calculated by ensemble forecasts without the large cost of development. However, the accurate calculation is limited to only the case that the hydrometeor exists at each analysis point in the first-guess fields of multiple ensemble members.
To assimilate radar reflectivity based on the more reasonable inter-variable correlation, we propose a new method for adding ensemble perturbations of radar reflectivity produced from the atmospheric state in points where the hydrometeor does not exist in the first-guess fields before assimilation with EnKF. This may be regarded as a kind of the additive inflation (Mitchell and Houtekamer 2000). However, the added perturbations are not given by random sampling but are created based on perturbations of wind, temperature, and water vapor at each analysis point multiplied by their sensitivities to radar reflectivity calculated in the whole computational domain including the rainfall region. Such perturbations are expected to have reasonable correlation with the atmospheric state.
To confirm the advantage of this state-dependent additive inflation, we conducted assimilation experiments of the reflectivity of the Meteorological Research Institute advanced C-band solid-state polarimetric radar with the 50-member local ensemble transform Kalman filter (LETKF) for the case of the tornadic supercell on 6 May 2012. As a consequence of the comparison between experiments with and without the state-dependent additive inflation, the assimilation with this inflation improved short-term rainfall forecasts through modifying wind and water vapor.
Direct assimilation of radar reflectivity can correct not only the hydrometeor but also the atmospheric state (e.g., wind, temperature, water vapor) based on the forecast error covariance including inter-variable correlation. This correlation can be given climatologically in three-dimensional variational method (3D-Var) or calculated by the linear model in four-dimensional variational method (4D-Var). However, reasonable estimation of the correlation has not been sufficiently accomplished. In ensemble Kalman filter (EnKF), the correlation can be calculated by ensemble forecasts without the large cost of development. However, the accurate calculation is limited to only the case that the hydrometeor exists at each analysis point in the first-guess fields of multiple ensemble members.
To assimilate radar reflectivity based on the more reasonable inter-variable correlation, we propose a new method for adding ensemble perturbations of radar reflectivity produced from the atmospheric state in points where the hydrometeor does not exist in the first-guess fields before assimilation with EnKF. This may be regarded as a kind of the additive inflation (Mitchell and Houtekamer 2000). However, the added perturbations are not given by random sampling but are created based on perturbations of wind, temperature, and water vapor at each analysis point multiplied by their sensitivities to radar reflectivity calculated in the whole computational domain including the rainfall region. Such perturbations are expected to have reasonable correlation with the atmospheric state.
To confirm the advantage of this state-dependent additive inflation, we conducted assimilation experiments of the reflectivity of the Meteorological Research Institute advanced C-band solid-state polarimetric radar with the 50-member local ensemble transform Kalman filter (LETKF) for the case of the tornadic supercell on 6 May 2012. As a consequence of the comparison between experiments with and without the state-dependent additive inflation, the assimilation with this inflation improved short-term rainfall forecasts through modifying wind and water vapor.
Thanks to McGill's Atmospheric and Oceanic Sciences department for hosting this web-site.