Ensemble based data assimilation at the mesoscale using large ensembles

GUO-YUAN LIEN; JUAN RUIZ; TAKEMASA MIYOSHI; KEIICHI KONDO

Reading

Simposio; International Symposium on Data Assimilation 2016; 2016

Large ensemble data assimilation experiments allows for a more accurate representation of the structure and temporal evolution of the background and analysis error covariance matrices. This provides useful information about the error dynamics in atmospheric systems at different spatio-temporal scales. Having an accurate representation of the error covariance matrices also help to improve current localization strategies in ensemble based data assimilation systems.In this work, large ensemble data assimilation experiments are performed using a mesoscale model at 1 km resolution. The SCALE-LETKF mesoscale data assimilating system is run using 1000 ensemble members and assimilating radar observations from the Osaka University phased array radar every 60 seconds. The 13th July 2013 heavy rain case over Kyoto is used to run the large ensemble experiment.The background error covariance matrix obtained from the large ensemble data assimilation experiment is analyzed in order to better understand its spatial structure and temporal evolution. The dependence of the error covariance patterns with the state variables is also analyzed. Results will be presented at the conference.