[M-GI24] Data assimilation: A fundamental approach in geosciences

[MGI24-P04] Preliminary result on implementing flow-dependent background error covariances in JMA Meso-scale analysis

*Akane Saya¹, Yasutaka Ikuta¹, Takuya Kawabata¹ (1.Meteorological Research Institute)

Keywords:Hybrid 4DVar, time lagged ensemble, Meso-scale analysis, background error covariance

Improvement of weather numerical prediction (NWP) system has been required to forecast extreme events more accurately. To tackle this issue, data assimilation systems have been developed and applied to provide more accurate initial conditions. At JMA, Meso-scale model (MSM) is operated to assist the provision of disaster prevention information. The initial condition of MSM is created by Meso-scale analysis using a four-dimensional variational data assimilation system (4DVar).
However, since its background error covariance (BEC) is not flow-dependent, there is a difficulty on predicting extreme weather events. In addition, development of adjoint model requires quite high development costs. As a preliminary report, this study focuses on the performance of hybrid 4DVar system using ensemble BEC. The results from the hybrid 4DVar system and the traditional 4DVar system are compared. The aim of this study is to quantify the performance of the hybrid 4DVar and explore its characteristics. The test case is the heavy precipitation event occurred in Tokai region, Japan, September 2022. In this study, for flow-dependent BEC, 3 hourly time-lagged ensemble is used to create the BEC. We will show the comparison of this system with the traditional 3- and 4- D-Vars.