Development of a cloud-resolving 4DVAR data assimilation system based on the JMA nonhydrostatic model

A cloud resolving 4-dimensional variational data assimilation system (4DVAR) based on the Japan Meteorological Agency nonhydrostatic model (JMA-NHM) is under development. One of the targets of this system is the analysis of mesoscale convective systems. Features of background error statistics for the model with a horizontal resolution of 2km (hereinafter abbreviated as 2km model) are much different from those with a 5km resolution (5km model). Thus, forecast error estimated by the scale-down method from that forecast error obtained from the 5km model was not applicable. To develop the cloud resolving system, background error statistics for the system with 2km horizontal intervals were calculated and a suitable set of control variables was designed. Using the new background error statistics and the new control variable set, a preliminary data assimilation experiment of the Global Positioning System (GPS)-derived precipitable water vapor (PWV) and radial wind observed by Doppler radars (RW) was performed. By assimilating GPS-PWV and RW, the convergence of horizontal wind was strengthened, and observed features of horizontal winds and PWV were reproduced in the analyzed field.