An application of a weakly constrained 4DVAR to satellite data assimilation and heavy rainfall simulation

Abstract

In this paper a simple weakly constrained four-dimensional variational data assimilation (4DVAR) technique (W4DVAR) is used in the assimilation of retrieved satellite data for a heavy rainfall simulation. The W4DVAR and the strongly constrained 4DVAR (S4DVAR) were compared through the assimilation of retrieved satellite data. In the assimilation of the retrieved satellite data, the W4DVAR technique provided an optimal initial condition for the MM5 model in the simulation of a heavy rainfall event. The W4DVAR reduced both boundary errors during 4DVAR assimilation and the large retrieval errors of Television Infrared Observational Satellite Operational Vertical Sounder (TOVS) sounding data near the surface. It was found that TOVS sounding data were useful for the simulation of typhoons and mesoscale convective systems over the ocean. The satellite data assimilated by W4DVAR contributed to favorable conditions for a heavy rainfall event by providing increased and balanced water vapor transport in the lower troposphere. The improvement in precipitation prediction using satellite data and W4DVAR was attributed to the improved spinup of moist physics processes due to the fast production of cloud water near the initial time. The moisture and temperature fields generated by W4DVAR were a significant factor in producing optimal initial data for the heavy rainfall prediction. The simulated precipitation indicated that the incremental approach was also useful for providing the initial data for a high-resolution forecast model obtained from low-resolution 4DVAR results. It is also suggested that it would be useful to employ W4DVAR in the assimilation of retrieved asynoptic satellite data over data-void areas.