局地モデルの降水の立ち上がりを改善するためのモデル同化と初期値化法

Abstract

Listen

Translate article

Forecast experiments are performed from various initial states using an operational mesoscale model of JMA in order to improve the spin-up of precipitation forecasts. The initial conditions are designed to investigate the impact of FA cycle, diabatic NNMI and moisture initialization. The diabatic effects in the NNMI are treated in two ways; diabatic heating rates calculated by the model physical processes, and those estimated from observed rain rates with an assumed parabolic heating profile.The examined case includes a mesoscale convective cloud system. The operational forecast is characterized by very slow spin-up of precipitation and position errors after late build-up of precipitation. In the forecast experiments, use of the FA cycle was effective for improving the description of the mesoscale features such as a low-level vorticity maxima in the initial state. As a result, position errors of the rain area in the late hours were reduced, while the model still showed slow spin-up of precipitation. The diabatic NNMI with the model physical processes did not introduce large changes of the divergent motion due to weak diabatic heating calculated in the NNMI. The NNMI using observed rain rates, in contrast, produced strong divergent motion and associated updrafts corresponding to the cloud system in the initial state. However, diabatic heating in the forecast is still weak, and the divergent motion in the initial state dissipated rapidly. In the case where all three methods, i. e., the FA cycle, the NNMI using observed rain rates, and the moisture initialization were combined, the model produced realistic precipitation from the first hour of the forecast, and the effects persisted for a long time. It was found that smooth transition from initialization to forecast is important for the rapid spin-up of precipitation.