Three methods for estimating a transfer function of a system with two inputs were compared. The first is a method based on the analysis in frequency domain (FD method). The second method makes use of the autoregressive model (AR method). The third one uses the multiple regression model (MR method). These methods were applied to 100 pairs of artificial data which were synthesized on the basis of three kinds of assumed transfer functions. It is concluded from the results of comparison that MR method is the best of the three.
Air concentration and deposition of radioactive pollutants in Europe after the Chernobyl nuclear power plant accident is simulated using a long-range transport model developed in the MRI. This model is composed of the weather forecasting part which was the routine regional weather forecasting model of the JMA and the Lagrangian advection-diffusion part. The source data distributed by the ATMES project is used to determine the initial release rate from the power plant. Calculated concentrations of Cs-137 and I-131 in the surface level atmosphere agree well with observation. Surface deposition of Cs-137 has, however, poor correlation with the observed deposition. Bad precipitation forecast of the weather forecasting part of the model and the difference between the horizontal scale represented by observation and by simulation are considered to be responsible for this poor correlation.
In order to study the orographic effect on the rainfall distribution in the Kii peninsula, airflow over a three-dimensional mountains is simulated by use of a non-hydrostatic model and compared with the observed distribution of rainfall. Numerical experiments for simplified orography show that low-level updraft areas have two peaks in weak wind cases. The first peak is on the windward slope of the mountain, while the secondary peak is over the sea, which is associated with the low-level blocking by the mountain. Locations of both updraft areas shift upstream as wind velocity decreases, while in cases of stronger wind or less atmospheric stability, no secondary peak appears over the sea. The characteristic curvature of the orography of the Kii peninsula intensifies the low-level blocking in the case of southeasterly environmental wind. Heavy rainfall events in the Kii peninsula in 1985 are analyzed and it is confirmed that 1) the rainfall amount in each event at Owase tends to be higher than that at Shionomisaki, and 2) heavy rains occur at Owase when low-level synoptic wind is east-southeasterly to southerly. In numerical experiments for real orography of the Kii peninsula, low-level updrafts are found around Owase and the southern part of the Kii peninsula in easterly to southerly winds. The positions are well coincident with the positions of two maxima in mean annual rainfall distribution shown by Sakakibara and Takeda (1973). No secondary updraft occurs in the case of southwesterly to westerly winds. Numerical simulations using atmospheric conditions observed in heavy rain events in 1985 are conducted. The locations of simulated low-level updrafts correspond well with the observed heavy rain areas around Owase and the southern part of the Kii peninsula. In some cases, middle-level updraft areas associated with mountain waves are computed aloft above the observed heavy rain areas to the northeast of Owase. The experimental results are consistent with the empirical relation between the location of heavy-rain areas and wind speed: the sea-shore region in weak-wind type and the region deeper inland in strong-wind type. It is also suggested that low-level convergence due to the blocking effect of the mountains affects the orographic enhancement of the convective rain band which is often seen over the sea in the windward side of the Kii peninsula.