The impact of SST anomalies on the dynamical long-range forecast of monthly mean fields in the early summer of 1983, which was studied by Tokioka et al. (1987; TYC 87), is examined again. The shortcomings in the results of TYC 87 are eliminated by including SST anomalies over the Indian Ocean, using a modified Arakawa-Schubert cumulus parameterization and using the 15 layer model.
   In the tropics, anomaly correlation coefficients are high both in the first and the second month in the present study, while they are low in the second month in TYC 87. The model simulates the intraseasonal variation in the tropics. The forecast skill in mid-latitudes for the second month is higher than that of TYC 87 both in the Northern and Southern Hemispheres.
   The results of the present study are encouraging for the dynamical long-range forecast of monthly mean fields beyond one month with the use of real SST and a good model.

View full abstract

   When the Joetsu-Shinkansen started its business from Omiya to Niigata in central Japan, 2Hz-tremors (henceforth the Tremor) with slightly larger amplitudes than usual microseisms' began to be regularly recorded on the seismometer at the Kumagaya Local Meteorological Observatory (KLMO) located about 1km away from the railroad site. In order to understand the Tremor's source and generation mechanism, we made temporary microseisms observations and train-speed measurements.
   The results are as follows: the Tremor's emergence corresponds with the time when Shinkansen trains bound for Omiya pass through the Kumagaya area; from the Tremor's particle motion and the examination of its source position mentioned below, the Tremor is a Rayleigh-wave coming from WNW or NW at a speed of 60-110m/s or 78-110 m/s. Referring to the results of other observations of Shinkansen noise-microseisms (Hosoyama & Nakai, 1972), and the noise level at KLMO, we understand that the distance from KLMO to the position of Tremor's source, is within 3km; so we searched for the Tremor's source in the area which is within 3km from KLMO. We found 4 piers whose features are different from those of other piers. They are located 2.4km to the west of KLMO. They have supports which reach deeper in the ground than other piers. The total length of their girders on them is 217m. If we assume these four piers as the source of the Tremor, we can explain the arrival times and maximum velocity amplitudes of the Tremor consistently, further, if we assume that the Tremor's predominant frequency and trains' speed are in proportion, we can interpret the relations among the Tremor's duration time, its predominant frequency, and its maximum velocity amplitude consistently.
   The Tremor's generation mechanism is as follows: the Shinkansen trains pass over the railroads on the piers A1-A4, which push the earth; then Rayleigh-waves, which are to be observed at KLMO as the Tremors, are generated and propagate with 60-110m/s or 78-110m/s group velocity.
   The source model is as follows: each of the A1-A4 piers receives forces from the two girders independently put on it; each pier vertically pushes the elastic body (the earth) that is assumed to be semi-infinite, isotropic, and homogeneous. The present elastic-wave-generation problem is dealt with as a two-dimensional problem for simplicity.
   After this model, synthetic waveforms are calculated. Though only 2.0-2.5Hz waves are observed, the calculated spectra show peaks at around 1Hz, 2-3Hz and 6-7Hz; the existence of selective propagation is thought of as a cause of this phenomenon. The Kumagaya area is covered with a low velocity surface layer whose thickness is larger than 16m and whose P-velocity is 345m/s. These values are obtained at the New Kumagaya Transformer Sub-station (NKTS) located 3km to the west of KLMO. The quarter-law (Tazime, 1956) suggests that the Rayleigh-wave with 2.0-2.5Hz can keep its energy density even after long distance propagation on a medium which has a 40m-thick surface layer of 345m/s P-velocity.
   It is shown by the similarities between filtered synthetic waveforms and observed ones that 110m/s is the most likely as the Tremor's propagation velocity. Even with the 345m/s as the surface layer P-velocity, the propagation velocity can be 110m/s, if the generated Rayleigh-wave is dispersive. This is because the ratio between the rigidity of the surface layer and that of the half-infinite layer below it can be calculated as 4.8-14 from the data obtained at NKTS.
   The above discussions reveal that the present model is acceptable, though the following points should be clarified by another observation: the hypothetical velocity structure & surface layer thickness, and the hypothesis that Tremor's predominant frequency and trains' speeds are in proportion.

View full abstract