Observations of the brightness temperature of the sea surface were carried out in 1982 and 1983, using the Bread Board Model (BBM) of the Microwave Scanning Radiometer (MSR) which will be on board MOS-1 in early 1987. The observational results show that at 23 GHz, one of the two frequencies selected for MSR, the brightness temperature observed with MSR is explained by a radiative model with an error smaller than 5 degrees. At 31 GHz, the other frequency of MSR, the error becomes larger than 10 degrees. At frequencies used for MSR, a linear regression method, applied to the surface water vapor pressure and the surface wind speed, gives the result that signs of coefficients of the surface wind effect are almost all negative. Instead of this method, two separate approaches give them 0.4 and 0.8 K/m/s at 31 GHz in the horizontal polarization. One is to seek an example in which the surface wind increases greatly within a short period, and the other is to eliminate the water vapor effect from the observed temperature, using the coefficients resulting from the radiative model.

View full abstract

   A linear characteristic signal, which is converted from the output amplitude of a logarithmic receiver, can be used for MTI weather radar's signal processing.
   This paper describes the input-to-output characteristics (normal and MTI signals) of precipitation echoes due to the technique investigated by simulation. There are two errors affecting the system's performance, that is, the limitation of the receiver dynamic range and the quantization effect of A-D conversion.
   The amplitude ratio between the adjacent levels which gives the A-D conversion's quantization resolution is uniform for the whole input range of the logarithmic receiver. Therefore, the reproducibility in the small amplitude region is essentially superior to that of the linear receiver technique. However, since the following log-lin conversion makes the sensitivity decrease in the small amplitude region, it is necessary to increase the bit number of the conversion by 4 bits, by which the accuracy can reach the limit decided by the A-D conversion's resolution of the log receiver itself.
   Concerning the linearity of the characteristics of the system, in the small signal region, the normal signal causes positive errors and the MTI signal negative errors. In the saturation region, on the contrary, the MTI mean amplitude decreases for the input signal amplitude beyond some specified value, because the MTI signal is processed from the amplitude difference between the adjacent pulses. Therefore, the overall characteristics of this technique are determined by the MTI signal processing. In cases where the bit numbers of the system dynamic range and the A-D conversion are both set in 8 bits, the minimum measurable rainfall intensity is 0.46 mm/hr and the unambiguous rainfall intensity reaches 138 mm/hr.
   It is stated that the MTI signal amplitude can be calibrated with the measured normal signal amplitude even though it is also accompanied with the system's quantization error.

View full abstract

   Recently, many seismologists have been analyzing the fine rupture process of earthquakes by the use of strong motion seismograms recorded near the source regions. Not only long-period waves, but also waves of periods less than several seconds are found to play an important role when analyze the detail of rupture processes. On the other hand, amplitudes and forms of seismic waves are strongly influenced by anelastic properties of the earth. In this paper, synthetic displacements are calculated on the surface of an anelastic layered half-space in order to estimate the effect of attenuation on the short-period waves near the source region.
   If effective Q value is constant over the whole range of observed frequency (e.g. Q is constant in the frequency range of τ₁^-1<<ω<<τ₂^-1.), the effect of anelasticity is introduced by the use of the dispersive complex velocities as;
   υ(ω) = υ₁ [1+1/πQ ln (ω/2π) -i/2Q] (i)
   where υ₁ is a phase velocity at ω=2π. Moreover, we must introduce the velocity at ω=0 to evaluate the static deformation. Based on the relaxation mechanism of Liu et al. (1976), the velocity at zero frequency is approximately given as
   υ₀∼υ₁ / [1+1/πQ ln (2π · τ₁)] (ii)
   In this paper, τ₁ is assumed to be 2×10⁴ sec, whose reciprocal approximately corresponds to the minimum angular frequency of the free oscillations of the earth.
   Extending the approach by Yao and Harkrider (1983), dynamic displacements and strains on the free surface are given in terms of reflection-transmission matrix [Kennett and Kerry (1979)], and discrete wavenumber method [Bouchon (1981)] and by using the layered half-space structure with the complex velocities of the equations (i) and (ii). The Fourier spectrums of each displacement and strain component are given in this paper.
   The conducted numerical calculation presented the ground displacements and velocities on the free surface sedimentary medium with low Q layer. We also calculated ground displacements and velocities without the effect of anelastic attenuation. A comparison of the synthetic waveforms shows the apparent effect of anelasticity. When the thickness of the sedimentary layer is more than 1km, the amplitudes of the later phases whose travel times are more than 10 seconds are attenuated remarkably. Though the synthetic waves may be within 10km of the epicenter, the amplitude calculated for the anelastic half-space is less than half of what is calculated for the elastic half-space.
   In order to investigate the complicated rupture process of the earthquake, such as multiple events, a more realistic estimation of the amplitudes of the later phases must need the analysis of the near-field strong motion seismograms which are recorded on the sedimentary layer. Our method must be useful for such analysis.

View full abstract