Simultaneous observations were made of the fall velocity, size and mass of snowflakes in the steady falling state. It was found that the fall velocity depended upon both the size and the density of the flakes, but could not be represented as a function of their mass. An aerodynamic equation which expresses the fall velocity as a function of the density of a snowflake σ, was obtained ; u=330(σ-ρ)1/4 which agrees very well with the empirical data, where ρ is the density of air.
The non-adiabatic heating due to the latent heat released by the condensation of water vapor through the simply parameterized convective activity is incorporated in the quasigeostrophic four-level numerical prediction model which is in operational use at Japan Meteorological Agency. This effect is so formulated that the heating takes place only over the cyclonic domain and the amount of heating is assumed to be proportional to the intensity of the relative vorticity at the lowest level. An air column resolved in the four level model is heated, at first over its lowest layer (700-900mb) alone (Version B), secondly over its lower two layers (500-700 and 700-900mb) at a certain distributive proportion of heating (Version C) and at the third over its whole three layers (300-500, 500-700 and 700-900mb) at some rate of distribution of heat (Version D). Comparing a few examples of numerical weather prediction by means of the nonadiabatic model mentioned above with the observed patterns and with results of the operational routine model which has no parameterized convective heating (Version A), we observe that (1) the relatively small scale cyclones and the cyclonic circulation of typhoons are well reproduced in surface prognostic charts by Versions B, C and D, though the predicted deepning of the disturbances does not cease, but that (2) the retardation of the predicted displacements of typhoons is common in all the versions, and also (3) when we cannot successfully forecast the upper level pattern, the inclusion of the heating exaggerates the error, and that (4) Version B distorts the vertical structure of the disturbances, that Version C well describes the development of cyclones in the lower troposphere and that Version D seems to be appropriate for simulating the behavior of typhoons.
Universal equations of diffusion are introduced. These equations can be applied to diffusion from a point source with infinite releasing time-interval to relative diffusion from an instantaneous source and to any other types of diffusion. Dependences of eddy diffusivity on time-distribution and spatial dimensions of clusters are clarified. In the course of the derivation of the equations, local time-space correlation of velocities is formulated, as a product of the correlation coefficient and local energy. The equation of local variance of stochastic time series is applied to the local energy in two-dimensional clusters with spatial and time-dimensions, by introducing the virtual one-dimensional scales of the clusters. Taylor's hypothesis about the relations between the Eulerian and Taylor-Karman correlations are made. The universal equations derived are solved numerically and for special cases analytically. The solutions for some cases are calculated and illustrated.
Potential differences of magnitude ranging from about 10 to 100 millivolts were found to exist between the surface layers and the interior of three hot pools in Yellowstone Park. Their long-term variations (of duration greater than 5 sec.) were linearly proportional to the simultaneous variations of temperature difference between the same points ; the gradients of the curves obtained ranged from about 400 to 700 mV.°C.-1 Two of the pools exhibited rapid and extremely regular pulsations of potential difference of the type observed by Sisler. These pulsations, which were unrelated to simultaneous variations in temperature difference, persisted with little change in frequency over many hours.