A newly designed spectro-pyranometer was used for the measurement of the global (direct+diffuse) and the diffuse sky radiation reaching the ground. By the subtraction of the diffuse component from the global radiation, we got the direct radiation component which leads to the spectral distribution of the optical thickness (extinction coefficient)of the turbid atmosphere. The measurement of the diffuse sky radiation reveals the scattering effect of aerosols and that of the global radiation allows the estimation of total attenuation caused by scattering and absorption of aerosols. The effects of the aerosols are represented by the deviation of the real atmosphere measured from the Rayleigh atmosphere. By the combination of the measured values with those obtained by theoretical calculation for the model atmosphere, we estimated the amount of absorption by the aerosols. Very strong absorption in the ultraviolet region was recognized.
Experimental studies were made on t he balloon-borne type dew (frost) -point hygrometer, i. e., the Dew (Frost) -Point Radiosonde, in order to estimate the systematic errors due to its instrumental mechanism and observation procedures. The experiments were performed by using two t ypes of humidity generator which were specially designed. Errors of measurement of the frost point become noticeable at the frost point about 20°C, and increase with the lowering of the frost point. The factors affecting these errors are as follows: (1) Too rapid cooling and heating of the mirror surface in the servo-system of the hygrometer. (2) Diminution of water-vapor mass transfer rate from ambient air to the mirror surface at very low water-vapor concentration. (3) Improper assumption that the frost point of the ambient air is identical with the temperature of the mirror at the moment of the detection of frost on its surface dring the cooling phase. Too rapid cooling a n d heating of the mirror, together with the decrease in the mass transfer rate of water vapor at a lower frost point, not only causes exaggeration of the amplitude of the mirror temperature swing by the servo-action, but also makes the observed value of the frost point always lower than the actual one because of the delayed detection of frost on the mirror surface. An experimental formula of e s timating errors was derived from laboratory experiments. According to this formula, the overall maximum observation error was estimated as about 6.5°C under the condition of 350-400 m/min balloon ascending rate at 50 m. b level. On the basis of laboratory experiments, an improved type of dew (frost) -point radiosonde was made as 'a trial. The on-off type electric relay used for the mirror heater in the normal type dew (frost) -point radiosonde was replaced by a proportional control circuit that is composed of IC and transistors. The purpose of this replacement is to reduce errors due to the above reasons. Comparative soundings between the normal type dew (frost) -point radiosonde and the improved one were carried out at the Tateno Aerological Observatory. The effects of water-vapor contamination on the balloon and the instrumental structures were evident on both types. The errors for the normal type at 50 mb level were estimated as -1°C from the voltage change of the electric light source batteries used for the frost detection, -7°C from the servo-system, and +4°C from the water-vapor contamination effect. Summed up correction was +4°C at 50 mb level. For the improved type, water-vapor contamination was the only factor to be considered in the correction. In the lower stratosphere, observed water-vapor mixing ratios of 2.5 mg/kg and 3.0 mg/kg were obtained from the normal type dew (frost) -point radiosonde and the improved type respectively, after all corrections were performed.