Abstract
This paper presents the results obtained from the statistical analysis of summer climate conditions in relation to the growth and the yield of rice crops. The climatic data used here were the yearly series (1889-1965) of mean air temperature for its vegetative period (July to August) at fourty stations. From these yearly series of the temperature were determined the coefficients of variation, return period, autocorrelation coefficient and spectrum distribution, respectively.
The obtained results may be summarized as follows:
1) Deviations of the July-August mean air temperature from the normal were fairly large in the estern part of Hokkaido, that is, the value-of coefficient of variation (C. V.) reached about 9% in this part, while the values of C. V. in south-western part of Japan were less than 3%. These results seem to indicate that there is difference in the stability of summer climate between both regions.
2) Return period (R. P.) of the July-August mean air temperature was determined by an empirical way. The mean air temperature for a most recent year with the cool summer damage of rice crops was adopted as a climatologically critical temperature for the occurrence of the cool summer damage at districts concerned. In Tohoku region the critical temperature has been reduced by 1-2°C on accont of the marked improvement on the cultivating technique since 1955. However, the critical temperature in Hokkaido district has hardly been lowered, although improved agricultural technique has widely been adopted to reduce the direct influence of weather conditions upon rice crops. It was found that the probability of being damaged by cool summer is still so high as its return period is 2.5 years.
3) Correlograms of yearly series of the July-August mean air temperature showed the high stability of the climate in the southern part of Kyusyu and Shikoku, while in the eastern part of Hokkaido the values of R(t) decreased rapidly with lag time, which show the large variability of temperature in summer. The shapes of correlograms in the other districts except for Hokkaido were too diversiform to interpret the variation of temperature.
4) Spectral analysis of both the yearly series of summer temperature and of the rice yield in Hokkaido and Aomori Prefec. was done with an electric computer (TOSBAC-3400). The peak of the spectral densities in both the series did not satisfactorily agree with each other. Periodicity of the air temperature was not so clear as found by Ozawa et al. (1952), mainly because the yearly series of the air temperature was not enough in the length to find statistically the pronounced periodicity.
