After pointing out the shortcomings in the traditional procedure for the investigation of relationship between yields and climate,new concepts -- 'normal yield' and 'reduction of yield' are developed to replace the common used 'trend yield' and 'climate yield'.
Climatic factors which impose a great influence on the yield of rice are selected by the stepwise regression analysis. Principal factor analysis is used to extract the main information from the factors, and then the scores of a new top factors are classified into patterns by the procedure of cluster analysis. The principal results are summarized as follows:
1. The yield of early rice is related to rainfall in March, sunshine in May and rainfall in the period from the late May to the middle June.
2. Sunshine in October and temperature in November have an influence
on the yield of late rice.
3. Most of the selected climatic factors have clear periodicity.
4. Yearly climates are classified into four kinds of climate pattern.Every climate pattern well correspond to a certain year's harvest