Monotonic trend and step changes in Japanese precipitation

For the purpose of detecting the possible long-term trends of Japanese precipitation, both parametric ttest and nonparametric Mann-Kendall and Mann-Whitney techniques are applied to the spatially averaged precipitation time series over Japan. The results indicate that although several step changes occurred in Japanese precipitation, the time series did not exhibit significant evidence of monotonic trend during the past century. When a step change is present, the number of observations required for detecting the trend of a given magnitude at a specified significance and power level is investigated with the power function of the t-test. Results indicate that if the magnitude of the step change reaches one or two times of its standard deviation, the previous 50-year of record together with 5 years or more of new data will be available for detecting the possible trend. This conclusion may be helpful for the detection of step changes in the regions where the precipitation has near-normal distributions.