ANALYSIS OF MULTIFRACTAL PROPERTIES OF TEMPORAL AND SPATIAL PRECIPITATION DATA IN JAPAN

摘要

First, we examined scaling and multifractal properties of temporal precipitation data with an emphasis on the structure of the rainy vs. non-rainy periods, using the so-called β model. These data have scaling properties in a range of 1 day ≤ L_n ≤ 32 days. We characterize the scaling of the rainy areas with the single parameter of the β model and classified surface observation stations into two groups according to the parameter; namely, along the Pacific Ocean and Japan Sea coasts. The Pacific Ocean data are more intermittent and peaky than the Japan Sea data and exhibit more multifractality than the latter. Second, we analyzed the time variation of scaling properties of several months of spatial radar data using the intercept of the regression estimate of the scaling of the fractional wetted area. These results showed that over the sea in the summer, there was a tendency for a different scaling of small vs. large scales. Finally, we extracted four independent precipitation events from the whole dataset and examined whether or not a one-to-one function can represent the relationship between mesoscale forcing and the parameter of the β model. We conclude that there is a one-to-one functional relationship if we consider each precipitation event separately and exclude multiple independent precipitation events.