논문 상세보기
pp.95-106
The characteristics on the transtion probabilities and periodicity for the daily precipitation occurrence in Korean peninsula are investigated by applying the Markov chain properties to daily precipitation occurrence. In order to examine the responses of Markov Chain properties to the applied period and their magnitudes, three cases (Case A: 1956∼1985 at 14 stations, Case B: 1965∼1994 at 14 stations, and Case C: 1985∼1994 at 63 stations) are considered in this study.
The transition probabilities from wet day to wet day for all cases are about 0.50 and in summer, especially July, are higher. In addition, considering them in each station we can find that they are the highest at Ullung-do and lowest at Inchon for all cases. The annual equilibrium probabilities of a wet day appear 0.31 in Case A, 0.30 Case B, and 0.29 Case C, respectively. This may explain that as the data-period used becomes shorter, the higher the equilibrium probability is. The seasonal distributions of equilibrium probabilities are appeared the lowest(0.23∼0.28) in winter and the highest(more than 0.39) in spring and monthly in July and in October, repectively. The annual mean wet duration for all cases is 2.04 days in Case A, 1.99 Case B, and 1.89 Case C, repectively. The weather cycle obtained from the annual mean wet and dry duration is 6.54∼6.59 days, which are closely associated with the movement of synoptic systems.
And the statistical tests show that the transitions of daily precipitation occurrence for all cases may have two-state first Markov chain property, being the stationarity in time and heterogeneity in space.
