干旱区生态环境知识资源中心

Knowledge of the pattern of occurrence of peak runoff events is valuable for managing activities in watercourses and their environs. Current approaches consider the inter-arrival intervals between events as being exponentially distributed with a fixed seasonal parameter. In a new approach (Ben-Zvi and Meirovich, Stochastic Hydrology and Hydraulics, 1997) the occurrence day is considered as a random variable to whose series a gradually-varied distribution, such as the normal or Pearson III function, can be fitted. The statistical moments of occurrence days facilitate such fits. The present work extends the consideration, via a case study of partial duration series, to the occurrence times of high-discharge events. The first three statistical moments of occurrence days in series of runoff peaks have been determined for a sample of nine hydrometric stations in Israel. The mean occurrence day of high peaks at eight stations falls earlier than that of respective low peaks, whilst for one station the opposite trend is observed. The shifts at five stations are found statistically significant. The standard deviations of occurrence days in partial duration series for the sample stations are mostly shorter than those in the respective complete series of events. The occurrence days of runoff events in the complete series at seven stations are about evenly distributed around their respective mean, whereas most of the other occurrence series are skewed.