中文摘要：

经典箱形图运用到有偏降雨数据时，常常将超过触须区间的有偏数据错误地判断为异常值。利用偏态的抗差度量参数，对经典触须区间进行偏态改造，以满足有偏降雨数据异常值的探测需求。分析和比较了度量参数的3种改造形式：线性函数、二次函数、幂函数，结果显示幂函数形式的探测风险最小。将经典箱形图和改造箱形图运用于无异常值和有异常值数据中，改造箱形图的的探测风险小，且探测效果良好。

英文摘要：

When the data are skewed, some data usually exceed the whisker zone of the classical boxplot and are often erroneoustly declared as outliers. For skewed rainfall observations, an adjustment of the boxplot is presented that includes a robust measure pa rameter of skewness in the determination of the new whisker zone. This results in a more accurate representation of the skewed data. The three adjusted functions- line function, quadratic function and exponential function are analyzed and compared. The results show the exponential function suits skewed rainfall data sets. The two methods are applied to uncontaminated and contaminated rainfall data. The results show that the adjusted boxplot has less detection risk and better detection effects.