中文摘要：

以蒙特卡罗理论为基础，采用P-Ⅲ型分布函数来概括流域汛期降雨量的概率特性，并以包含预测值样本系列的均值不超出历史均值范围为控制条件，经有限次试验确定预测值。结合嫩江流域尼尔基站控制断面以上1970年-2009年汛期（6月～9月）降雨资料，以1970年-2005年汛期降雨量为计算样本，预报2006年汛期降雨量，并以2006年实测降雨量进行检验，不断外推预测与检验，直至2010年。预测结果表明：当实际汛期降雨量属平水时，都会获得较高的预报精度，如2006年和2008年；在极枯情况，如2007年（降雨保证率为99．82％）的相对误差较大，为0．75；在较丰情况，如2009年（降雨保证率为19．63％）的相对误差也较大，为-0．20。所以，本方法适合平水年的预测，对于如何进行丰枯极值预报，最后从原理和方法上进行了探讨性思考。

英文摘要：

Based on Monte Carlo theory, this paper uses P-Ⅲ type distribution tuncnon to summarize me probability cnaracteristics of basin precipitation in flood season Under the control condition that the average of the sample series which includes the predicted values does not exceed the historical average range, the predictive values are determined after limited times tests. Com- bined with the flood season （from June to September） rainfall data from 1970 to 2009 up the control section of Nierji station in Nenjiang River Valley, and with the flood season rainfall data from 1970 to 2005 as the calculation sample, flood season rainfall in 2006 is then forecast. After the forecasting, the actual measured rainfall in 2006 is used to test this prediction, and then the extrapolating predictions and tests are continuously undertaken till 2010. The predicted results indicate that the prediction is of high accuracy with the actual rainfall in flood season in normal year, such as in the year of 2006 and 2008;in extremely dry con- ditions, the relative error is big, for example,in 2007, when the rainfall assurance is 99.82%,the error is as big as 0.75;in relatively water-abundant circumstances,the relative error is also big, such as in 2009, when the rainfall assurance is 19. 63%, the error is as large as -0. 20. So,this method is suitable for the flat water forecasting. This paper finally discussed how to forecast the extremums of abundant and dry years from the viewpint of theory and method.