Monte-carlo与NNBR模型结合在年降水量预测中的应用

以内蒙古五原气象站近40 a的降水资料为例，从降水发生的随机性、统计规律性和统计相似性入手，将蒙特卡罗方法(Monte-carlo)和最近邻抽样回归模型(NNBR)结合，对降水量进行预报。Monte-carlo方法的预报结果虽能很好地体现研究区降水发生的随机性和统计的规律性，但却不能准确反映出降水序列之间的排列顺序，使预报序列具有多解性。利用NNBR模型对Monte-carlo的预报序列依据最相似理论重新排列，确定预报序列中各个成员之间的排列次序，从而实现预报序列不仅能体现出研究区内降水序列的统计规律性和相似性，同时也能反映出单一个体出现的随机性。通过检验证明模型具有合理性和较好的适用性。

Application of Monte-carlo and NNBR Combination in Predicting Annual Precipitation

In this paper,the Monte-carlo method and the Nearest Neighbor Bootstrap Regressive model was combined to predict annual precipitation in Inner Mongolia based on the 40-year meteorological data observed by Wuyuan Meteorological Station in the study area.The results predicted with the Monte-carlo method can reflect well the randomicity and regularity of precipitation in the study area,but the order of precipitation series cannot be reflected exactly,and there may be a misunderstanding about the predicted series of precipitation.In this paper,the Monte-carlo prediction series was used to resort the predicted series based on the most similar theory,thus the statistical regularity and similarity of precipitation series can be reflected by the predicted series,and the randomicity of individual precipitation series can also be reflected.The test verified that the model is quite rational and applicable.