数据驱动模型在渭河流域来水预报中的开发和应用研究

现有的水文预报方法分为过程驱动模型方法和数据驱动模型方法两大类,近年来随着水文数据获取能力和计算能力的发展,数据驱动模型在水文预报中受到了广泛的关注,回归模型、神经网络模型均属此类.本文首先采用距离平方反比与泰森多边形相结合的方法,由雨量站观测到的降水量求得渭河流域各四级区的降水量,然后采用自回归模型结合年～月～旬逐级修正的方法对渭河流域10个四级区进行降水预报,采用降水～气温～径流多元线性回归模型对渭河流域28个水库进行径流预报,取得了较好的效果.

Development and Application of Data-driven Model in Hydrological Forecasting for Weihe River Basin

There are two kinds of existing methods for hydrological forecasting,which are data-driven models and process-driven models.As the development of calculation and acquirement of hydrological data,data-driven models draw more attention in hydrological forecasting,such as regression models and neural network models.Firstly,this paper completed the interpolation of point rainfall to sub-basins based on the Reverse Distance Square(RDS) method or Thiessen Polygon method,which depend on the integrity of rainfall d...