摘要
鉴于合理选择预报因子和预测方法是提高中长期水文预测准确性的重要环节,以青海省西宁市为例,采用单相关系数法和相关概率法对气温和降水资料进行相关性分析,并在此基础上进行线性和非线性回归预测。结果表明,在给定95%置信度条件下两种相关性分析方法均通过检验,气温和降水线性相关显著,可将气温作为预报因子进行降水预测,但线性回归预测结果不理想,而采用最小二乘支持向量机的非线性方法预测结果较理想。
Rational choice of predictor and prediction method is an important part of improving mid and long term hydrologic forecasting accuracy.Taking Xining City in Qinghai Province for an example,single correlation coefficient method and relative probability method are adopted to analyze the correlation between air temperature and precipitation data.And the linear and nonlinear regression prediction is implemented.The results show that the correlation analysis of two methods pass the test under the given 95% confidence level;linear dependence of temperature and precipitation is pronounced;but the temperature as a predictor of precipitation forecast,the linear regression forecast result is not satisfactory while the nonlinear least squares support vector machine method has satisfactory forecasting effect.
出处
《水电能源科学》
北大核心
2012年第11期18-20,213,共4页
Water Resources and Power
关键词
水文预报
相关系数法
相关概率法
回归分析预测
最小二乘支持向量机
hydrological forecast
correlation coefficient method
correlation probabilistic method
regression analysis prediction
least square support vector machine
