Research on Adjusting Effect for Different Mathematical Model of Precipitation Series
-
摘要: 采用逐步多元线性回归模型、一元线性回归模型、综合法模型和比值法模型对分布于全国31个台站的降水量资料进行了模型订正效果的试验与分析。结果表明:(1)逐步多元线性回归模型对年降水量序列的订正效果较好,其相对拟合误差总平均在0.08以下,而其它3种模型与月降水量序列的订正效果较差,相对拟合误差总平均在0.11以上;(2)若将年降水量序列相对拟合误差控制在0.10以内,则要求确定订正方程式的平行资料年数为10年或以上,要求订正站与基本站序列的相关系数在0.85以上;(3)较湿润地区拟合误差较小,反之较大。Abstract: Based on the model of successive multiple linear regression, one-variate linear regression, ratio value method and aggregate method the analyses and tests of precipitation series correction are carried out by using the data observed at thirty-one stations in twenty-five Provinces of China. The results show that: the correction effect of multiple linear regression method for annual precipitation data series is better than those of other methods, the average relative fitting error is less than 0.08, and the other is more than 0.11 for monthly data series; if the relative fitting error is less than 0.10, the time of paralled data requested in the equation should be longer than 10 years, the correlation coefficient above 0.05; if the precipitation is above 300mm, the error is less than 0.10 and if below 300 mm, the error is more than 0.11
计量
- 摘要浏览量: 3360
- HTML全文浏览量: 260
- PDF下载量: 1356
- 被引次数: 0