【摘要】 将经度、纬度、海拔、坡度、坡向、分区(干燥度)和植被等七个因子作为影响现代降水的地理因子,结合R语言程序、Arcgis空间分析和SPSS统计分析工具,探讨对现代降水进行空间模拟时,其影响因子与重建方法的选择。基于广义可加模型(GAM)用R语言实现了对现代降水影响因子的非线性分析,得到除坡度外其它各因子均与多年平均降水量之间大体呈线性相关;继而对各地理因子进行共线性检验,通过检验的因子使用最小二乘法建立与多年平均降水之间的多元线性回归模型。计算结果表明:逐次叠加地理因子进行回归分析时,其方程的R2均有提升,显示地理因子的类型对降水空间模拟具有一定的影响;非线性和线性建模中坡向因子的P值都大于0.05,说明在本研究中其对降水的影响不显著;线性回归建模的模型拟合优度R2为0.849,非线性模型的解释达到89.6%,两种建模结果都具有一定的可信度,对分析今后中国大范围区域中现代降水的影响因子及空间模拟方法具有重要的参考意义。更多还原

【Abstract】 The seven geographical factors, such as longitude, latitude, altitude, slope, aspect, regionalization(dryness) and vegetation, were used as influencing factors of modern precipitation. Using R programming language, Arcgis spatial analysis and SPSS(Statistic Package for Social Science) statistical analysis software, this paper discussed on how to select the influence factors and reconstruction method in spatial simulation of modern precipitation. Based on generalized additive model(GAM) by R language to realize the nonlinear analysis between modern precipitation and its influence factors, and the results were obtained that all the geographical factors are roughly linearly related to the mean annual precipitation except the factor of slope. The multivariate linear regression model was established between the geography factors and the multi-year average precipitation, and these factors were tested in multiple collinearity. The calculation shows that R2 of model was promoted in successive geography factors regression analysis, which means the type of geographical factors have a certain influence on the precipitation space simulation; the P values of the slope orientation factor in nonlinear and linear models are larger than 0.05, indicating that the influence of slope orientation factor on the precipitation is not significant; R-squared of linear regression model is 0.849 and deviance explanation of non-linear model is 89.6%, so as that the results of the two models have a certain credibility.更多还原