GPS测站坐标非线性变化研究方法的比较与分析

以GPS测站残差时间序列为研究对象,采用自回归滑动平均模型（ARMA）的两种建模方法（Box-Jenkins方法和Pandit-Wu方法）和三角函数法对其进行建模拟合与分析比较。通过实验发现Pandit-Wu方法拟合效果最好,Box-Jenkins方法次之,三角函数法最差,但是通过不同的测站残差数据实验比较也发现三角函数法能更加直观地体现出非线性变化的规律,有利于进一步分析非线性变化规律的性质与机制,因此,同样具有深入研究的价值。

Comparative analysis of studying on GPS station non-linear variation

Taking GPS residuals as research object,we use three methods(Box-Jenkins method,Pandit-Wu method,triangle method) to deal with the GPS residuals,and take them for comparing.It is found that Pandit-Wu method is the best one,and triangle method is worse than the others.However,through different experiments,it is also found that triangle method can exhibit the law of non-linear variation directly,which can help us go deep into studying on the reason of non-linear variation.So it is worth for further study too.