利用直接求导法和递推法的高阶解析延拓算法

A High-Order Analytical Continuation Algorithm Based on Direct Derivation and Recurrence Methods

  • 摘要: 解析延拓算法在全球重力场模型构建、区域大地水准面精化等领域发挥着至关重要的作用。过去解析延拓计算普遍采用一阶算法,高阶延拓算法鲜有应用。为提高解析延拓算法的精度,本文对高阶解析延拓算法展开深入研究。高阶径向导数的计算是实现高阶解析延拓算法的关键环节。首先,本文利用直接求导法推导了调和函数与重力扰动的高阶径向导数的通用计算公式,并在此基础上给出了基于直接求导法的前六阶径向导数的具体算式。随后,本文利用递推法推导了调和函数与重力扰动的前六阶径向导数的具体算式。数值实验结果表明,利用直接求导法推导的高阶径向导数算式存在理论缺陷,而利用递推法推导的高阶径向导数算式是有效可行的。根据本文的研究,应用解析延拓算法时应使用递推法推得的径向导数算式,延拓阶数推荐使用三阶或四阶。

     

    Abstract: The analytical continuation algorithm plays a vital role in constructing the global gravity field model, refining the local geoid, etc. The one-order continuation algorithm is widely adopted in conventional analytical continuation calculations, while the high-order counterpart is rarely applied. To enhance the accuracy of the analytical continuation algorithm, an in-depth investigation on the high-order analytical continuation algorithm is carried out in this paper. The calculation of the high-order radial derivatives is critical to the realization of the high-order analytical continuation algorithm. To begin with, the general formulae for the radial derivatives of harmonic function and gravity disturbance are deduced by the direct derivation method. Subsequently, the specific expressions for the radial derivatives of the first six orders based on direct derivation method are given according to the general formulae. Next, the radial derivatives of the first through sixth orders of harmonic function and gravity disturbance are deduced using the recurrence method. Obvious discrepancies exist between the formulae for the high-order radial derivatives obtained using the two derivation methods. Numerical tests demonstrate that the formulae for the high-order radial derivatives obtained by the direct derivation method have inherent theoretical defects, while those deduced via the recurrence method are theoretically sound and practically applicable. Therefore, the high-order analytical continuation algorithm should adopt the radial derivative formulae from the recurrence method rather than those from the direct derivation method. For quantitative assessment of the performance of the high-order analytical continuation algorithm, numerical experiments are carried out on full-degree model gravity data with a resolution of 2′×2′ using the formulae for the radial derivatives obtained via the recurrence method. The experimental results indicate that, under the noise perturbation of 1mGal, the measurement-surface continuation accuracies of the two-, three-, four-, five- and six-order analytical continuation algorithms are improved by 36.26%, 47.30%, 50.00%, 50.68% and 50.68% compared with the conventional one-order algorithm, respectively. Lastly, analytical continuation tests on the terrestrial observed gravity data are performed using the radial derivative formulae obtained using the recurrence method. The results show that the two-, three- and four-order continuation algorithms outperform the conventional one-order algorithm by 31.65%, 39.66% and 41.77% in terms of the continuation accuracy, respectively. Compared with the four-order analytical continuation algorithm, both the five- and six-order algorithms achieve only a marginal accuracy improvement of 0.01mGal. According to the findings of this study, the three- or four-order continuation is recommended for practical applications of analytical continuation algorithm.

     

/

返回文章
返回