中文摘要：

多种类型高分辨率重力场数据的不断增加,使得在局部范围内精化重力场模型成为了可能。本文采用Abel-Poisson核将重力场量表示成有限个径向基函数线性求和的形式,对局部区域的多种重力场数据进行联合建模。为了提高运算速度,运用了基于自适应精化格网算法的最小均方根误差准则（RMS）来求解径向基函数平均带宽。以南海核心地区为例,联合两种不同类型、不同分辨率的重力场资料（大地水准面起伏6＇×6＇、重力异常2＇×2＇）,构建了局部区域高分辨率的重力场模型。所建模型表示的重力场参量达到了2＇×2＇的分辨率,对原始的重力异常数据（2＇×2＇）拟合的符合程度达到±0.8×10^-5m/s^2。结果表明,利用径向基函数方法进行局部重力场建模,避免了球谐函数建模收敛慢的问题,有效提高了模型表示重力场的分辨率。

英文摘要：

With the increasing number of various types of high-resolution gravity observations,earth gravity models can be regionally refined. We use Abel-Poisson kernel to represent the gravity as the linear summation of finite radial basis functions and combine the multiple gravity data to build a regional gravity model with high resolution.The minimum root mean square criterion based on the data adaptive algorithm is proposed to calculate the base function,which promote the speed of computation significantly. Taking the central South China Sea as an example,two different types of gravity data,namely geoid undulations with resolution of 6＇ ×6＇ and gravity anomaly with resolution of 2＇×2＇,are used to construct the high-resolution regional gravity model. The model has a resolution of2＇×2＇,and has a great agreement with original gravity anomaly,reaching to ±0. 8×10^-5m/s^2. Our results show that using radial basis functions to construct the regional gravity field can avoid the problem of slow convergence of spherical harmonic functions,and can improve the resolution remarkably.