拟合推估法可用于拟合系统误差,削弱随机误差的影响,提高空间数据质量。与正常拟合推估法不同,提出了一种抗差拟合推估法以控制GIS空间数据的系统误差和异常误差的影响。在此基础上,提出了对剩余残差再拟合的思想,从而对拟合推估剩余系统误差进行控制。利用该方法,一方面可以有效地抵制系统误差的影响,另一方面又可以较好地遏制异常误差的污染。此外,对因协方差函数误差导致的信号估计偏差,也具有一定的补偿能力。最后通过实际计算与分析验证了该方法的有效性。
Collocation method is a helpful tool to fit the systematic errors and weaken the effects of the stochastic errors. A robust collocation method which was different from the standard collocation was introduced in this paper. It could control the influence of systematic errors and outliers of spatial datum in GIS. Furthermore, the idea of re-fitting for residual, in employed in order to compensate the remained systematic errors, was also given. This method could not only resist the influence of systematic errors, but also reduce the effects of outliers. In addition, it could reduce some estimated signal bias caused by inaccuracy covariance function. An actual calculation and corresponding analysis showed that the proposed method was effective in improving the quality of digital maps.