为了克服数字高程模型（DEM）精度估计中传统方法受误差分布非正态性的影响，发展了基于自适应M（AM）估计的DEM精度评价。AM估计以高崩溃污染率的估计值为DEM误差指标初值，以误差残差分布为误差权重选择依据，通过对估值结果多次迭代，最终实现DEM误差估计。以DEM误差均值和标准差为精度指标，数值试验和实例分析表明，传统的非抗差估计法受粗差影响最为严重，使其估计结果远偏离真值；尽管“3?滓准则”和传统的抗差估计法能够剔除部分粗差，但一定程度上仍受粗差影响，使结果的精度较低；AM估计受粗差影响最小，对DEM误差指标估计精度最高，可用于误差非正态分布的DEM精度评价。
The report on of DEM accuracy is commonly based on some global statistical measures, such as the mean and standard deviation of DEM errors. Usually, the specification of these accuracy indices is based on the assumption that the error distribution is normal and there is no outlier and systematic error. However, such an assumption is rarely an exact statement owing to the malfunction or improper calibration of instruments, mistaken readings, gross recording and calculation, and improper execution, etc, particularly when a DEM is directly derived from digital photogrammetric systems and active airborne sensors including Light Detection and Ranging (LiDAR) and InSAR. A robust estimator based on the adaptive M-estimation principle (REMP) has been developed for the DEM accuracy assessment. The iteration of REMP starts from estimations with a high breakdown point and selection of the weights of errors with the residual distribution. In terms of DEM mean and standard deviation errors, two examples including a numerical test and a real-world one were employed to comparatively analyze the results of REMP and the classical non-robust and robust estimators. Results indicate that under the non-normal distribution of DEM errors, the classical non-robust estimators are seriously influenced by the non-normality. Some robust estimators, such as 10% trimmed or Winsorized mean, normalized median absolute deviation are not very robust to resist the influence of outliers. REMP that is slightly affected by the non-normal distribution of DEM errors is more accurate than the classical estimators. The robust methodology can adapt to the DEMs, especially the ones derived from remote sensing, such as LiDAR or digital photogrammetry in the non-open terrain.
