文摘
|
在剖析随机误差与逼近误差区别的基础上,指出数字高程模型(DEM)内插误差属于逼近误差,精度评价方法不应继续延用“中误差”概念,而应采用”逼近误差”,给出了常用DEM内插方法的误差评定模型,分析了等高线间距对DEM内插精度影响的作用机制,最后解释了DEM精度分析中用“中误差”概念无法解释的几个问题。 |
其他语种文摘
|
As an important information source, DEMs (Digital Elevation Models) have found applications in a wide variety of areas, such as view shed analysis, watershed partition, derivation of geomorphic fea-tures, etc. Assessing the accuracy and quality of DEMs scientifically will assure the production, construc-tion and right applications of DEMs. Except factors of data sources, instruments and man-made, the ma-jor error of DEM roots in interpolation arithmetics. So, the assessment of the accuracy of DEMs is a kind of error analysis aiming at the interpolation accuracy mainly. There exist two methods that have the specific theoretical context and accuracy analysis among numerous methods to generate DEMs. One method is TIN (Triangulated Irregular Network) interpolation based on discrete ground elevation points, the other is bilinear interpolation used for grid ground elevation points obtained by photogrammetry. This paper gives error models of the above two interpolation methods. The two error models explain some problems that can't be interpreted using RMSE. Nowadays, the two meth-ods to generate DEMs have a wide range of application in practical DEMs productions. Both have own ad-vantages and application fields. At present, there are major methods to assess the accuracy of DEMs : test, transfer function, covari-ance and contour maps produced from DEMs. For test method, the sampling check points lack rigorous theoretical basis. While both methods of transfer function and covariance have difficulty to assess the accu-racy of the whole DEM and render the factual distribution of errors, they also can' t estimate the status of the DEM coinciding with the factual topography. In the situation of lacking a valid theoretical accuracy as-sessment about DEMs presently, contour maps produced from DEMs serve as a natural method to evaluate the quality of DEMs. In fact, as long as the method applied to DEMs interpolation is right, the maximal difference of the original topographic map and the map generated from the corresponding DEM will be less than a half of the numerical value of contour - interval. And the overlay map of the original topographic map and the map obtained from the DEM is just an elaborate render of the whole DEMs error. In a word, assessing the accuracy of DEMs rightly is of momentous practical significance and theoreti-cal value. We must adopt truncation error to assess the accuracy of DEM abandoning RMSE. Comparison of the original topographic map with the map generated from the DEM is the most effective method to check the quality of DEMs. |
来源
|
地球信息科学
,2003,5(3):64-70 【扩展库】
|
关键词
|
DEM
;
DEM精度
;
误差分析
;
逼近误差
;
精度评价
|
地址
|
武汉大学资源与环境科学学院, 武汉, 430079
|
语种
|
中文 |
文献类型
|
研究性论文 |
ISSN
|
1560-8999 |
学科
|
测绘学 |
文献收藏号
|
CSCD:1251969
|