The Positive and Negative Terrain of Loess Plateau Extraction Based on the Multi-azimuth DEM Shaded Relief
Received date: 2011-02-20
Request revised date: 2011-04-11
Online published: 2012-01-20
Copyright
Based on the DEM data of Jiuyuangou wateshed in Suide County, Shaanxi of China, with a spatial resolution of 5 m, employing the digital terrain analysis, multivariate statistics and data mining methods, multiple indexes of multi-azimuth DEM shaded relief and slope are established. The solutions of extraction positive and negative terrain of loess plateau by Principle Components Analysis, and Logistic regression model is proposed. The result indicates that: the extracting model has better consistency and accuracy, of which the accuracy is 82.1% and Kappa statistics is 0.6298. Tested on 6 samples of different valley by this model, the mean accuracy of positive and negative terrain are at 77.6% and 84.9%, and weighed mean accuracy is 81.3%. It is suitable to extract positive and negative terrain of loess plateau by PCA and Logistic regression model.
CHEN Yong-gang , TANG Guo-an , ZHOU Yi , LI Fa-yuan , YAN Shi-jiang , ZHANG Lei . The Positive and Negative Terrain of Loess Plateau Extraction Based on the Multi-azimuth DEM Shaded Relief[J]. SCIENTIA GEOGRAPHICA SINICA, 2012 , 32(1) : 105 -109 . DOI: 10.13249/j.cnki.sgs.2012.01.105
Fig.1 Sketch of tudy area图1 研究区 |
Fig. 2 Principal component screeplot (a) and loading (b)图2 主成分碎石(a)、载荷(b) |
Table 1 Eigenvalue and contribution rate表1 特征值与贡献率 |
项目 | 1 | 2 | 3 | 4 | 5 | … | 17 | 18 | 19 |
---|---|---|---|---|---|---|---|---|---|
特征值 | 7.6459 | 6.5736 | 3.8480 | 0.8922 | 0.0395 | … | 0.0000 | 0.0000 | 0.0000 |
贡献率 | 0.4020 | 0.3460 | 0.2030 | 0.0470 | 0.0020 | … | 0.0000 | 0.0000 | 0.0000 |
累计贡献率 | 0.4020 | 0.7480 | 0.9510 | 0.9980 | 1.0000 | … | 1.0000 | 1.0000 | 1.0000 |
Table 2 Logistic regression coefficients and significance test表2 Logistic回归系数及显著性检验 |
变量名 | 系数 | 标准差 | Z值 | 概率P |
---|---|---|---|---|
常数项 | 0.5471 | 0.0136 | 40.240 | 0.000 |
Y1 | 0.5950 | 0.0063 | 94.300 | 0.000 |
Y2 | 0.0802 | 0.0051 | 15.680 | 0.000 |
Y3 | 0.0842 | 0.0066 | 12.680 | 0.000 |
Y4 | 1.6539 | 0.0178 | 92.920 | 0.000 |
Table 3 Model accuracy evaluation表3 模型精度评价 |
项 目 | TP | FP | Precision | Recall | F-Measure | ROC |
---|---|---|---|---|---|---|
正地形 | 0.868 | 0.244 | 0.832 | 0.868 | 0.85 | 0.897 |
负地形 | 0.756 | 0.132 | 0.805 | 0.756 | 0.78 | 0.897 |
加权平均值 | 0.821 | 0.197 | 0.821 | 0.821 | 0.82 | 0.897 |
Table 4 Accuracy of testing samples表4 测试样区精度 |
测试样区 | 正地形精度 | 负地形精度 | 加权平均精度 |
---|---|---|---|
1 | 0.788 | 0.855 | 0.820 |
2 | 0.809 | 0.809 | 0.809 |
3 | 0.703 | 0.725 | 0.713 |
4 | 0.748 | 0.951 | 0.862 |
5 | 0.819 | 0.882 | 0.848 |
6 | 0.787 | 0.872 | 0.826 |
平均值 | 0.776 | 0.849 | 0.813 |
Fig.3 Comparison of study area图3 研究样区对比 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
/
〈 |
|
〉 |