| 基于决策树方法的干旱区盐渍地信息提取——以渭干河-库车河三角洲绿洲为例 |
| |
| 引用本文： | 何祺胜,塔西甫拉提·特依拜,丁建丽.基于决策树方法的干旱区盐渍地信息提取——以渭干河-库车河三角洲绿洲为例[J].资源科学,2006,28(6):134-140. |
| |
| 作者姓名： | 何祺胜 塔西甫拉提·特依拜 丁建丽 |
| |
| 作者单位： | 1. 新疆大学资源与环境科学学院,乌鲁木齐,830046;新疆大学绿洲生态教育部重点实验室,乌鲁木齐,830046
2. 新疆大学资源与环境科学学院,乌鲁木齐,830046;新疆大学绿洲生态教育部重点实验室,乌鲁木齐,830046;新疆大学理论经济学博士后流动站,乌鲁木齐,830046 |
| |
| 基金项目： | 国家自然科学基金;新疆高校科研项目;教育厅创新研究群体基金 |
| |
| 摘 要： | 以渭干河-库车河三角洲绿洲为例，探讨了干旱区盐生植被红柳覆盖的盐渍地信息的提取方法。利用TM卫星图像数据，分析了研究区主要地物的光谱特征及其波段间的相互运算，从而分析不同地物之间的可分性。研究表明：K-L-3（第三主成分）是提取重度盐渍地信息的最佳波段，TM1是区分红柳覆盖区（轻、中度）盐渍地信息的最佳波段，提取盐渍地信息时混分的水体信息可以通过MNDWI（改进归一化差异水体指数）设定一定的阈值予以剔除，混分的植被信息可以通过NDVI设定一定的阈值予以剔除。根据以上分析，建立决策树模型，在各节点设计不同的分类器，最后得到盐渍地信息的提取结果。然后进行精度评价，结果表明，该方法的总体提取效果较好，是干旱区监测盐渍地变化的有效手段。
|
| 关 键 词： | 遥感 决策树 盐渍地信息提取 NDVI MNDWI K-L变换 |
| 文章编号： | 1007-7588（2006）06-0134-07 |
| 收稿时间： | 2005-12-19 |
| 修稿时间： | 2006-06-21 |
The Extraction of Saline Soil Information in Arid Area based on Decision Tree Algorithm:A Case Study in the Delta Oasis of Weigan and Kuqa Rivers |
| |
| HE Qi-sheng,Tashpolat·Tiyip,DING Jian-li.The Extraction of Saline Soil Information in Arid Area based on Decision Tree Algorithm:A Case Study in the Delta Oasis of Weigan and Kuqa Rivers[J].Resources Science,2006,28(6):134-140. |
| |
| Authors: | HE Qi-sheng Tashpolat·Tiyip DING Jian-li |
| |
| Abstract: | Extraction of saline soil information from remotely sensed images is significant in arid area for surveying the change of saline soil.In this paper,taking the delta oasis of Weigan and Kuqa rivers as the example,TM image collected on Aug 2001 was used.The approach of effective remote-sensing information extraction for saline soil was discussed.The mechanism and characteristics of saline soil and other objects in TM imagery were analyzed to find the possibility of extracting saline soil from the background.With the detailed analysis and clarification of several existing indices,the study selects four indices,the third principal component(K-L-3),the blue wavelength(TM1),the modified normalized difference water index(MNDWI) and the normalized difference vegetation index(NDVI).The research shows that K-L-3 is the best band to extract severe saline soil information,TM1 is the best band to differentiate saline soil information with the salt-tolerant vegetation of Hongliu and the mixed water body and vegetation information can be separated by the index of MNDWI and NDVI.The index of NDVI can accurately differentiate vegetation information from other objects.There is only vegetation whose spectrum lightness value in NDVI band surpasses 0.So it is easy to extract vegetation information from the background.The modified normalized difference water index(MNDWI) is defined by Xu Han-qiu(2005) based on the normalized difference water index(NDWI) of Mcfeeters(1996),which uses MIR(TM5) instead of NIR(TM4) to construct the index.It has been tested that the MNDWI can accurately extract the water body information from the study area more than the NDWI.The paper also analyzed the application of Karhunen-Loeve transformation for extracting severe saline soil information.The study shows that although the first and second principal component account for 98.69% of total information,they cannot effectively extract saline soil information.The third principal component only accounts for 1.09% of total information,but it is the best band for extracting severe saline soil information.The third principal component enhances the information of water body and severe saline soil information.Then,based on the analysis,a simple model of decision tree was applied for extracting saline soil information.Finally,the results were checked by statistical accuracy assessment.The results suggest that the model of decision tree is simple and effective and the precision of this approach is very high,so it is an effective method for monitoring saline soil changes in arid area. |
| |
| Keywords: | Remote Sensing Decision tree Extraction of saline soil information NDVI MNDWI K-L transformation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《资源科学》浏览原始摘要信息 |
| 点击此处可从《资源科学》下载免费的PDF全文 |
|
