| 摘要: |
| 遥感技术是快速而又准确获取水体信息的有效途径，对水资源管理与评估有着重要意义。以福州市闽江、贵阳市百花水库、曲靖市南盘江和泰安市东平湖区域Landsat 8影像为基础，构建改进的自动水体提取指数（MAWEI）对水体信息进行提取，利用目视解译结果作为精度验证数据，量化不同水体指数在多种环境条件下的水体提取精度。研究结果表明：（1）面向对象的分类方法改善了面向像素方法存在“椒盐现象”的缺点；（2）MAWEI、AWEInsh、AWEIsh和MNDWI四种水体指数在最优提取阈值的情况下，MAWEI指数提取水体的精度最高，效果最佳。且通过实验发现不同环境条件下MAWEI水体指数的稳定性较好。MAWEI指数可用于快速有效地提取大面积水体信息，能满足大面积的水体快速制图和土地利用的分类，对水体信息快速提取具有较强的应用价值，值得推广。 |
| 关键词: 水体指数 遥感影像 水体提取 精度评价 |
| DOI：10.7515/JEE182080 |
| CSTR： |
| 分类号: |
| 基金项目:福建省自然科学基金项目（2014J01149）；福建省公益基金项目（2017R1034-4，2013R02） |
|
| Research on water information extraction based on MAWEI index |
|
NIE Yawen, YU Ming, LAN Ting
|
|
1. College of Geographical Science, Fujian Normal University, Fuzhou 350007, China
2. Institute of Geography, Fujian Normal University, Fuzhou 350007, China
|
| Abstract: |
| Background, aim, and scope Remote sensing technology is an effective way to obtain water information quickly and accurately. It is of great significance for water resources management and evaluation. Aiming at improving AWEI water body extraction index, this paper proposes MAWEI method. The research content is to compare MAWEI method with other methods to verify its effectiveness, so as to provide a reference method for effectively extracting large area watershed water body. Materials and methods Based on Landsat 8 images of Minjiang River in Fuzhou City, Baihua Reservoir in Guiyang City, Nanpanjiang River in Qujing City and Dongping Lake in Tai’an City, this paper constructs an Modified Automatic Water Extraction Index (MAWEI) to extract water information. And the visual interpretation results are used as the accuracy verification data to quantify the water extraction accuracy of different water body indices. Results (1) It is found that e-Cognition can improve the “salt and pepper phenomenon” in the water extraction by ENVI. (2) Under the optimal extraction threshold, compared with AWEInsh, AWEIsh and NDWI, MAWEI index has the highest accuracy and the best effect. The stability of MAWEI water body index is better under different environmental conditions. Discussion All four methods extract the main contours of water body information. In these four research areas, the MAWEI method has relatively less errors and relatively high extraction accuracy. Conclusions MAWEI index can be used to extract large area water body information quickly and effectively. It can satisfy large area water body rapid mapping and land use classification. It has a strong application value for water body information fast extraction. Recommendations and perspectives The MAWEI index can quickly and effectively extract large-scale water body information with good results and high precision, and can meet the requirements for rapid mapping of large-scale water bodies and classification of land use. However, No water extraction method can extract all water body information completely and ensure that no other species mix in. A variety of methods can be synthesized to remove specific ground disturbance factors, such as shadows, buildings and so on. |
| Key words: water index remote sensing image water extraction accuracy evaluation |
