摘要
鄱阳湖水体范围参数对揭示该流域生态环境演变和旱灾形成机制具有重要意义。鄱阳湖枯水期地物类型复杂且分布不连续,水体微观构成机制复杂多样,导致水体及其邻界地物光谱特征复杂,现有卫星影像水体提取方法效率和精度难以适用景观异质性强的水体范围提取。基于环境减灾卫星多光谱数据,协同使用阈值法、决策树法和区域生长法开展卫星影像鄱阳湖枯水期水体提取,并以人工目视解译结果作为真值,开展精度验证和效率评价。结果表明:该方法可将目视解译需要2 d的时间缩短到2 h以内,显著提高了效率,正确识别率为目视解译的95.8%。
The water extent parameter of Poyang Lake is significant to reveal the ecological environment evolution and drought formation mechanism of the basin. In dry season,the terrain features of Poyang Lake are complicated and discontinuous,the microscopic formation mechanism of water body are highly complicated and diverse,so the spectral features of water body and adjacent terrain are complicated and the existing water body extraction approaches are difficult to meet the extraction efficiency and accuracy requirements of water body with highly landscape heterogeneity. Decision tree method,threshold method and region growing method were combined to perform water extraction by using HJ-1 data. Then,the accuracy validation and efficiency evaluation were performed by using the manual interpretation results as reference. The results indicate that the method significantly improved the efficiency,only 2 hours required instead of 2 days by using traditional visual interpretation method. The correct recognition accuracy of the method reaches 95.8% of the visual interpretation.
出处
《人民长江》
北大核心
2016年第17期1-5,共5页
Yangtze River
基金
国家自然科学基金项目"冬小麦旱灾致灾因子危险性高光谱遥感定量诊断研究"(41401605)
