基于TM数据的敦煌绿洲土壤盐碱化信息提取研究

干旱、半干旱地区的绿洲盐碱化威胁着生态环境.随着空间信息技术的发展,遥感已被广泛应用于土壤盐碱化监测与制图.当前基于遥感技术的盐碱化信息提取方法主要归为基于图像光谱的直接提取方法和基于辅助数据的间接提取方法两大类.间接提取方法精度相对较高,但在缺乏辅助信息时并不可行,此时提高基于图像光谱特征的直接提取精度就尤为重要.本研究提出了一种优化波段组合的盐碱地提取方法,并引入了热红外波段,提高了直接提取方法的精度.以敦煌绿洲为研究区域,基于Landsat TM遥感图像中盐碱地的光谱特征,采用最佳指数因子确定盐碱化信息提取的最佳波段,结合图像变换及数据融合等图像处理方法,利用最大似然分类器进行了图像分类,提取出了敦煌绿洲盐碱地的分布区域.将不同波段组合、不同图像处理方法的分类精度检验结果进行比较得知,陆地卫星TM3,5,6和7波段组合的总体分类精度和盐碱地提取精度最高,增加热红外波段可使总体分类精度提高9%,盐碱地提取用户精度改善最高可达18%.

Soil Salinization Extraction Research in Dunhuang Oasis Based on TM Data

Soil salinization is a threat to the ecological environment of the Oasis in arid and semi-arid region. With the rapid development of space information technology, remote sensing has been widely used in the monitoring and cartography of soil salinization. The current methods of soil salinization extraction are generally concluded into two categories, direct extraction method based on remote sensing image spectra and indirect extraction method based on auxiliary data. Generally, the indirect methods are of higher precision compared with the direct methods. However, the indirect method is infeasible without auxiliary data. In this circumstance, improving the precision of the direct method is significantly important. In this work, an improved direct extraction method based on the optimal band combination is proposed. The use of thermal band greatly improves the precision. This work took Dunhuang Oasis as the research area, band combinations for soil salinization using OIF （Optimum Index Factor） based on Landsat TM spectra of saline-alkali soil were compared and optimized. After the image transformation and the combination of thermal band, the processed optimized band combination image was classified with maximum likelihood algorithm, and finally the distribution of saline-alkali soil was extracted. The precision validation reveals that, first, band combination of TM 3, 5, 6 and 7 is of the best performance both in classification and extraction; second, the introduction of thermal band improves the overall accuracy by 9% and the user accuracy of soil salinization by up to 18%.