干旱区绿洲土壤盐渍化程度遥感定量评价

土壤盐渍化是制约渭干河-库车河三角洲绿洲内部植被生长最主要的生态环境地质问题,也是影响区域农业生产的第一障碍性问题.利用Landsat TM遥感数据,构建了多维向量空间下的11个遥感定量指标.首先,采用因子分析方法分析不同程度的盐渍化与这11个指标之间的相关性,并建立因子得分模型.其次,采用Fisher逐步判别分析法从这些指标中筛选出8个与盐渍化程度密切相关的指标:氧化铁、归一化盐分指数、地表反照率、植被指数、TM1、TM5、地表温度、湿度指标,并对这些指标建立判别方程,然后对渭干河—库车河三角洲绿洲典型样区盐渍化程度进行评价,将所有的实地样本分别代入判别方程,通过分析,发现判别方程的精度较高,与实际情况较吻合.轻度盐渍化的判别精度为80.3%,中度盐渍化的判别精度为68.75%,重度盐渍化的判别精度为83.65%.中度盐渍化土地的判别精度与轻度盐渍化、重度盐渍化土地的判别精度相比较低,这与中度盐渍化的光谱特征与其它地物易混淆有关.总之,所选取的指标简单,易于获取,有利于盐渍化的定量分析与评价.

Qualified evaluating on salinization degree by the remote sensing in the arid oasis

Soil salinization is one of the most important eco-environmental problems constraining the vegetation growth in the Delta Oasis of Weigan and Kuqa Rivers. It is first obstruction for the regional agricultural production. In this paper, under the multi-dimensional vector space, 11 quantitive indices of remote sensing were established by using the spectrum information of Landsat TM. Firstly, It is analyzed that the correlation relationship between the different degree of salinization and 11 indices, and established the score model by adopting the method of factor analysis. Secondly, By adopting the method of stepwise discriminant analysis of Fisher, we chose Fe_2O_3, NDSI, Albedo, NDVI, TM1, TM5, LST, Wet as the closely correlate indices with the degree of salinization. Then, these indices were use to establish a discriminant function for appraising the salinization degree of the study area, and we put these indices into the discriminant function. Through the analysis, it is showed that the average precisions is relatively high, up to 77.57%, and inosculate with the present conditions: light soil salinization discriminant precision is 80.3%, moderate soil salinization discriminant precision is 68.75%, severe soil salinization discriminant precision is 83.65%. Moderate soil salinization discriminant precision is lower than light soil salinization discriminant precision and severe soil salinization discriminant precision, because the moderate soil salinization shape is very abnormity and spectrum characteristic is confused with other ground matters. All in all, the chose indices is simple, easy obtaining, and in favor of quantitative analysis and evaluate for us.