基于电磁感应仪的土壤盐渍化剖面特征解译研究

Interpretation of Soil Profile Salinization Characteristics Based on Electromagnetic Inductor

为了快速、精准解译区域尺度土壤盐分特征,有必要建立土壤剖面盐分信息精确解译模型。以新疆农灌区不同土壤质地的盐渍化土壤为研究对象,利用电磁感应式大地电导率仪EM38获取土壤表观电导率,构建了基于EM38的两种土壤质地剖面分层盐分解译模型,并对盐分解译回归模型进行了精度检验。结果表明:两种质地的土壤剖面盐分含量变异均较大,中壤土各层电导率变异系数在58%~98%,呈现中等变异强度,而砂土表层(0~40 cm)变异系数达到100%以上,属于强变异强度,深层土壤变异系数介于76%~88%属于中度变异强度。两种质地土壤电导率与磁感表观电导率EMh、EMv间呈显著的相关关系,水平和垂直测量模式都能够对不同深度土盐层分进行预测,且以EMh+EMv为自变量的二元回归解译模型具有较高的精度,相关系数R达到0.94以上。中壤土EM38的盐分预测在不同深度土层的验证结果决定系数达到0.59以上,砂土质地土壤盐分预测验证R2达到0.36以上,预测精度与土壤剖面盐分变异性呈现显著负相关,其相关系数为-0.86,中壤土质地的解译效果优于砂土质地。分析EM38在预测不同土壤质地盐分精度上的差异性,构建了电磁感应式土壤剖面盐分含量的预测模型。研究结果引入土壤质地变量,可为大面积土壤盐渍化的快速精确测定提供理论依据。

It is necessary that establishing interpreting model of soil salinity in different soil layers in order to quickly and accurately interpret soil salinity in the regional scale. Taking salinized soil with different textures in the same agriculture irrigation district as research object in Xinjiang, electromagnetic inductor （EM38） was applied to measure soil apparent electrical conductivity （EC）, its model was built at different layers of two soil textures based on electromagnetic induction technique, and the accuracy of regression model of soil salinity was be validated. The results indicated that soil salinity at two soil texture layers showed significant spatial variability. Soil EC appeared intermediate variability with coefficient of variation （C.V.） of 58% - 98% at loam soil layers and strong variability with that of exceeded 100% at 0 - 40 cm layers and intermediate variability with that of 76% - 88% at 40 - 100 cm layers of sandy soil. Soil electromagnetic conductivity （EMh, EMv） were significant correlated with EC in two soil textures. The binary regression model, taking EMh and EMv as the independent variables, was with high accuracy and with C.V. value of more than 0.94, which could be used to prediction soil salinity in different horizontal and vertical soil layers. The coefficient of determination was more than 0.59 in loam and more than 0.36 in sandy soil layers based on the validation result of this prediction model. There was a significant negative correlation between prediction accuracy and spatial variability of soil salinity, The greater spatial variance, the worse prediction accuracy. The interpretation effect was better in loam than in sandy soil, The research introduced the variable of soil texture, which would provide theoretical foundation for accurately and rapidly testing soil salinization at most regions.