摘要
为提高潍坊北部地区土壤氯化钠含量的监测精度,研究土壤氯化钠含量的高光谱估算方法,本研究利用光谱测量技术,以潍坊市北部海咸水入侵典型区域作为研究区,野外采集41个土壤光谱样本,并测定理化性质,对实测光谱反射率进行数学变换,与土壤样本氯化钠含量进行相关性分析,利用偏最小二乘回归分析法建立土壤氯化钠含量监测模型。研究结果表明:反射率一阶微分、二阶微分土壤氯化钠含量监测模型的决定系数均高于0.8,建模精度较高。经检验,反射率二阶微分是预测土壤氯化钠含量的最佳光谱指标,所建模型为Y=-828.278x6-635.832x7-348.817x2-583.591x9+360.329x4+1 044.751,建模决定系数为0.859,总均方根误差为0.274。说明将实测反射率光谱变换后利用偏最小二乘回归法建立的潍坊北部地区土壤氯化钠含量估算模型具有较高的可行性,能够较好地反演研究区土壤氯化钠含量,可为潍坊北部地区土壤盐渍化监测和评价提供参考。
To develop an approach to estimating the soil sodium chloride content, spectral measurement technique was applied to 41 soil samples collected from the area eroded by northern seawater of Weifang,Shandong province. The correlation between soil sodium chloride and methematically converted spectral reflectance was analyzed,and a model monitoring soil sodium chloride content was developed using partial least square regression. The models of first derivative reflectance and the second derivative reflectance showed high accuracy of soil sodium chloride content,with both determination coefficients higher than 0. 8,in which the first derivative reflectance was better due to lower errors in inverting the content.It was indicated that the model developed in this study using partial least square regression analysis after conversion of spectral reflectance performed well in estimating the soil sodium chloride content.
出处
《江苏农业学报》
CSCD
北大核心
2016年第4期817-823,共7页
Jiangsu Journal of Agricultural Sciences
基金
国家自然科学基金项目(41371395)
山东省国土资源厅项目[鲁勘字(2011)14号]
山东省水利厅项目[鲁水财字(2012)49号-2]
关键词
高光谱技术
氯化钠含量
反射率
偏最小二乘回归法
hyperspectral technique
sodium chloride content
partial least square regression
reflectance
