摘要
【目的】探讨南疆盐碱土在不同水、氮、盐条件下的光谱特征,构建适合南疆盐碱土的水、氮、盐反演模型。【方法】以南疆代表性盐碱土为研究对象,设置不同的土壤水、盐和氮量,分析不同处理的土壤光谱特征,采用偏最小二乘回归(PLSR)、支持向量机回归(SVR)和BP神经网络(BPNN)建立土壤水、氮、盐反演模型。【结果】土壤水的特征波段在1900 nm附近,土壤氮的特征波段在1490~1506、1540~2006、2011~2500 nm之间,土壤盐的特征波段在1880~1883、1890~1942 nm之间;PLSR模型对水、氮、盐量的反演效果最好,BPNN模型次之,SVR模型最差。【结论】1900 nm波段是水、氮、盐共同的特征波段,南疆盐碱土水、氮、盐量的最优反演方法为:采用Savitzky-Golay方法进行平滑处理,运用主成分分析降维后通过偏最小二乘回归建立反演模型。
【Objective】Soil nitrogen and water are crucial factors influencing crop growth.Understanding their spatiotemporal variation at large scales is essential for improving agricultural management but challenging.This paper aims to investigate the application of airborne technologies for inversely estimating the spatiotemporal change in nitrogen and water in saline soils.【Method】The research area is located in southern Xinjiang.Remote sensing images were used to analyze the spectral characteristics of saline soils with different water,nitrogen,and salt contents.Inversion models for estimating water,nitrogen and salt contents were developed,using partial least squares regression(PLSR),support vector regression(SVR),and BP neural network(BPNN),respectively.The accuracy of each model was evaluated against ground-truth data.【Result】The characteristic bands of soil water are around 1900 nm,the characteristic bands of soil nitrogen are between 1490~1506,1540~2006,2011~2500 nm,and the characteristic bands of soil salt are between 1880~1883 and 1890~1942 nm.The PLSR model has the best inversion effect on water,nitrogen and salt,followed by BPNN model and SVR model.【Conclusion】The characteristic spectral bands around 1900 nm were sensitive to changes in soil water,nitrogen,and salt content.The optimal inversion model for estimating soil water,nitrogen,and salt involved using the Savitzky-Golay method for smoothing,principal component analysis for dimensionality reduction,and partial least squares regression for developing the inverse model.
作者
赵泽艺
李朝阳
王洪博
张楠
李国辉
唐茂淞
王兴鹏
高阳
ZHAO Zeyi;LI Zhaoyang;WANG Hongbo;ZHANG Nan;LI Guohui;TANG Maosong;WANG Xingpeng;GAO Yang(College of Water Resource and Architecture Engineering,Tarim University,Aral 843300,China;Farmland lrrigation Research lnstitute,Chinese Academy of Agricultural Sciences,Xinxiang 453002,China)
出处
《灌溉排水学报》
CAS
CSCD
北大核心
2023年第7期93-100,共8页
Journal of Irrigation and Drainage
基金
兵团财政科技计划项目(2022BC009)
国家自然科学基金项目(51879267,51669032)
兵团节水灌溉试验计划项目(BTJSSY-202210)
现代农业工程重点实验室2022年度开放课题项目(TDNG2022103)。
关键词
土壤光谱特征
盐碱土
反演模型
土壤含盐量
土壤含氮量
土壤含水率
soil spectral characteristics
saline soils
inversion model
soil salinity
soil nitrogen content
soil moisture
