中国农学通报
中國農學通報
중국농학통보
CHINESE AGRICULTURAL SCIENCE BULLETIN
2013年
23期
146-152
,共7页
于士凯%姚艳敏%王德营%司海青
于士凱%姚豔敏%王德營%司海青
우사개%요염민%왕덕영%사해청
土壤有机质%高光谱%多元回归%光谱预测模型
土壤有機質%高光譜%多元迴歸%光譜預測模型
토양유궤질%고광보%다원회귀%광보예측모형
soil organic matter%hyperspectrum%multiple regression%spectral prediction model
土壤有机质含量是土壤肥力的一个重要指标,利用高光谱对土壤有机质含量进行定量化反演,为精准农业地表土壤有机质含量的快速测定提供参考。利用美国ASD FieldSpec FR地物光谱仪,在室内条件下对经过处理的土壤样品进行光谱测量,通过对土壤样品光谱反射率不同变换形式与有机质含量进行相关性分析,建立土壤光谱变量与土壤有机质含量的多元回归关系模型。结果表明:在波长492 nm、663 nm、1221 nm、1317 nm、1835 nm 和2130 nm 处,采用光谱反射率一阶微分建立的土壤有机质含量反演回归模型,预测精度最好,决定系数R2为0.909。建立的土壤有机质含量高光谱反演模型,可以较好地预测土壤有机质含量,从而为精准农业土壤有机质含量的快速测定提供新的途径。
土壤有機質含量是土壤肥力的一箇重要指標,利用高光譜對土壤有機質含量進行定量化反縯,為精準農業地錶土壤有機質含量的快速測定提供參攷。利用美國ASD FieldSpec FR地物光譜儀,在室內條件下對經過處理的土壤樣品進行光譜測量,通過對土壤樣品光譜反射率不同變換形式與有機質含量進行相關性分析,建立土壤光譜變量與土壤有機質含量的多元迴歸關繫模型。結果錶明:在波長492 nm、663 nm、1221 nm、1317 nm、1835 nm 和2130 nm 處,採用光譜反射率一階微分建立的土壤有機質含量反縯迴歸模型,預測精度最好,決定繫數R2為0.909。建立的土壤有機質含量高光譜反縯模型,可以較好地預測土壤有機質含量,從而為精準農業土壤有機質含量的快速測定提供新的途徑。
토양유궤질함량시토양비력적일개중요지표,이용고광보대토양유궤질함량진행정양화반연,위정준농업지표토양유궤질함량적쾌속측정제공삼고。이용미국ASD FieldSpec FR지물광보의,재실내조건하대경과처리적토양양품진행광보측량,통과대토양양품광보반사솔불동변환형식여유궤질함량진행상관성분석,건립토양광보변량여토양유궤질함량적다원회귀관계모형。결과표명:재파장492 nm、663 nm、1221 nm、1317 nm、1835 nm 화2130 nm 처,채용광보반사솔일계미분건립적토양유궤질함량반연회귀모형,예측정도최호,결정계수R2위0.909。건립적토양유궤질함량고광보반연모형,가이교호지예측토양유궤질함량,종이위정준농업토양유궤질함량적쾌속측정제공신적도경。
The use of hyperspectral can conduct quantitative inversion on soil organic matter content, which is an important indicator of soil fertility, and then provide a reference for the rapid determination of surface soil organic matter content of accurate agricultural. The author conducted the spectral measurements on treated soil samples under laboratory conditions by using the spectroradiometer-U.S.ASD FieldSpec FR, established the multiple regression relationship model between the soil spectral variables and soil organic matter content through the correlation analysis between the different variations of the spectral reflectance of the soil samples and the organic matter content of the soil. The results showed that: the soil organic matter content inversed regression model, which was established by employing the first-order differential spectral reflectance at the wavelength of 492 nm, 663 nm, 1221 nm, 1317 nm, 1835 nm and 2130 nm, possessed the best prediction accuracy, the coefficient of determination R2 was 0.909. The established hyperspectral inversion model of the soil organic matter content could predict the soil organic matter content with the most accurate, and it also provide a new approach for the rapid determination of soil organic matter content of precision agriculture.