Estimation of generalized soil structure index based on differential spectra of different orders by multivariate assessment

扫码查看

原文链接

NETL
NSTL
万方数据

外文摘要：Better soil structure promotes extension of plant roots thereby improving plant growth and yield.Dif-ferences in soil structure can be determined by changes in the three phases of soil,which in turn affect soil function and fertility levels.To compare the quality of soil structure under different conditions,we used Generalized Soil Structure Index(GSSI)as an indicator to determine the relationship between the"input"of soil three phases and the"output"of soil structure.To achieve optimum monitoring of comprehensive indicators,we used Successive Projections Algorithm(SPA)for differential processing based on 0.0-2.0 fractional orders and 3.0-10.0 integer orders and select important wavelengths to process soil spectral data.In addition,we also applied multivariate regression learning models including Gaussian Process Regression(GPR)and Artificial Neural Network(ANN),exploring potential capabilities of hyperspectral in predicting GSSI.The results showed that spectral reflection,mainly contributed by long-wave near-infrared radiation had an inverse relationship with GSSI values.The wavelengths be-tween 404-418 nm and 2193-2400 nm were important GSSI wavelengths in fractional differential spectroscopy data,while those ranging from 543 to 999 nm were important GSSI wavelengths in integer differential spectroscopy data.Also,non-linear models were more accurate than linear models.In addition,wide neural networks were best suited for establishing fractional-order differentiation and second-order differentiation models,while fine Gaussian support vector machines were best suited for establishing first-order differentiation models.In terms of preprocessing,a differential order of 0.9 was found as the best choice.From the results,we propose that when constructing optimal prediction models,it is necessary to consider indicators,differential orders,and model adaptability.Above all,this study provided a new method for an in-depth analyses of generalized soil structure.This also fills the gap limiting the detection of soil three phases structural characteristics and their dynamic changes and provides a technical references for quantitative and rapid evaluation of soil structure,function,and quality.

外文关键词：

Three-phase soilGeneralized soil structure indexHyperspectralDifferential spectrumRegression learning model

作者：

Sha Yang、Zhigang Wang、Chenbo Yang、Chao Wang、Ziyang Wang、Xiaobin Yan、Xingxing Qiao、Meichen Feng、Lujie Xiao、Fahad Shafiq、Wude Yang

展开 >

作者单位：

College of Agriculture,Shanxi Agriculture University,Taigu,China

College of Smart Agriculture,Shanxi Agriculture University,Taigu,China

Department of Botany,Government College University Lahore,Pakistan

基金：

National Natural Science Foundation of ChinaNational Natural Science Foundation of Chinaearmarked fund for Shanxi Province Graduate Education Innovation ProjectProject was also supported by Modern Agroindustry Technology Research SystemScientific and Technological Innovation Fund of Shanxi Agricultural UniversityScientific and Technological Innovation Fund of Shanxi Agricultural UniversityKey Technologies R & D Program of Shanxi ProvinceKey Technologies R & D Program of Shanxi ProvinceNational Key R&D Program of China

项目编号：

31871571313715722022Y3122023CYJSTX02-232018YJ172020BQ32201903D211002201603D31110052019YFC1710800

出版年：

2024

DOI：

10.1016/j.iswcr.2023.08.008

国际水土保持研究(英文)

ISSN：2095-6339

年,卷(期)：2024.12(2)

参考文献量21