A systematic study on the application of scatter-corrective and spectral-derivative preprocessing for multivariate prediction of soil organic carbon by Vis-NIR spectra

被引:202
作者
Dotto, Andre Carnieletto [1 ]
Diniz Dalmolin, Ricardo Simao [1 ]
ten Caten, Alexandre [2 ]
Grunwald, Sabine [3 ]
机构
[1] Univ Fed Santa Maria, Dept Soil, Room 3314,CCR 42,Av Roraima 1000, BR-97105900 Santa Maria, RS, Brazil
[2] Univ Fed Santa Catarina, Dept Biodivers Agr & Forestry, Rod Ulysses Gaboardi,Km 3,Caixa Postal 101, BR-89520000 Curitibanos, SC, Brazil
[3] Univ Florida, Soil & Water Sci Dept, 2181 McCarty Hall,POB 110290, Gainesville, FL 32611 USA
关键词
Visible-near infrared spectroscopy; Modeling; Prediction; Soil property; Pedometrics; PARTIAL LEAST-SQUARES; DIFFUSE-REFLECTANCE SPECTROSCOPY; PRINCIPAL COMPONENTS REGRESSION; NEAR-INFRARED SPECTROSCOPY; ARTIFICIAL NEURAL-NETWORK; RANDOM FORESTS; CLAY CONTENT; MATTER; MODEL; QUALITY;
D O I
10.1016/j.geoderma.2017.11.006
中图分类号
S15 [土壤学];
学科分类号
0903 ; 090301 ;
摘要
Soil organic carbon (SOC) plays a crucial role as an ecosystems indicator. Its quantification requires an affordable, and less time-consuming method. Visible and near infrared (Vis-NIR) reflectance spectroscopy has demonstrated its applicability to predict SOC over the years. The aims of this study were to i) to compare the influence of preprocessing techniques on prediction performance, ii) assess the modeling performance of a wide range of multivariate methods, and iii) evaluate the potential of Vis-NIR spectroscopy to predict SOC. Soil sampling was conducted over an area of approximately 1800 km(2) in the Southern region of Brazil, where a total of 595 soil samples were collected. Oxisols are predominant in the area following by Entisols and Inceptisols. The seven preprocessing techniques that were employed can be divided into two categories based on their SOC prediction performance: scatter-correction and spectral-derivatives. A total of nine different methods were evaluated to predict SOC from Vis-NIR spectra. The models that use scatter-corrective preprocessing exhibited superior prediction compared to the spectral-derivatives group. In the scatter-correction group, continuum removal was the most suitable preprocessing method for SOC prediction. Except for random forest (RF), all the multivariate methods presented robust predictions. The best fit and highest model accuracy for SOC models in validation mode were achieved when applying the weighted average partial least-squares (WAPLS) method and normalization by range (NBR) preprocessing (R-2 = 0.82, root mean square error = 0.48%, and ratio of the performance to the interquartile range = 3.18). Findings from this systematic methodology study identified the reliability of SOC determinations by examining how preprocessing techniques and multivariate methods affect spectral analyses. It also guides future studies to select the most appropriate methods on similar soils.
引用
收藏
页码:262 / 274
页数:13
相关论文
共 96 条
[1]   The soil management assessment framework: A quantitative soil quality evaluation method [J].
Andrews, SS ;
Karlen, DL ;
Cambardella, CA .
SOIL SCIENCE SOCIETY OF AMERICA JOURNAL, 2004, 68 (06) :1945-1962
[2]  
[Anonymous], 2014, GLOBALSOILMAP BASIS
[3]  
[Anonymous], R PACKAGE VERSION 12
[4]  
[Anonymous], 2012, GEODERMA, DOI DOI 10.1016/j.geoderma.2012.03.011
[5]  
[Anonymous], 1979, THESIS PURDUE U W LA
[6]  
[Anonymous], 2007, Large Scale Kernel Machines
[7]  
[Anonymous], 1980, GEOMORPHOLOGY MINERA
[8]  
[Anonymous], 2001, R NEWS
[9]  
[Anonymous], 2014, 2014 6 WORKSH HYP IM, DOI DOI 10.1109/WHISPERS.2014.8077520
[10]  
[Anonymous], R PACKAGE VERSION 26