Effective agricultural planning requires basic soil information. In recent decades visible near-infrared diffuse reflectance spectroscopy (vis-NIR) has been shown to be a viable alternative for rapidly analysing soil properties. We studied 7172 samples of seven different soil types collected from several regions of Brazil and varying in organic matter (OM) (0.2-10.3%) and clay content (0.2-99.0%). The aim was to explore the possibility of enhancing the performance of vis-NIR data in predicting organic matter and clay content in this library by dividing it into smaller sub-libraries on the basis of their vis-NIR spectra. We used partial least square regression (PLSR) models on the sub-libraries and compared the results with PLSR and two non-linear calibration techniques, boosted regression trees (BT) and support vector machines (SVM) applied to the whole library. The whole library calibrations for clay performed well (ME (modelling efficiency) > 0.82; RMSE (root mean squared error) < 10.9%), reflecting the influence of the direct spectral responses of this property in the vis-NIR range. Calibrations for OM were reasonably good, especially in view of the very small variation in this property (ME > 0.60; RMSE < 0.55%). The best results were, however, found when dividing the large library into smaller subsets by using variation in the mean-normalized or first derivative spectra. This divided the global data set into clusters that were more uniform in mineralogy, regardless of geographical origin, and improved predictive performance. The best clustering method improved the RMSE in the validation to 8.6% clay and 0.47% OM, which corresponds to a 21% and 15% reduction, respectively, as compared with whole library PLSR. For the whole library, SVM performed almost equally well, reducing RMSE to 8.9% clay and 0.48% OM.
SUMMARYAmong the toxic elements, Cd has received considerable attention in view of its association with a number of human health problems. The objectives of this study were to evaluate the Cd availability and accumulation in soil, transfer rate and toxicity in lettuce and rice plants grown in a Cd-contaminated Typic Hapludox. Two simultaneous greenhouse experiments with lettuce and rice test plants were conducted in a randomized complete block design with four replications. The treatments consisted of four Cd rates (CdCl 2 ), 0.0; 1.3; 3.0 and 6.0 mg kg -1 , based on the guidelines recommended by the Environmental Agency of the State of São Paulo, Brazil (Cetesb). Higher Cd rates increased extractable Cd (using Mehlich-3, Mehlich-1 and DTPA chemical extractants) and decreased lettuce and rice dry matter yields. However, no visual toxicity symptoms were observed in plants. Mehlich-1, Mehlich-3 and DTPA extractants were effective in predicting soil Cd availability as well as the Cd concentration and accumulation in plant parts. Cadmium concentration in rice remained below the threshold for human consumption established by Brazilian legislation. On the other hand, lettuce Cd concentration in edible parts exceeded the acceptable limit.Index terms: Lactuca sativa L., Oryza sativa L., soil pollution, chemical extractants, heavy metals, human health.(1) Received for publication in January 2010 approved in december 2010.
ResumoO uso de técnicas de sensoriamento remoto nos estudos de solo é relativamente recente e pode abrir novos campos nesta área. Objetivou-se com este trabalho estimar atributos químicos e físicos de solos da região de Ibaté e São Carlos (SP), por meio de equações de regressão linear e múltipla, geradas a partir de informações eletromagnéticas captadas por sensores instalados em laboratório e satélite. Foi realizada uma análise da viabilidade econômica da utilização de sensores na quantificação de elementos do solo em comparação ao método convencional de análises de solo. Para tanto, foram georreferenciadas e realizadas análises espectrais em laboratório (sensor FieldSpec, 400-2500 nm) e por sensor orbital (ASTER) de 319 amostras de terra. Assim, elaboraram-se modelos de predição para atributos de amostras de terra desconhecidas, nos dois níveis de aquisição. Foi possível quantificar o teor de argila (R² = 0,69) e areia (R² = 0,53) do solo, utilizando-se a energia eletromagnética refletida e mensurada por sensor terrestre. Através do sensor orbital obtiveram-se coeficientes de determinação de 0,45 e 0,41 para areia grossa e areia total respectivamente. Verificaram-se que a análise de atributos do solo por sensores pode ser um método menos oneroso, mais rápido e não poluente, como apoio ao método tradicional de análise de terra.Palavras-chave: sensores; quantificação de atributos; agricultura de precisão.Terrestrial and orbital spectral models for the determination of soil attributes: potential and costs AbstractThe use of remote sensing techniques in soils studies is relatively new. Further research in this area should be made. The objective of this work was to estimate chemical and granulometric attributes of soil from Ibaté and São Carlos counties, State of São Paulo, through multiple and linear regression equations generated with reflected electromagnetic information collect by sensors installed in laboratory and satellite. An analysis of the economic viability of the sensor uses for quantifying soil attributes were carried out in comparison with the conventional method of soil analysis. It was collected and georeferenced 319 soil samples. The soil samples were evaluated by laboratory sensor (FielSpec) and orbital image (ASTER). Spectral prediction models were elaborated for both acquisition levels. The models were used to determine the attributes on unknown samples. It was possible to quantify some soil attributes, such as clay (R² = 0.69) and sand (R² = 0.53) content using the reflected electromagnetic energy measures by terrestrial sensor. The orbital sensor showed good results to predict coarse sand (R² = 0.45) e total sand (R² = 0.41). It was verified that the attribute analysis by sensors can be cheaper and faster to assist the traditional method of soil analysis.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.