The matrix effect is one of the challenges to be overcome for a successful analysis of soil samples using X-ray fluorescence (XRF) sensors. This work aimed at evaluation of a simple modeling approach consisted of Compton normalization (CN) and multivariate regressions (e.g., multiple linear regressions (MLR) and partial least squares regression (PLSR)) to overcome the soil matrix effect, and subsequently improve the prediction accuracy of key soil fertility attributes. A portable XRF was used for analyzing 102 soil samples collected from two agricultural fields with contrasting soil matrices. Using the intensity of emission lines as input, preprocessing methods included with and without the CN. Univariate regression models for the prediction of clay, cation exchange capacity (CEC), and exchangeable (ex-) K and Ca were compared with the corresponding MLR models to assess matrix effect mitigation. The MLR and PLSR models improved the prediction results of the univariate models for both preprocessing methods, proving to be promising strategies for mitigating the matrix effect. In turn, the CN also mitigated part of the matrix effect for ex-K, ex-Ca, and CEC predictions, by improving the predictive performance of these elements when used in univariate and multivariate models. The CN has not improved the prediction accuracy of clay. The prediction performances obtained using MLR and PLSR were comparable for all evaluated attributes. The combined use of CN with multivariate regressions (MLR or PLSR) achieved excellent prediction results for CEC (R² = 0.87), ex-K (R² ≥ 0.94), and ex-Ca (R² ≥ 0.96), whereas clay predictions were comparable with and without CN (0.89 ≤ R² ≤ 0.92). We suggest using multivariate regressions (MLR or PLSR) combined with the CN to remove the soil matrix effects and consequently result in optimal prediction results of the studied key soil fertility attributes. The prediction performance observed for this solution showed comparable results to the approach based on the preprogrammed measurement package tested (Geo Exploration package, Bruker AXS, Madison, WI, USA).

The determination of organic matter in soil and sediment requires several laboratory steps and use of reagents. The objective of this study is to propose an alternative method for organic matter determination in lake sediments samples, by the combination of energy dispersive X-ray fluorescence and partial least squares regression, allowing the quantification of these parameters rapidly and nondestructively. The prediction values of the model presented a relative deviation up to 8%. It could be concluded that multivariate calibration using spectral data from X-ray fluorescence is a fast and feasible technique, with potential to replace the conventional techniques commonly used.

The aim of this study was to present a method for sediment texture characterization combining three spectroscopy techniques with multivariate analysis. Specifically, data were used from flame atomic absorption spectroscopy (FAAS), energy dispersive X-ray fluorescence (EDXRF) and photoacoustic spectroscopy (PAS) measurements in superficial lake sediment samples, which were combined with principal component analysis for exploratory data analysis and with partial least square (PLS) regression for indirect estimative of clay and sand textures . It was possible to infer quantitative values of grain size using the data from the three techniques allied with PLS regression, which allowed to obtain implicit information in each data set. The same classifications in accordance with samples texture were verified for all the applied spectroscopic methods. However, the classification by EDXRF or PAS was simpler, faster, cheaper and non destructive when compared with FAAS. In the case of granulometry quantitative prediction by PLS regression, the results for PAS were less accurate, but were satisfactory for EDXRF and FAAS. In general the results are promising, indicating the method viability, although a larger sampling is necessary to implement the methodology. DOI: 10.5935/1984-6835.20140109O objetivo deste estudo foi apresentar um método para caracterização da textura de sedimentos combinando três técnicas espectroscópicas com analise multivariada. Em particular, foram utilizados dados de medidas de espectroscopia de absorção atômica de chama (FAAS), fluorescência de raios X por dispersão em energia (EDXRF) e espectroscopia fotoacústica (PAS) em amostras de sedimento superficial de lago, os quais foram combinados com o método de análise de componentes principais para a analise exploratória de dados e o método de regressão dos mínimos quadrados parciais (PLS) para a estimativa indireta da textura dos sedimentos. Foi possível inferir valores quantitativos de tamanho de partícula usando dados das três técnicas aliada a regressão por PLS, a qual permitiu obter dados implícitos em cada conjunto de dados. A mesma classificação em função da textura das amostras foi verificada para todas as técnicas espectroscópicas. No entanto, a classificação por EDXRF ou PAS foi mais simples, rápida, barata e não destrutiva quando comparada com a FAAS. No caso da previsão quantitativa da granulometria pela regressão por PLS, os resultados foram menos acurados por PAS, porém foram satisfatórios por EDXRF e FAAS. Em geral, os resultados são promissores indicando a viabilidade do método. Contudo, um maior numero de amostras precisa ser analisado para implementar a metodologia. DOI: 10.5935/1984-6835.2014010