Predictions of Cu, Zn, and Cd Concentrations in Soil Using Portable X-Ray Fluorescence Measurements.

Karl Adler,Mats Söderström,Jan Eriksson,Omran Alshihabi,Kristin Piikki

doi:10.3390/s20020474

Abstract

Portable X-ray fluorescence (PXRF) measurements on 1520 soil samples were used to create national prediction models for copper (Cu), zinc (Zn), and cadmium (Cd) concentrations in agricultural soil. The models were validated at both national and farm scales. Multiple linear regression (MLR), random forest (RF), and multivariate adaptive regression spline (MARS) models were created and compared. National scale cross-validation of the models gave the following R2 values for predictions of Cu (R2 = 0.63), Zn (R2 = 0.92), and Cd (R2 = 0.70) concentrations. Independent validation at the farm scale revealed that Zn predictions were relatively successful regardless of the model used (R2 > 0.90), showing that a simple MLR model can be sufficient for certain predictions. However, predictions at the farm scale revealed that the non-linear models, especially MARS, were more accurate than MLR for Cu (R2 = 0.94) and Cd (R2 = 0.80). These results show that multivariate modelling can compensate for some of the shortcomings of the PXRF device (e.g., high limits of detection for certain elements and some elements not being directly measurable), making PXRF sensors capable of predicting elemental concentrations in soil at comparable levels of accuracy to conventional laboratory analyses.

Highlights

Mapping concentrations of micronutrients or toxic elements in agricultural soil is important but is not commonly done
Descriptive statistics of the elements used to calibrate the multiple linear regression (MLR), random forest (RF), and multivariate adaptive regression splines (MARS) models are presented in were calculated after removal of samples with not a number” (NaN) values in any of the included variables, which resulted in exclusion of 313 samples out of the original 1833 samples (i.e., 1520 samples were used for modelling)
The results in this study demonstrated that an approach based on Portable X-ray fluorescence (PXRF) measurements coupled with machine learning algorithms is capable of predicting concentrations of Cu, Zn, and Cd in non-organic (SOM < 20%; Table 1) Swedish agricultural soils that can be used for risk assessments

Summary

Introduction

Mapping concentrations of micronutrients or toxic elements in agricultural soil is important but is not commonly done. This kind of information could be useful in precision agriculture, where the goal is optimal management in space and time [1]. Excessively high concentrations can be toxic for crops (e.g., an excessive concentration of Cu can lead to malformation of root systems) [3]. Cadmium (Cd) is toxic to consumers of crop products above certain threshold concentrations [2]. It can be useful to map Zn, Cu, and Cd at the field scale in order to rectify deficiencies and toxicities, and to safeguard crop quality and food safety. There are no public field-scale maps of these elements in Sweden

Objectives

Methods

Results

Discussion

Conclusion