Measuring soluble salts in soils via portable x-ray fluorescence spectrometry

Abstract

Soil scientists have been using the same quantification methods for soluble salts for decades. Yet they have long struggled with an effective method for quantifying gypsum content, because current methods are fraught with problems. Saline soil has been historically defined as soil containing salts more soluble than gypsum (e.g., various combinations of Na+, Mg2+, Ca2+, K+, Cl-, SO42-, HCO3- and CO32-) to the extent that soil fertility is severely reduced across a wide array of climates and geological settings. Since salinity is not germane to specific soil characteristics such as texture or parent material, it can be challenging to predict salt concentrations within a profile. Given the success of previous studies using portable x-ray fluorescence (PXRF) as a tool for measuring soil characteristics, the evaluation of soluble salts in soil with PXRF seems timely. Not only does this newer technology offer more accurate, quantifiable data to investigators, it produces results in-situ, in seconds. Recent enhancements to PXRF spectrometers have provided better detection limits especially for lighter elements such as S and Cl, a key component of gypsum and other salts. Thus, this research aimed to test the effectiveness of PXRF as a means of directly quantifying gypsum and salinity in soils. A total of 102 soil samples containing a wide variety of gypsum (~2–95%) and 122 samples with various salt concentrations were subjected to both traditional laboratory analysis (thermogravimetry and electrical conductivity, respectively) and elemental analysis via PXRF. Simple and multiple linear regression were used to establish the relationship between the two data sets. Log transformation of some data sets was necessary to normalize the data. Using simple linear regression between laboratory and PXRF data, an R2 of 0.88 was produced for the gypsum data, and an R2 of 0.839 was produced for salinity data. Similarly, simple linear regression for laboratory-quantified gypsum vs. PXRF S produced an R2 of 0.91. Multiple linear regression of laboratory quantified gypsum vs. both PXRF S and Ca produced an R2 of 0.91, and 0.8669 for laboratory determined EC (dS m-1) vs. PXRF Cl, S, Ca, and K. No significant differences were observed between model generation and validation data sets. Overall, PXRF shows great promise for the direct quantification of soluble salts in soils.