| Autor: |
Wangeci, Alex, Knadel, Maria, De Pascale, Olga, Greve, Mogens H., Senesi, Giorgio S. |
| Předmět: |
|
| Zdroj: |
JAAS (Journal of Analytical Atomic Spectrometry); Nov2024, Vol. 39 Issue 11, p2903-2916, 14p |
| Abstrakt: |
Laser-induced breakdown spectroscopy (LIBS) has contributed to the advanced and rapid determination of soil properties including soil organic carbon (SOC) and texture. Recent developments of commercial handheld LIBS (hLIBS) have allowed the use of the technique directly in the field. However, to date, the performance of hLIBS on different types of soils covering wide geographical distributions has not been evaluated. In this study, a total of 305 soil samples covering a continental scale were used to assess the repeatability and reproducibility of LIBS data acquired using a commercially available hLIBS instrument. Furthermore, the performance of the prediction models for SOC and texture was evaluated based on the prediction error. The repeatability and reproducibility of LIBS data were evaluated based on the relative standard deviation (RSD) for measurements performed under similar and different environmental conditions (temperature and humidity). First, the RSD of the signal ratios and the predicted values for soil properties under investigation were calculated. Then, the prediction accuracy of the various soil properties was compared based on the standardized root mean error of prediction (SRMSEP) and the ratio of performance to interquartile distance (RPIQ). The signal ratios assessed using the C, Si, Ca, and K LIBS emission lines achieved a repeatability of 4–9% and a reproducibility of 7–10%, whereas the repeatability and reproducibility for predicting SOC and texture were <25%. The prediction of sand content exhibited the lowest error (SRMSEP = 0.14) followed by clay and silt (SRMSEP = 0.15), and then SOC (SRMSEP = 0.16). The results of this work underscore the promising potential of hLIBS for large-scale SOC and texture determination, with the opportunity to enhance the prediction accuracy by integrating soil mineralogy information for soil classification before applying the modeling process. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
