Estimating soil texture from vis–NIR spectra

Abstract

Summary Vis–NIR spectroscopy is a low‐cost method for proximal soil sensing, enabling rapid analysis of soil texture as an alternative to more laborious analytical methods. In this study we used partial least squares regression (PLSR) and random forest (RF) models trained on vis–NIR spectra of 173 soil samples from across Germany to estimate soil sand (63–2000 µm), silt (2–63 µm) and clay (< 2 µm) contents. Models were trained with different spectral processing methods. Texture was also estimated by averaging results across different models and by calculating each fraction as the difference from 100% of the sum of the other fractions. The PLSR models predicted clay and sand best, whereas RF performed better for predicting silt fractions. Spectral processing did not improve clay predictions, but standardized normal variate spectra predicted the silt fraction best, and log‐inverse spectra improved modelling of the sand fraction. The best models explained > 90% of variance in the evaluation samples of the textural fractions. Residual prediction deviations were > 3 for all fractions, which together with good accuracy indicated excellent model performance. Model averaging across the top three performing models improved the predictive performance of all fractions. Calculating each fraction from the sum of the other two fractions predicted only the sand fraction well, but increased model bias. In addition, we evaluated the models for their reliability at predicting the cumulative sum of the three fractions for samples of unknown textural composition (n = 1186). In a novel approach, we identified poorly predicted samples by propagating errors from the model uncertainties for the individual soil fractions, namely those samples whose cumulative sum lies outside the range covered by 100% ± propagated error from the three fractions. This new method provides the opportunity to optimize analyses efficiently because these samples can be prioritized and used to check and update the models. Highlights Partial least squares regression and random forest models of vis–NIR spectra predict soil texture reliably. Cumulative sums of independently predicted fractions enabled error estimation. Error propagation allowed robust classification of the whole soil texture. Error propagation identified poorly predicted samples, optimizing laboratory resources.

Iron isotope fractionation in soil and graminaceous crops after 100 years of liming in the long‐term agricultural experimental site at Berlin‐Dahlem, Germany

“…Only two plots (+L2 and +L3) were alike in their Fe concentrations at every depth. The smallest Fe concentrations below 40 cm were found in plot −L2, where the subsoil contained more silt and less clay than in the other plots (Hobley & Prater, 2019). Correspondingly, total Fe stocks varied in the subsoil, with the mean values of 1.86 × 10 5 (standard error ( SE ) 0.33 × 10 5 ) and 1.76 × 10 5 ( SE 0.38 × 10 5 ) kg ha −1 for limed and non‐limed plots (Figure 2), respectively, without significant difference between the two treatments.…”

“…Only two plots (+L2 and +L3) were alike in their Fe concentrations at every depth. The smallest Fe concentrations below 40 cm were found in plot −L2, where the subsoil contained more silt and less clay than in the other plots (Hobley & Prater, 2019).…”

A century of liming affects the Mg isotopic composition of the soil and crops in a long‐term agricultural field at Berlin‐Dahlem, Germany

“…According to the FAO (IUSS Working Group WRB, 2014), the soil at this experimental site is classified as an Albic Luvisol, developed from glacial till. The soil texture is sandy to loamy sand, dominated by sand particles that comprise 65-75% of soil weight in the topsoil (0-30 cm) and 40-60% in the subsoil (30-100 cm) (Hobley & Prater, 2019). The field site is heterogeneous with regard to textural changes with soil depth from 40 to >100 cm ( Figure S1, Sümer, 2012).…”

Reducing the Moisture Effect and Improving the Prediction of Soil Organic Matter With VIS-NIR Spectroscopy in Black Soil Area

“…It is widely used to quantify physical, chemical, and biological properties of soil [3,6]. Many previous studies have successfully predicted soil texture, colour and mineral composition [7,8,9,10], chemical and biological proper-ties [11,12,13,14,15,16], soil moisture [12,17,18], and SOM [3,19,20]. A lot of absorption bands related to organic molecules are within VIS-NIR range [21], and the reflectance spectrum also carries integrative information of organic and inorganic soil components [12].…”