Rapid Acquisition of Site Specific Lime Requirement with Mid-Infrared Spectroscopy
M. Leenen, S. Pätzold, T. Heggemann, G. Welp
Institute of Crop Science and Resource Conservation, Division Soil Science, University of Bonn, Nussallee 13, 53115 Bonn, Germany
In Germany, the lime requirement of arable topsoils is derived from the organic matter content, clay content, and pH(CaCl2). For this purpose, it is common practice to determine the lime requirement of a field size up to three hectares from only one composite soil sample, whereby site heterogeneity is regularly not taken into account.
To consider site heterogeneity, a measurement technique is required which allows a rapid and high resolution data acquisition. Mid-infrared spectroscopy (MIRS) is a cost-effective, high throughput technique and in combination with partial least squares regression (PLSR) an accurate method for prediction of a large number of soil properties. All input variables for lime requirement estimation can potentially be predicted by MIRS.
The objectives of our study were (i) to test the performance of separate organic matter, clay, and pH(CaCl2) prediction for lime requirement derivation, and (ii) to calibrate prediction models for direct lime requirement recommendations. However, on many fields with partially calcareous parent material, topsoils reveal a small scale pattern of non-calcareous areas and calcareous zones; the latter should not be limed. Consequently, these zones have to be identified. Thus, another objective of the study was, (iii), to find a spectral index to distinguish between non-calcareous and calcareous soil samples.
We surveyed seven study sites across Germany and a total of 458 soil samples in the frame of the Bonares-I4S project funded by the German Federal Ministry of Education and Research. The topsoil of four study sites was partially calcareous. The prediction models were calibrated by using PLSR. To identify calcareous soil samples, we used the spectral bands for CaCO3 at 2646 - 2423 cm-1, 1833-1782 cm-1, and 930-730 cm-1.
Compared with the conventional method, single parameter prediction and subsequent derivation of lime requirement yielded satisfactory results (R² up to 0.93; RMSEP = 0.66 – 1.01 t CaO ha-1). We successfully calibrated local PLSR models for direct lime requirement prediction (R² = 0.56 - 0.76; RMSECV = 0.66 – 1.42 t CaO ha-1; RPD = 1.44 - 2.03). The difference of absorption between wavenumbers 2513 and 2439 cm-1 was highly correlated with the CaCO3 content (R² = 0.94, all sites) and enabled us to identify calcareous samples having no lime requirement.
We conclude that MIRS is a promising tool for the determination of site specific lime requirement as basic information for variable rate lime application.
