Globally, surface and subsurface soil acidity pose a serious threat to crop production and require effective management to prevent yield losses. The application of lime (calcium carbonate, CaCO3) is a well-recognised remediation technique for surface soil acidity. However, for subsurface acidity, improved application and detection methods are required to manage acidity due to the stratification of acidic layers. Mid infrared (MIR) spectroscopy has potential as a rapid, cost effective and accurate method of CaCO3 detection in soils. However, for this technology to be valuable in an agricultural context, there is a need to develop MIR spectroscopy methods for the detection of CaCO3 at agriculturally relevant concentrations of <1%, which was the aim of this study. MIR Partial Least Square Regression (PLSR) carbonate prediction models were developed using carbonate-specific spectral regions to predict CaCO3 concentrations in soil. Prediction models were initially developed using a single soil type spiked with reagent grade CaCO3 and were tested further with soils collected from field trials that had been treated with various rates and sources of lime. Prediction models using specific carbonate peak regions were established and compared with models using the entire MIR spectrum to assess the model performance. Models built using a single peak range at 2560–2460 cm−1 gave the best results (highest R2 values and lowest RMSE values) followed by the single peak range at 1840–1760 cm−1. Specifically, model performance was improved when developed with a single soil type and reagent grade CaCO3 in the 2560–2460 cm−1 spectral region, with PLSR R2 of 0.834 and RMSE of 0.030. Performance was reduced but still satisfactory (R2 of 0.735 and RMSE of 0.069) when applied to field trial soils using a range of different lime sources, including dolomitic lime. The MIR method enabled accurate prediction of carbonate concentrations and had a detection limit of 0.05% CaCO3, similar to what can be achieved using a pressure calcimeter. Findings suggest that MIR spectroscopy measurements of agriculturally applied CaCO3 in soils may provide valuable insights into monitoring the movement and efficacy of lime treatments and will enable more informed lime applications in the future. The method may also be used to detect naturally-occurring carbonates at low concentrations and may be useful for rapid isolation of inorganic carbon from organic carbon in soil carbon measurements and carbon accounting.
Read full abstract