Estimation of wet aggregation indices using soil properties and diffuse reflectance near infrared spectroscopy: an application of classification and regression tree analysis
Soil aggregation is critical for assessing soil health; however, conventional aggregationmeasurement is laborious and expensive. The performance of near infrared diffusereflectance spectroscopy (NIR) and basic soil properties for estimation of wet aggregationindices was investigated. Two samples sets representing different soils from across LakeVictoria Basin in Kenya were used for the study. A model calibration set (n ¼ 136) wasobtained following a conditioned Latin hypercube sampling, and validation set (n ¼ 120)using a spatially stratified random sampling strategy. Spectral measurements wereobtained for air-dried (<2 mm) soil using a Fourier-transform NIR spectrometer. Soillaboratory reference data were also obtained for wet aggregation indices (WSA): macro,micro and unstable fractions using two different wet-sieving pretreatments. Soil propertieswere screened as candidate predictors of WSA using Classification and Regression Tree(CART regression) analysis. WSA were calibrated to soil predictors and to smoothed firstderivative NIR spectra using partial least squares (PLS) regression. Key soil predictors were:soil organic carbon and pH water (macro), water dispersible clay (WDC) (micro) andexchangeable sodium (unstable). Full cross validation of NIR PLS prediction of stablemacro, micro, unstable aggregates, and for WDC gave RPD (ratio of prediction deviation) of1.4e2.0. Independent testing of NIR PLS gave RPD ¼ 1.4 for macro and RPD ¼ 1.2e1.0 forunstable and soil predictors. NIR could estimate macro and unstable fractions withmoderate reliability, and; NIR was superior over soil properties for stability pedotransferpurposes. Further efforts should widely test performance for a wider range of soil types andcalibration strategies for improved geographic transferability of models.