Sampling reproducibility and error estimation in near infrared calibrationof lake sediments for water quality monitoring

This study forms part of a wider project designed to develop methods for ro utine lake monitoring using near infrared (NIR) spectrometry of surface sed iment samples. During calibration, linear relationships (y = Xb + f) betwee n water chemistry variables (y) and the NIR spectra (X) were evaluated by r egression analysis. The principal objectives of this study were to investig ate sources of error, both in the X-data (i.e. the NIR spectra), due to nat ural variation, sediment sampling, subsequent sample handling and measureme nts and also in the estimation of y-data, here measured lake-water pH value s for use in calibration. The error in the NIR spectral data was investigat ed in two different ways. First, lake-water pH was predicted by a PLS model derived from triplicate lake sediment spectra, and an ANOVA was carried ou t on the predicted pH. Using this strategy, the within-lake variance of NIR -predicted pH of each lake was found to be significantly lower than the bet ween-lake variance at the p = 0.01 confidence level. In an alternative appr oach, lakes which were very similar, according to principal component analy sis (PCA) score plots, were selected and PLS-DA (Partial Least Squares-Disc riminant Analysis) was used to show that the triplicate sediment spectra fr om each lake were clearly resolved from spectra of other lakes. For 33 lake s, pH measurements of their waters allowed estimation of an arithmetic mean and variance in the y-data. This variance was pooled over all the lakes an d compared to the total variance in the y-variable. For pH, the temporal wi thin-lake variability, pooled over all lakes, accounted for only 1.7% of th e between-lake variability Thus, the sampling strategy and temporal resolut ion of measured lake-water pH allow accurate estimates of lake-water pH fro m NIR spectra.