Inferring lake water chemistry from filtered seston using NIR spectrometry

Near-infrared spectrometry (NIR) is a rapid, inexpensive and reagent-free t echnique, widely used in industry in areas such as quality control and proc ess management. The technique has great potential for environmental monitor ing of aqueous systems. This study assesses relationships, using PLS regres sion, between NIR spectra of seston collected on glass fibre filters and th e following measured lake water parameters: total organic carbon (TOC), tot al phosphorus (TP), Abs420 and pH. Water samples were collected from 271 ol igotrophic lakes during autumn 1995. The predictive model for TOC explained 68% of the variance (SEP=2.1 mg L-1. range 14.9 mg L-1), and that for colo ur 71% (SEP=0.04 A, range 0.36 A), while the explained variances for pH and TP were 72% (SEP=0.36, mu g L-1 range 3.13 mu g L-1) and 45% (SEP =4 eta g L-1, range 41 mu g L-1), respectively. A model correlating NIR spectra and the actual amount of phosphorus in the seston captured on filters explaine d 86% of the variance (SEP = 0.044 mu g/filter, range 0.47). Several pretre atments and regression techniques were used in an attempt to enhance modeli ng performance. However, straightforward PLS on raw data performed best in all cases. (C) 2000 Elsevier Science Ltd. All rights reserved.