Simonsson, Magnus
- Department of Soil Sciences, Swedish University of Agricultural Sciences
Research article2005Peer reviewed
Simonsson, M; Kaiser, K; Danielsson, R; Andreux, F; Ranger, J
Multivariate regressions on ultraviolet (UV) absorbance (210-300 nm) data were tested as a potential tool to estimate concentrations of total dissolved organic carbon (DOC), DOC in the XAD-8-adsorbable (hydrophobic) fraction (HoDOC), and dissolved nitrate (NO3-N) in forest floor leachates. Partial least square (PLS) regressions were established for 20 zero-tension lysimeter solutions sampled in plots of coniferous and deciduous forest stands from May 2002 to January 2003. These regressions were validated against data from 24 other zero-tension solutions sampled in the same stands. The root mean square error of prediction (RMSEP) was 2.0 mg/l for DOC and 1.6 mg/l for HoDOC. These errors were 39% (DOC) and 20% (HoDOC) smaller than the errors obtained with regressions using only absorbance at 280 nm. For NO3-N, RMSEP was 0.12 mg/l (estimation by single wavelength regression is not possible). DOC and HoDOC covaried strongly and could not be estimated independently based on absorbances; HoDOC varied only between 61% and 81% of the DOC. Partly for this reason, an attempt to estimate the HoDOC/DOC ratio based on specific absorbance spectra (absorbance units per mg/l of total DOC) was little successful. Partly, however, this problem resulted from the fact that both HoDOC and HiDOC fractions had a considerable UV absorptivity, which was not consistent between samples. (C) 2004 Elsevier B.V. All rights reserved
Geoderma
2005, volume: 124, number: 1-2, pages: 157-168
Publisher: ELSEVIER SCIENCE BV
Agricultural Science
https://res.slu.se/id/publ/6563