Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4573367 | Geoderma | 2014 | 10 Pages |
•Successfully mapped the measurable fractions of carbon related to RothC model•This is the first time these fractions were mapped at catchment scale.•Assessed the uncertainties associated with the mapped fractions•Discuss possibility of using mapped outputs in future process modelling of carbon
This study aims to map the measurable fractions of soil organic carbon related to the RothC carbon model at the catchment scale and to assess the model and prediction quality. It also discusses how the outputs can be used to provide initial pool estimates for process modelling of soil carbon in a spatial context. The study was carried out in Cox's Creek catchment in northern New South Wales, Australia. Samples were collected in 2010 using a design-based sampling scheme. The measurable fractions of the RothC soil carbon model considered in this study were resistant organic carbon, humus organic carbon and particulate organic carbon. It has been reported that these measurable fractions of soil organic carbon can successfully substitute for the conceptual pools of carbon in the RothC soil carbon model. All the samples were scanned to create MIR spectra and recently developed spectroscopic models by Commonwealth Scientific and Industrial Research Organisation (CSIRO) under the national soil carbon research programme (2009–2012) were used to carry out the prediction of respective fractions. We used linear mixed models to create a model for mapping the measurable fractions of soil organic carbon across the catchment. The cross validation results revealed that the highest Lin's concordance correlation between measured and predicted values was recorded for resistant organic carbon (0.78), followed by humus organic carbon (0.74) and particulate organic carbon (0.58). Finally, to assess the uncertainty of the predictions we carried out conditional sequential Gaussian simulations. We demonstrated that measurable fractions of carbon related to the RothC model can be mapped at catchment scale with reasonable accuracy. The derived maps could be used in future studies to initialize the RothC model at any location across the landscape with quantified uncertainties.