Uncertainty and sensitivity analyses of a model of dissolved organic carbon in a lake
(English)Manuscript (preprint) (Other academic)
Uncertainty and sensitivity analyses have been performed on a lake model describing the dissolved organic carbon (DOC) turnover in a humic lake. The DOC model is based on a Lagrangian dispersion model where trajectories of fluid particles, representing DOC containing water parcels, are simulated. Thus the history of the individual parcel influences its present degradation and DOC concentration. Together the parcels describe the DOC concentration of the lake. A Monte Carlo simulation based on a random sampling of DOC input parameters was used for the analyses. The results were studied one day a monthbetween May 1993 and December 1997. The mean DOC concentration (9-13 mg C L"1) has a range of about 2 mg C L-1 and a coefficient of variation (CV) of 2-4%. That is rather large, but still less than the variability of the input parameters. The daily mean degradation variables (bacterial degradation and photooxidation) had larger variability (CV approx. 20%). Different methods were combined to give multiple views of the sensitivity: scatter plots, the chosen parameters in stepwise regression, and the parameters in multiple linear regression with high partial correlation coefficient and/or high standardised coefficient. In addition, both ordinary and rank based multiple linear regression were used. The combined methods show that the mean DOC concentration in the lake is most sensitive to primary production and bacterial degradation rate, while bacterial degradation and photooxidation are most sensitive to bacterialdegradation rate and the attenuation rate of light respectively. Both seasonal and annual variations in sensitivity indices were found for mean DOC concentration revealing a flow dependence of the crucial parameters. Overall, the DOC model would benefit most from a better estimation of bacterial degradation rate, a better model for primary production, and a better determination of DOC concentration of inflow.
uncertainty analysis, sensitivity analysis, Monte Carlo simulation, lake model, dissolved organic carbon
Social Sciences Interdisciplinary
IdentifiersURN: urn:nbn:se:liu:diva-79297OAI: oai:DiVA.org:liu-79297DiVA: diva2:540287
Short title: Sensitivity of a Doc Model2012-07-092012-07-092012-07-09Bibliographically approved