Use of water quality model uncertainty analysis to develop sampling design criteria for in-stream carbon

The present research is motivated by the need to include recognized uncertainty within water quality sampling methods performed in agricultural and urban watersheds. We use uncertainty analysis of water quality model results to help define a sampling method for particulate organic carbon in disturbed watershed systems. Low-gradient watersheds with agriculture and urban land use disturbances receive a significant portion of organic matter from in-stream benthic carbon production in which coupled physical and biological processes govern the degree of organic carbon accrual. A lack of uniformity in methodological approaches and temporal domains has driven the need to develop an appropriate and repeatable sampling strategy to capture the range and distribution of transported organic carbon. Therefore for this study a conceptual model for sediment carbon fate and transport was applied to a lowland watershed system with pronounced fluvial storage and agriculturally and urban disturbed lands. A suite of sampling routines was implemented to test the sensitivity of the transported carbon (CT) distribution, aiming to isolate the importance of physical and biological processes. Results of the study highlight CT followed a Gamma distribution, with a root mean square error approximation of 0.066. Low flow sampling routines performed as well as routines including high and low flows, contradicting previous studies that emphasize capturing high flow events for CT variability. Likewise, monthly and bimonthly sampling routines were as adequate as a weekly or bimonthly routine. Although the two-year routine did well at capturing variability, the equality of the central measure of tendency was not statistically significant, requiring us to recommend a three-plus-year study if feasible, especially if consecutive years have similar hydrologic patterns.