“…Dynamic models that are built directly from the information contained within time-series observations (i.e., data-based models without predetermined model structures) have the greatest potential to have parsimonious structures. Only a few dynamic models for the simulation of stream DOC CONC (or DOC LOAD ) over time have been developed, notably the dynamic conceptual models of Hornberger et al, Taugbøl et al, Boyer et al, , Futter et al, , and Ågren et al These models differ from steady-state models of DOC (e.g., Grieve, and Liu et al) in that dynamic models are able to simulate the effects of initial conditions or “memory effects” (i.e., effects of past values of DOC or associated mobilization processes or transport variables) as well as lags to response and exhaustion of stores. Although some of these dynamic models may be described as parsimonious, none have utilized regularly sampled, subdaily data from contiguous storm events or have had their structure (and associated parameters) derived directly from the dynamics contained within observed, subdaily time series of DOC LOAD (i.e., data-based models).…”