We analyzed a five year, high frequency time series generated by an in situ fluorescent dissolved organic matter (fDOM) sensor installed near the Connecticut River’s mouth, investigating high temporal resolution DOM dynamics in a larger watershed and longer time series than previously addressed. We identified a gradient between large, saturating summer fDOM responses to discharge and linear, subdued responses during colder months. Seasonal response patterns were not consistent with multiple linear regression. Alternatively, we binned measurements across the yearly cycle using environmental indices, such as temperature, and applied moving regression, a novel approach which produced superior fits to calendar day binning. Spatially averaged watershed soil temperature at 10 cm was the best overall index of discharge-fDOM response. DOM fractionation showed fDOM was primarily a surrogate for hydrophobic organic acid (HPOA) concentrations. HPOAs were highly correlated with discharge, but hydrophilics (HPIs) were not. Discharge dependent DOM concentrations driven by the HPOA fraction may be controlled by soil temperature and water table position relative to organic and mineral soil horizons. HPI concentrations were correlated with average watershed soil temperature at 10 cm but were rather stationary throughout the year, further indicating a consistent groundwater source for this nonfluorescent DOM. We present a resolved subseasonal empirical model of DOM concentrations and fluxes, showing that riverine DOM flux and quality depend heavily on seasonal terrestrial carbon dynamics and hydrologic flow paths. High frequency monitoring reveals readily discernible patterns demonstrating that upland biogeochemical signals are maintained even at this large watershed scale.