Simulating dissolved organic carbon during dryness/wetness periods based on hydrological characteristics under multiple timescales

Abstract

Understanding and simulating dissolved organic carbon (DOC) concentrations is a challenge with important implications for water quality projections and carbon balance estimates, particularly at the event-scale. Hydrological dryness/wetness represents water shortage/surplus in surface water. Previous studies have analyzed the dynamic patterns of DOC concentrations during either dryness or wetness (or rewetting) periods. However, the dynamic patterns of DOC concentrations and corresponding simulations during both dryness and wetness events, based on their characteristics (i.e., duration and severity), remain to be fully understood, particularly under multiple timescales. This study proposed a framework to help overcome this limitation. The standardized streamflow index (SSI) combined with run theory was used to describe the dryness/wetness characteristics at the monthly, seasonal, and annual scales. Simulation models were constructed and validated based on the multivariable linear regression model and cross-validation method. Three evaluation indicators, namely, the coefficient of determination (R2), Nash-Sutcliffe efficiency (NSE), and mean absolute percentage error (MAPE, %), were employed to determine the optimal model. This model was then applied to additional basins based on the correction coefficient method. Three nearby basins (HP3a, HP4, and HP6), with different catchment areas (9.28 to 122.80 ha) and long-term (41 years) daily streamflow and weekly (or biweekly) DOC measurements in the Harp Lake catchment, south-central Ontario, Canada, were used to demonstrate the usefulness of the framework. Data from HP3a was used to construct and validate the simulation model, and the model was applied to the HP4 and HP6 basins. Results show that the average DOC concentration during the dryness events is lower than that in the wetness events at the monthly, seasonal, and annual scales. The average, maximum, and minimum DOC concentrations during the dryness events increased with the dryness/wetness index timescales, but decreased during the wetness events. The optimal model was able to accurately simulate the DOC concentrations (R2 ≥ 0.90, NSE ≥ 0.90, and MAPE ≤ 10 %) at the monthly, seasonal, and annual scales based on the hydrological characteristics during the dryness and wetness periods, respectively, and the application results exhibit a good performance. The proposed framework can help improve the forecasting of DOC concentrations and the corresponding ecological coping strategies at the event-scale under changing environments.