Watershed Processes Laboratory  

The identification of water sources contributing to runoff is critical to understand the relation between water and  biogeochemical cycles. During rainstorm events, runoff can be composed of fractions of event and pre-event water. Tracers in two-component end member mixing analysis are commonly used to investigate these relative contributions to total runoff.  However, tracer data are often only available at low temporal resolution, leading to high uncertainties in the estimation of flow components.  Here we present TraSPAN a new numerical tracer based streamflow-partitioning model that simulates both the tracer mass balance and the water flux response at the event scale. TraSPAN has four different structures representing different internal catchment hydrologic characteristics.  We used high-resolution (0.25-5 hours) hydrometric and tracer (water stable isotopes (WSI) and electrical conductivity (EC)) data to simulate flow partitioning and compare the results between tracers for a storm in a forest headwater catchment at the western Oregon Cascades. Our results show that flow partitioning and transit time functions (TTFs) of event and pre-event water are well defined using either EC or WSI. The same model structure provided the best fit in both cases (Nash Sutcliffe Efficiency > 0.9). This structure includes two reservoirs in parallel to route the event and pre-event water fractions following independent TTFs and allows a time-variant fraction of precipitation routed as event water over the course of the storm. The level of agreement between the results attained with EC and WSI is remarkable in terms of parameter values and TTFs. Given the high cost and effort associated with the collection and analysis of WSI at high temporal resolution, our results provide great promise for the use of EC as a tracer in high-resolution flow partitioning modelling. The use of such an inexpensive tracer could allow for detailed investigation of the relative importance of internal (e.g. geology, topography) and external (e.g. event magnitude, season) features in the hydrologic response of catchments in other settings.  The use of EC could also be incorporated in alternative nonstationary modelling approaches of hydrologic response that require longer-term high-resolution data.

Strong level of agreement between the results attained with EC and WSI.  Panels a, b, d, & e present observed and predicted stream flow and tracer data for at least 500 different sets of parameters all yielding a NSE> 0.75. The best prediction is depicted in dark blue. Panel c and f present the predicted pre-event and event water contributions during the storm.

TRaSPAN conceptual model structures used in this study. 1) Assumes a constant fraction of effective precipitation (Peff) routed as event water and a single reservoir for the event and pre-event runoff components 2) Assumes a constant fraction of Peff routed as event water and two reservoirs in parallel for the event and prevent runoff components 3) Assumes that the fraction of Peff routed as event water varies over time and a single reservoir for the event and prevent runoff components.  4) Assumes that the fraction of Peff routed as event water varies over time and two reservoirs in parallel for the event and prevent runoff components.

High-resolution modeling of tracer-based flow partitioning at the catchment scale: A comparison of classic water stable isotopes and electrical conductivity approaches (2015-2017)

Funded by

22. Mosquera G., Segura C, Crespo P. 2018.Flow Partitioning Modelling Using High-Resolution Water Stable Isotopes and Electrical Conductivity. Water 2018, 10 (7), 904; https://doi.org/10.3390/w10070904.