Evaluating electrical conductivity as a surrogate tracer to determine the factors controlling event and pre-event water flow partitioning at a tropical montane Andean ecosystem

Abstract

The results of this thesis showed that electrical conductivity could be used as an alternative to Oxygen-18 as a tracer for event and pre-event water flow partitioning analyses which highly increases the temporal resolution of tracer concentrations while reducing the uncertainty of the tracer-aided model. The results showed great similarity under a large range of flow conditions, reassuring the consistency of the estimated fractions with 89% of the monitored events showing differences lower than 20% in pre-event water fraction regardless of the antecedent moisture and rainfall conditions. The use of electrical conductivity was possible due to a quasi-conservative behavior related with the presence of organic-rich riparian soils (peat-type) overlying compact bedrock across the catchment. This highlights the potential of electrical conductivity to obtain high temporal frequency data while lowering the costs needed to implement and keep tracer data collection up for long time periods. In addition, results from the comparison of models with different complexities also showed great similarity in their estimations of the event and pre-event water fractions as long as appropriate concentrations of event (Ce) and pre-event (Cp) water for the simpler model are set. In fact, Cp showed to be the most important factor for improving accuracy while Ce had little influence on the results. Hence, the best way to determine Cp was the concentration of a streamflow sample taken before the beginning of each event. These findings will allow to reduce the logistical and economical resources needed to adequately assess hydrograph separation and to carry out quasi-continuous assessments of flow partitioning with high accuracy in high-Andean montane ecosystems. The previous results led to the analysis of several spatial and temporal factors controlling event and pre-event water fractions. This allowed us to obtain tracer and hydrometeorological high-frequency data from a large number of rainfall-runoff events (n=72). The correlations showed that the main temporal controlling factor was rainfall amount with a strong (i.e., r>0.7) and significant (i.e., p<0.05) positive correlation with the event water fraction, whereas, soil type, vegetation cover, and topography were highly correlated when considering the spatial factors. These results suggested an enhanced shallow subsurface hydrological connectivity between hillslopes and riparian wetlands which follow in an increase of event water fraction for events that show higher peak flows while a threshold is exceeded. Overall, our findings suggests that high temporal resolution data is extremely necessary to adequately assess event and pre-event water fraction flow partitioning as it helps to obtain a complete understanding of catchment hydrological behavior at scale event. This improved understanding could aid in the implementation of science-based water management strategies that includes many processes that are often overlooked.