On the value of isotope-informed model calibration

Abstract:

In regions of mid to high-latitude, or regions where data scarcity or paucity are an issue, expansion of traditional hydrometric data networks is often neither cost effective nor realistic. With limited data available (both spatially and temporally), there is a need for hydrologic models to assess watershed-scale response to climate and anthropogenic change, and enhanced data networks to support these modelling efforts. Limited hydrometric coverage means less information for model calibration, which can result in increased parameter uncertainty. The use of coupled frameworks that integrate hydrology and stable water isotopes for the purpose of assisting with model calibration is at the forefront of our efforts. Using an isotopically-enabled hydrologic model (isoWATFLOOD) we demonstrate the impact tracers have on model calibration outcomes, process-based apportioning within the model, and the added information content these data provide. Though isotope data do not necessarily improve calibration scores, they do not degrade model performance, and are shown to improve long-term model performance. There are distinct and significant impact on process function within the model that affect travel times and evapotranspiration loss predicted by the model, having implications for soil moisture storage. Isotope-enabled calibration is shown to provide a more realistic representation of long-term flow path and storage apportionment, which is crucial for climate change and water balance assessment.