Can we parameterise Subsurface Stormflow in a conceptual simulation model using flow duration curve percentiles for calibration?

Subsurface stormflow (SSF) is an important runoff-generation process, especially in humid, mountainous regions. It can play a major role in flood generation and contaminant transport at the catchment scale. However, as it is a subsurface and heterogeneous process, its monitoring can be very challenging. In turn, the identification of SSF parameters in hydrological models is a difficult task and is often affected by equifinality.  Our study uses the HBV-light model, a conceptual model at the catchment scale, to simulate SSF (and catchment discharge) at a catchment in the Ore Mountains in Saxony, Germany. To see whether it is possible to improve the identifiability of SSF parameters by looking at different flow conditions separately, we divide discharge data according to flow duration curve (FDC) percentiles. We then calibrate the conceptual model several times, each time using only discharge data within one percentile of the FDC. Using a Monte Carlo based calibration, we select the same number of behavioural parameter sets for every FDC percentile based on the Kling-Gupta-Efficiency as an objective function. With a regional sensitivity analysis as well as a GLUE uncertainty estimation, we analyse and compare the parameter sets and discharge simulations of the different percentile calibrations. In this way, we analyse whether there is more information content on SSF hidden in a specific part of the FDC and thus, SSF parameters and processes become better quantifiable.