3.2.1 Natural tracers
Time series of EC available at the gauging station were used in combination with individual samples analysed for EC and stable isotopes (δ18O and δ2H) taken during various field trips between 31.07.2017 and 15.09.2017. These samples include glacial meltwater (4 samples), a cumulative rainfall sample collected during the same period using a funnel connected via a flexible tube to a bottle (placed 1 m deep within the coarse blocky surface to avoid evaporation by direct radiation), as well as 18 samples taken by an automatic sampler between 01.08.2017 and 06.08.2017 (Figure 2; sampling locations indicated in Figure 1b). During field campaigns no direct access to permafrost ice (below the active layer) was obtained, so the actual signature of permafrost ice in this catchment remains unknown.
Natural tracer data (EC, isotopes) allow to determine variations in event water contributions from the endmembers of rain, snowmelt and ice melt (e.g. Winkler et al., 2016). The restricted rock-water interaction in the alpine catchment composed of metamorphic bedrock allows regarding EC as a conservative tracer (cf. Winkler et al., 2016). Based on this assumption a two component mixing model was applied to identify groundwater and event water components at diurnal and seasonal time scales (Harrington et al., 2018; Williams et al., 2006; Winkler et al., 2016). The event water component is composed of low mineralized rainfall, snowmelt and glacier ice melt, which is distinguishable from the higher mineralized groundwater component.
3.3 Rainfall-Runoff Modelling
For a better understanding of the discharge pattern and to further constrain the contribution of rainfall, snowmelt and ice melt within the rock glacier catchment, a parsimonious lumped parameter rainfall-runoff model was deployed. The model was successfully applied for a relict rock glacier and catchments downstream of relict rock glaciers by Wagner et al. (2016) and was shown to be especially valuable in alpine catchments where input data are usually scarce (e.g. Wagner et al., 2013, 2016). Here, the model was extended by an ice store module (Figure 3) to account for ice melt within the spring catchment. The module is based on a degree-day factor and contributes ice melt as input in the production store if seasonal snow is melted as described in Nepal et al. (2017). This model setup relates ice melt to melt from glaciers and not from permafrost. For simplicity the initial glacier ice water volume is assumed to be large so that over the analysed period of more than 4 years, ice melt is warranted. The actual percentage of runoff contribution within the catchment is related to the areal coverage of the cirque glaciers within the total catchment of ~ 10 % (Figure 1b) and is assumed to remain constant over the time period of 4 years. Input data for the model are daily values of precipitation and average temperature for the catchment.
Discharge data from the gauging station are used to calibrate and validate the model using split sample tests (e.g. Klemes, 1986). The results are compared to the hydrograph analyses and natural and artificial tracer data thereby allowing to differentiate the various input (“recharge”) components over time (in contrast to the other data that indicate discharge variations). Moreover, the model is then used to hypothetically remove/ignore the cirque glacier ice stores and explore potential changes in the runoff characteristics. In addition, the model parameters representing a relict rock glacier (Wagner et al., 2016) can be used to account for a hypothetical decay from an active to a relict rock glacier (a trading space-for-time approach accounting for changing watershed behaviour; Singh et al., 2011) and observe a potential change in storage-discharge characteristics.
[Insert Figure 3]