Adaptation of root zone storage capacity to climate change and its effects on future streamflow in Alpine catchments: towards non-stationary model parameters

Abstract. Hydrological models play a vital role in projecting future changes in streamflow. Despite the strong awareness of non-stationarity in hydrological system characteristics, model parameters are typically assumed to be stationary and derived through calibration on past conditions. Integrating the dynamics of system change in hydrological models remains challenging due to uncertainties related to future changes in climate and ecosystems.

Nevertheless, there is increasing evidence that vegetation adjusts its root zone storage capacity – considered a critical parameter in hydrological models – to prevailing hydroclimatic conditions. This adaptation of the root zone to moisture deficits can be estimated by the Memory Method. When combined with long-term water budget estimates in the Budyko framework, the Memory method offers a promising approach to estimate future climate-vegetation interaction and thus time-variable parameters in process-based hydrological models.

Our study provides an exploratory analysis of non-stationary parameters for root zone storage capacity in hydrological models for projecting streamflow in six catchments in the Austrian Alps, specifically investigating how future changes in root zone storage impact modeled streamflow. Using the Memory method, we derive climate-based parameter estimates of the root zone storage capacity under historical and projected future climate conditions. These climate-based estimates are then implemented in our hydrological model to assess the resultant impact on modeled past and future streamflow.

Our findings indicate that climate-based parameter estimations significantly narrow the parameter ranges linked to root zone storage capacity. This contrasts with the broader ranges obtained solely through calibration. Moreover, using projections from 14 climate models, our findings indicate a substantial increase in the root zone storage capacity parameters across all catchments in the future, ranging from +10 % to +100 %. Despite these alterations, the model performance remains relatively consistent when evaluating past streamflow, independent of using calibrated or climate-based estimations for the root zone storage capacity parameter. Additionally, no significant differences are found when modeling future streamflow when including future climate-induced adaptation of the root zone storage capacity in the hydrological model. Variations in annual mean, maximum, and minimum flows remain within a 5 % range, with slight increases found for monthly streamflow and runoff coefficients. Our research shows that although climate-induced changes in root zone storage capacity occur, they do not notably affect future streamflow projections in the Alpine catchments under study. Our findings suggest that incorporating a dynamic representation of the root zone storage capacity parameter may not be crucial for modeling streamflow in humid and energy-limited catchments. However, our observations indicate relatively larger changes in root zone storage capacity within the less humid catchments, corresponding to higher variations in modeled future streamflow. This suggests a potentially higher importance of dynamic representations of root zone characteristics in arid regions and underscores the necessity for further research on non-stationarity in these regions.