On the Importance of groundwater constraining in hydrological modeling of the Pamir region

Central Asia is a climate change hot spot, facing an unprecedented juxtaposition of regional climate- and water-related issues. Meltwater from the Pamir Mountains plays a crucial role in Central Asia's hydrological cycle, and its response to climate change has been widely investigated using glacial-hydrological models. However, the hydrological simulation in Pamirs is highly uncertain, primarily driven by data scarcity and the complex interplay between climatic factors and glacier dynamics. Ongoing efforts concentrate on including more calibration data and constraining the uncertainty about the exact internal process representation of hydrological models. However, the quality of the groundwater simulation is often neglected. Groundwater reservoirs, buffering meltwaters and providing river flow when little to no surface runoff occurs, are extremely important in the Pamir region. Although physically based groundwater models provide a more detailed picture of the possible evolution of the system, empirical groundwater models are often used in hydrological modeling due to their minimal input data requirements and low computational cost compared to physically based models. However, the traditional empirical groundwater model with single linear storage is not suitable for the Pamir region. The region is characterized by a variety of sedimentary deposits in different landscape morphologies, resulting in varying delays in water recharge, release, and storage capacities. We improved the baseflow representation by coupling two linear groundwater reservoirs (one fast and one slow) into a widely-used hydrological model in the region. A representative catchment in the central Pamir, the Gunt River basin, is used as a case study to demonstrate the importance of groundwater in constraining the hydrological calibration process. Groundwater is the only contribution to winter river discharge in the Gunt basin and can thus be used as an indicator of groundwater parameter constraint. Here we show that the hydrological model can achieve good performance (in terms of daily discharge, seasonal snow cover fraction, and annual glacier mass balance) even when calibrated with only total daily discharge and winter baseflow. Especially the baseflow calibration helps constraining snowmelt onset in spring and improving adjustments of precipitation and temperature, which are the most uncertain sources in hydrological modeling in the region. Despite improvements, degree day factors still show a large variability. The resulting model equifinality problem still leads to predictive uncertainty, indicating that more glacier observations are needed for a sound process understanding. Based on the simulated results, the hydrological cycle in Gunt was analyzed and compared with previous studies.