AUTHORS' NOTE: The paper was reviewed by three reviewers in WRR and did not pass the review. There are some technical issues with our methodology. We advise the readers to consult the authors before using any results from this work. We are sorry for this inconvenience. For readers who interested in the reviewers' comments, please contact me .Abstract:The transit time (TT) of streamflow encapsulates information about how catchments store and release water and solutes of different ages. The young water fraction (Fyw), the fraction of streamflow that is younger than a certain age (normally 2–3 months), has been increasingly used as an alternative metric to the commonly used mean TT (mTT). In the commonly used (‘traditional’) procedure presented by Kirchner (2016), the age threshold (τyw) of Fyw separating young from old water is not pre-defined and differs from catchment to catchment depending on the shape of the (gamma) transit time distribution. However, it can be argued that it is important to use the same pre-defined τyw for inter-catchment comparison of Fyw. In this study, we propose an alternative (‘proposed’) procedure for the estimation of Fyw with any pre-defined τyw. This allows us to also compare the effects of data sampling frequencies on the results of Fyw estimation using the same τyw. We applied the traditional and proposed procedures using daily oxygen isotope (δ18O) data in the Alp and Erlenbach catchments, Switzerland. We found that our proposed and the traditional procedure can give very different Fyw values. With the proposed procedure, the estimated Fyw significantly increases when the sampling frequency changes from sub-monthly to monthly time steps. Overall, our study highlights the importance of the selection of τyw and the sampling frequency in Fyw estimation, which should be given more attention.

Understanding groundwater behavior is essential for water resources management in alluvial deltas. This study investigated the trends of groundwater levels (GWLs), the memory effect of alluvial aquifers, and the response times between surface water and groundwater across the Vietnamese Mekong Delta (VMD). 88 time series of GWL between 1996 and 2017 were collected at 27 national stations. Trend analysis, auto- and cross-correlation, and time-series decomposition were applied within a moving window approach to examine nonstationary behavior. Our study revealed high ratios of the seasonal component in shallow aquifers, and dominating ratios of the trend component in deep aquifers. These findings indicate an effective connection between the Holocene aquifer and surface water, and a high potential for shallow groundwater recharge. On the other hand, low-permeable aquicludes separating the aquifers behave as low-pass filters that reduce the high‐frequency signals in the GWL variations, and limit the recharge to the deep groundwater. Declining GWLs (0.01-0.55 m/year) were detected for all aquifers throughout the 22 years of observation, indicating that the groundwater system is currently not fully recharged. Stronger declining trends were detected for deep groundwater. While the slight decline of GWLs in the Holocene aquifer (0.01-0.11 m/year) is likely caused by natural conditions, the significant decline in the Pliocene and Miocene aquifers (0.30-0.55 m/year) is attributed to the overexploitation of groundwater. The time-variant trend analysis indicates that the decrease of GWL accelerated continuously. The groundwater memory effect varies according to the geographical location, being shorter in shallow aquifers and flood-prone areas and longer in deep aquifers and coastal areas. Variation of the response time between the river and alluvial aquifers is controlled by groundwater depth, seasonal variability, and the location with shorter response times for shallow groundwater, during the flood season, and in flood-prone areas. Our findings are not only essential for groundwater resource management in the VMD, but they also characterize general mechanisms of aquifer systems in alluvial settings.