Assessment of ensemble data assimilation based heat tracer method for estimating surface water-groundwater interaction at seasonal timescale under complex field conditions

Quantifying fluid fluxes between surface water (SW) and groundwater (GW) is essential for understanding nutrient transport processes across the terrestrial-aquatic interface and river ecosystems. For highly dynamic rivers (e.g., dam-regulated river, tidal river), frequent fluctuations in river stages present challenges in the assessment of SW-GW interactions. To address this challenge, an ensemble smoother-multiple data assimilation (ES-MDA) based heat tracer method was proposed previously, which outperforms analytical heat tracer methods and is able to capture sub-daily fluxes accurately. However, the performance of this method under non-ideal conditions (e.g., unsaturated flow, heterogeneous sediment) has not been thoroughly investigated, especially at the seasonal time scale when the flow patterns vary distinctively. In this study, we examined the influences of seasonal flow patterns, riverbed heterogeneity, and unsaturated flow conditions on the performance of the ES-MDA method for flux estimation. Our findings indicate that the ES-MDA method yields robust results under saturated flow conditions with heterogeneous permeability fields across all seasons. Both long-term low fluxes (e.g., ± 1 m/day) and temporary flux peaks (e.g., 3 m/day) can be accurately captured. Temperature differences between different depths significantly affect the estimation uncertainty, and the co-occurrence of low temperature differences and high heterogeneity may weaken the method's performance. In variably saturated zones, while the flow direction remains identifiable, the estimated fluxes may be unrealistic. Our work demonstrates that the ES-MDA method has the potential for application under complex field conditions for long-term monitoring of SW-GW interactions.