Climate Forcings Across Multiple Timescales Control Stream Drying and Wetting in a Headwater Catchment

Hydrologists commonly use climate metrics to understand watershed function and streamflow response due to their widespread availability and effectiveness in hydrologic models used to predict streamflow. This is particularly true in research aimed at characterising how streamflow response varies under changing climate conditions. In semi-arid to arid environments, researchers typically hypothesise that in drier, warmer years, vegetation and streamflow are in direct competition for stored water. Conceptual models suggest that seasonal plant water use and evapotranspiration (ET) can ultimately cause stream drying (or non-perennial flow) in summer months. While past research has directly investigated links between riparian ET and stream drying, we have yet to explore the link between the climate conditions that drive ET and stream non-perenniality, particularly over longer and varying timescales. Here, we used commonly observed climate variables in combination with machine learning and explainable artificial intelligence techniques to identify drivers and their impacts on the wetting and drying characteristics in the semi-arid Dry Creek Experimental Watershed. Our results highlight that actual ET (AET) and precipitation are highly correlated to non-perennial behaviour. We found an inverse relationship between ET and drying: higher AET corresponds to less drying. This may suggest that watershed scale AET and streamflow might not be in direct competition; rather, AET is an indicator of overall water availability in this semi-arid climate. Finally, we found that wetting and drying behaviour is driven by climate processes on shorter (weekly) and longer (seasonal) timescales, respectively. Non-perennial streamflow processes rely on antecedent climate conditions active over a range of timescales. Therefore, it is not sufficient to use a singular temporal discretisation of climate metrics to determine hydrologic response. These results can be used to inform the design of interdisciplinary studies that link climate, hydrology and ecological function to climate change impacts.