Search for: All records

ABSTRACT Analysis of PRISM and SNOTEL station data paired with USGS streamflow gage data in the western United States shows that, in snow‐dominated mountainous watersheds, streamflow regimes differ between watersheds with karst geology and their non‐karst neighbours. These carbonate aquifers exhibit a spectrum of flow paths encompassing karst conduits, including large fractures or voids that transmit water readily to springs and other surface waters, and matrix flow paths through soils, highly fractured bedrock, or porous media bedrock grains. A well‐connected karst aquifer will discharge a large portion of its accumulated precipitation to surface water via springs and other groundwater flow paths on an annual scale, exhibiting a lagged response to precipitation presenting as a “memory effect” in hydrograph time series. These patterns were observed in the hydrologic records of gaged watersheds with exposed or near‐surface carbonate layers accounting for > 30% of their drainage area. In western snow‐dominated watersheds, where paired streamflow and SNOTEL data are available, analysis of the precipitation and flow time series shows low‐flow volume is strongly related to karst aquifer conditions and winter precipitation when compared to low‐flow volumes present in non‐karst watersheds, which have a complex relationship to multiple driving metrics. Analysis of normalised streamflow and cumulative precipitation in karst watersheds show that low‐flow conditions are highly dependent on the preceding winter precipitation and streamflow in both wet and dry periods. In non‐karst watersheds, increased precipitation primarily impacts high‐flow, spring runoff volumes with no clear relationship to low‐flow periods. When comparing cumulative streamflow and precipitation volumes within each water year and over longer timescales, karst watersheds show the potential filling and draining of large amounts of karst storage, whereas non‐karst watersheds demonstrate a more stable storage regime. Communities in many western US watersheds are dependent on snow‐dominated karst watersheds for their water supply. This analysis, using widely available hydrologic data, can provide insight into the recharge and storage processes within these watersheds, improve our ability to assess current flow regimes, anticipate the impacts of climate change on water availability, and help manage water supplies. 

Abstract In many regions globally, snowmelt‐recharged mountainous karst aquifers serve as crucial sources for municipal and agricultural water supplies. In these watersheds, complex interplay of meteorological, topographical, and hydrogeological factors leads to intricate recharge‐discharge pathways. This study introduces a spatially distributed deep learning precipitation‐runoff model that combines Convolutional Long Short‐Term Memory (ConvLSTM) with a spatial attention mechanism. The effectiveness of the deep learning model was evaluated using data from the Logan River watershed and subwatersheds, a characteristically karst‐dominated hydrological system in northern Utah. Compared to the ConvLSTM baseline, the inclusion of a spatial attention mechanism improved performance for simulating discharge at the watershed outlet. Analysis of attention weights in the trained model unveiled distinct areas contributing the most to discharge under snowmelt and recession conditions. Furthermore, fine‐tuning the model at subwatershed scales provided insights into cross‐subwatershed subsurface connectivity. These findings align with results obtained from detailed hydrogeochemical tracer studies. Results highlight the potential of the proposed deep learning approach to unravel the complexities of karst aquifer systems, offering valuable insights for water resource management under future climate conditions. Furthermore, results suggest that the proposed explainable, spatially distributed, deep learning approach to hydrologic modeling holds promise for non‐karstic watersheds.