Paired Air and Stream Temperature Analysis (PASTA) to Evaluate Groundwater Influence on Streams

Abstract

Groundwater discharge along the stream corridor is a critical component of many surface water ecosystems, particularly perennial streams and rivers (Boulton & Hancock, 2006;Briggs & Hare, 2018). Aggregate groundwater discharge is a dominant component of baseflow (the portion of the streamflow i.e., sustained between runoff and quickflow events). Baseflow represents a median of 55% of streamflow across the United States and ranges from 14% to 90% (Winter et al., 1998). Stream temperature may show a varied sensitivity to climate change (e.g., Isaak et al., 2016) than expected based on local air temperature change due to numerous factors including groundwater influence (Snyder et al., 2013). Despite their importance, measurements of groundwater discharge are often not readily scalable because they are labor-intensive or depend on extensive streamflow records that are only available at limited sites, with particularly poor coverage in headwater streams (Krabbenhoft et al., 2022). The development of new tools and techniques to rapidly identify streams and rivers with substantial

Train, Inform, Borrow, or Combine? Approaches to Process‐Guided Deep Learning for Groundwater‐Influenced Stream Temperature Prediction

“…The remaining reaches were included in the model and in the temperature component of the loss function, but were unobserved with regard to the three custom loss function terms. When observed water temperature data are plentiful, however, calculating the annual signal properties is relatively straightforward using existing methods (Hare et al., 2021; Johnson et al., 2020) and recently developed tools (e.g., https://cuahsi.shinyapps.io/pasta/, Hare et al., 2023). In addition, the custom loss function requires the user to identify appropriate weights for the components of the loss function.…”