The drying regimes of non-perennial rivers

The paradigm of surface water flow regimes is central to the aquatic sciences, where the timing, duration, frequency, magnitude, and rate of change of flow drive physical, chemical, and biological functions in aquatic systems. However, non-perennial streams comprise the majority of the global river network and there is a need to understand not just whether or not a stream periodically dries, but how it dries. Here, we propose to flip the script on flow regimes by presenting a comprehensive 'drying regime' framework to characterize stream drying.  We then identify similar drying characteristics in streams across watersheds with a broad range of climates, physiographic regions, and land uses. Using daily streamflow from 894 U.S. Geological Survey streamflow gages we isolated over 25,000 unique drying events over a period from 1979 - 2018. From these drying events we identified and calculated streamflow metrics that describe timing, duration, magnitude, frequency, and rate of change of stream drying. Using multivariate statistics, k-means clustering, and random forest analyses we grouped drying events into distinct drying regimes and determined the drivers of the clustered regimes. K-means clustering resulted in 4 distinct drying regimes characterized by (1) more frequent drying, (2) longer no-flow duration, (3) drying occurring following low antecedent flows, and (4) flashy high frequency drying, respectively. The majority of gages had more than one drying regime present at different times within each year, suggesting that dominant flow paths or drivers varied through time  Clustered drying regimes show low event-scale spatial coherence, and while drying regimes (1) and (2) show similar frequency throughout the year, (3) and (4) are substantially more frequent during summer months. Based on random forest analysis, land cover characteristics appear to drive drying event assignment to drying regimes more than climate variables. Furthermore, increased importance of individual watershed properties shows that the structural makeup of the watershed is notably more important to how an intermittent system dries than climate or physiographic characteristics. Non-perennial systems have unique functions due to the occurrence of both flowing and dry states, yet most of the past efforts rely on frameworks built around perennial streamflow behavior. Our work presents a novel drying regime framework to allow future studies to more effectively connect river drying to the physical, chemical, and biological functioning in these systems. This framework may also aid in current sustainable river management, including engineered flow regimes that are designed to balance water allocations with ecosystem requirements.