Base flow estimation via optimal hydrograph separation at CONUS watersheds and comparison to the National Hydrologic Model - Precipitation-Runoff Modeling System by HRU calibrated version

Optimal hydrograph separation (OHS) is a two-component, hydrograph separation method that uses a two-parameter, recursive digital filter (RDF) constrained via chemical mass balance to estimate the base flow contribution to a stream or river (Rimmer and Hartman, 2014; Raffensperger et al., 2017). A recursive digital filter distinguishes between high-frequency and low-frequency discharge data within a hydrograph, where high-frequency data corresponds to quick flow or storms and low-frequency data corresponds to base flow. The two parameters within the RDF are alpha and beta, both are unitless. Alpha is defined as the recession constant and typically found through recession analysis. For the purposes of this data release and study, we derived alpha from a groundwater flow coefficient (gwflow_coef) defined in the National Hydrologic Model Infrastructure run with the Precipitation-Runoff Modeling System (NHM-PRMS) (Regan et al., 2018). The second parameter, beta, is defined as the maximum value of the base flow index (Eckhardt, 2005). Beta is optimized using specific conductance and mass balance techniques, where a hydrograph is split into quick flow and base flow and specific conductance values are proposed for these streamflow components. OHS uses two model types to estimate base flow specific conductance from stream specific conductance, referred to as 'SCfit' and 'sin-cos' model types. The 'SCfit' model type uses a peak-fitting algorithm to define time periods where the stream is entirely comprised of base flow, whereas the 'sin-cos' model type emulates seasonal variation in streamflow specific conductance with a sine-cosine function to pinpoint when base flow contributes to streamflow. For more information and equations regarding model type and OHS methods, please see the associated publication (Foks et al., 2019). OHS was applied to 1076 stream gages within the conterminous United States (CONUS) where daily streamflow and daily or discrete measurements of specific conductance were collected. Gages were selected for this method if they were of "reference quality" as defined by the Geospatial Attributes of Gages for Evaluating Streamflow (GAGES-II) dataset (Falcone, 2011). Of these 1076 sites, 825 had "successful" OHS models - implying good agreement between observed and simulated stream specific conductance. This data release contains the results of applying OHS to hundreds of stream gages of varying watershed characteristics, summary of watershed and hydro-climatological characteristics for each site (Falcone, 2011; USGS, 2003), and a comparison of OHS-defined base flow to base flow -analogous flow components within the NHM-PRMS (gwres_flow and slow_flow) (Regan et al., 2018; Regan et al., 2019). For this data release and study, comparisons of OHS-defined base flow were made to the "by HRU" calibration of the NHM-PRMS (Hay, 2019).