ABSTRACT:

This dataset contains hand-labeled classifications of baseflow-dominant (BFD) periods from 182 USGS stream gages across the continental United States, representing the first systematically developed ground-truth dataset for baseflow period identification. The dataset encompasses daily streamflow records spanning from 1890-2024, with each record classified as either baseflow-dominant (BFD=1) or non-baseflow-dominant (BFD=0) based on expert hydrological analysis using graphical hydrograph separation principles. BFD periods were identified as flow conditions occurring without quickflow contributions, focusing on streamflow magnitude relative to long-term averages, hydrograph stability during recession periods, and characteristic slope patterns of baseflow behavior. Quality assurance included independent labeling by multiple experts to ensure consistency and reliability. This benchmark dataset enables systematic evaluation of automated BFD identification algorithms, supports machine learning model development, and facilitates continental-scale hydrological research for improved low-flow forecasting and groundwater-surface water interaction assessments.

ABSTRACT:

This repository contains CSV files of selected gauges, evenly distributed across different geological sections according to USGS divisions. Each data point has been manually labeled to indicate whether baseflow is the only source of the streamflow.

ABSTRACT:

In this repository, 178 streamflow data from several gauges are gathered and for each of them, the baseflow value, based on the Chapman's model is calculated.

ABSTRACT:

This HydroShare resource contains all datasets, reproducible workflows, and Python/Jupyter Notebook scripts used in the manuscript “How Well Do U.S. National Water Model Short-Range Forecasts Predict Flood Event Timing and Magnitude?” (Maghami et al., 2025; under revision). The study evaluates the U.S. National Water Model (NWM) v2.1 short-range (0–18 hr) forecasts for 306 USGS stream gauges across 16 study areas in the continental United States, covering flood events occurring between April 2021 and September 2023.

The resource provides a complete end-to-end reproducible pipeline, including:

- identification and selection of USGS gauges and flood peaks,
- watershed delineation and integration of land-cover, climate-zone, regulation status, stream order, and drainage-area attributes,
- gauge-to-COMID matching using NHDPlus V2.1,
- extraction of short-range NWM forecasts and return-period estimates via the NWM BigQuery API,
- quality-control screening of USGS observations,
- multi-event flood selection using consistent peak-based criteria, and
- computation of evaluation metrics (scaled KGE, time-to-peak bias, peak-discharge ratio, and flow-volume ratio) across 18 lead times.

The workflows also generate the hydrographs, performance plots, and stratified analyses used in the manuscript (by return period, climate zone, imperviousness, stream order, and regulation status). The notebooks (JN1–JN10) and directory structure are fully documented to support transparency and reproducibility.

A README file is included with instructions for running each workflow, required software environments, and detailed descriptions of all input and output files. This resource will be made public upon manuscript acceptance and assigned a DOI to support citation, reproducibility, and long-term accessibility.