2imi9's picture
Update README.md
4bf3445 verified
metadata
dataset_info:
  features:
    - name: date
      dtype: timestamp[ns, tz=UTC]
    - name: plant_code
      dtype: int64
    - name: gauge_id
      dtype: string
    - name: discharge_cfs
      dtype: float64
    - name: gauge_lat
      dtype: float64
    - name: gauge_lon
      dtype: float64
  splits:
    - name: train
      num_bytes: 684740
      num_examples: 13140
  download_size: 162361
  dataset_size: 684740
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

USGS Streamflow Near U.S. Power Plants (2022–2023)

This dataset provides daily streamflow observations from the U.S. Geological Survey (USGS) National Water Information System (NWIS) for rivers and streams located near a sampled subset of U.S. electric power plants. It is intended for studies of hydrologic constraints on power generation, climate vulnerability analysis, and machine-learning applications requiring hydrology–infrastructure relationships.

Contents

The dataset includes:

  • 13,140 daily streamflow records
  • 18 matched USGS gauges
  • 50 power plants sampled from the EIA-860 registry
  • Daily streamflow from 2022-01-01 to 2023-12-31 (where available)

Each record represents the discharge at the nearest available USGS gauge to a specific power plant on a specific day.

Variables

Column Description
date Timestamp of observation (UTC, daily mean)
plant_code EIA plant identifier (from EIA-860 plant registry)
gauge_id USGS site identifier
discharge_cfs Mean daily discharge in cubic feet per second (CFS)
gauge_lat Latitude of USGS gauge (WGS84)
gauge_lon Longitude of USGS gauge (WGS84)

Methodology

  1. For each sampled power plant, a bounding box of ±0.5° latitude/longitude was used to search for USGS gauges with daily discharge data (parameterCd = 00060).
  2. If multiple gauges were found, the closest in geographic distance was selected.
  3. Daily discharge was retrieved for January 1, 2022 through December 31, 2023.
  4. Observations were merged with plant identifiers for cross-analysis.

Source

Processing Notes

  • Only gauges with available daily discharge time series were retained.
  • Some plants have no nearby gauge; resulting dataset includes streamflow for 18 of 50 sampled plants.
  • All timestamps were converted to pandas datetime format for consistency.
  • All columns use lowercase snake_case formatting.

Citation

If you use this dataset, please cite:

2imi9. USGS Streamflow Near U.S. Power Plants (2022–2023). Hugging Face Datasets. https://huggingface.co/datasets/2imi9/us-power-climate-data-streamflow