Caravan CAMELS-US Update

Updated CAMELS-US Streamflow Data for Caravan (Through Dec 2023) Description This repository contains an update to the streamflow target data for the CAMELS (US) subset of the Caravan hydrology dataset. While the original CAMELS-US data within Caravan extends only through 30/09/2014, this update extends the streamflow records for these gauges through 30/12/2023, which matches the maximum date range of other datasets in the original Caravan dataset. This dataset does not add new basins/locations; rather, it extends the temporal coverage of the existing CAMELS-US basins found in Caravan (this is the camels dataset in Caravan). The data allows for training and evaluating hydrological models on the most recent decade of hydrological variability in the United States. One basin from the original CAMELS dataset (ID 08171300) is no longer available through the USGS data service – this basin was retained in this dataset with the original CAMELS streamflow data. Repository Content This repository includes both the updated dataset and the code used to generate it. 1. Data (timeseries/) The data follows the standard Caravan file structure and formatting. It includes both NetCDF and CSV formats. timeseries/netcdf/camels/: NetCDF files containing the original Caravan forcing data (meteorological data) merged with the updated USGS streamflow observations. timeseries/csv/camels/: CSV files containing the same time series data. The streamflow variable has been converted from the raw USGS units (cubic feet per second) to mm/day, consistent with the Caravan standard, using the catchment area attributes provided in the Caravan camels attributes file. 2. Code (CAMELS-US-2025-USGS-Download.ipynb) The Jupyter Notebook used to generate this dataset is included. It documents the full retrieval and processing workflow: Identification: Identifies target gauges using the Caravan attributes_other_camels.csv file. Retrieval: Fetches daily mean streamflow (Parameter 00060, Statistic 00003) from the USGS National Water Information System (NWIS) using the dataretrieval Python package. Processing: Converts discharge from cubic feet per second (cfs) to mm/day using basin area. Merging: Injects the new streamflow data into the existing Caravan NetCDF files (preserving the original forcing data) and exports to CSV. Usage This dataset is designed to be a drop-in replacement or update for the camels folder in the standard Caravan dataset. To use this data with the GoogleHydrology library or other Caravan-compatible pipelines, simply point your data configuration to the timeseries directory contained in this repository. Methodology Notes Source: USGS National Water Information System (NWIS). Time Period: 1980-01-01 through 2025-12-26. Variable: Daily Mean Discharge (converted to mm/day). Missing Data: Days with invalid or missing USGS data are represented as NaN. Related Works If you use this dataset, please cite the original Caravan paper and the CAMELS dataset: Caravan Dataset: Kratzert, F., et al. "Caravan - A global community dataset for large-sample hydrology." Scientific Data 10, 61 (2023). https://doi.org/10.1038/s41597-023-01975-w Caravan GitHub: https://github.com/kratzert/Caravan CAMELS-US: Newman, A. J., et al. "Development of a large-sample watershed-scale hydrometeorological data set for the contiguous USA: data set characteristics and assessment of regional variability in hydrologic model performance." Hydrology and Earth System Sciences 19.1 (2015): 209-231. License This repository is provided under the BSD 3-Clause License. See the LICENSE file for details.