Sink et al. (2026) MACH: A Multi-Attribute Catchment Hydrometeorological dataset

Identification

• Journal: Scientific Data
• Year: 2026
• Date: 2026-04-04
• Authors: Katharine Sink, Tom Brikowski
• DOI: 10.1038/s41597-026-07162-x

Research Groups

• Sustainable Earth Systems Sciences, University of Texas at Dallas

Short Summary

This paper introduces MACH, a comprehensive hydrometeorological dataset for 1,014 watersheds across the contiguous United States, unifying and extending existing large-sample hydrology resources with consistent daily meteorological forcings, streamflow observations, and diverse watershed attributes over a 44-year period (1980-2023), with a 75-year extension for a subset of basins.

Objective

• To provide a standardized, comprehensive, and high-resolution hydrometeorological dataset (MACH) to advance large-sample hydrology, hydroclimatic assessment, and data-driven water science, facilitating investigations of climate-runoff interactions, hydrologic sensitivity, model calibration, and long-term watershed change.

Study Configuration

• Spatial Scale: 1,014 watersheds across the contiguous United States (CONUS), with drainage basin areas ranging from 4.7 km² to 25,546 km².
• Temporal Scale: Daily resolution for a 44-year period (January 1, 1980 – December 31, 2023) for all 1,014 watersheds, extended to a 75-year period (January 1, 1948 – December 31, 2023) for 395 MOPEX watersheds.

Methodology and Data

• Models used:
  • Daymet Version 4 (for gridded meteorological forcings)
  • GLEAM4.2a (Global Land surface Evaporation: the Amsterdam Methodology, for actual and potential evapotranspiration)
• Data sources:
  • Streamflow observations: U.S. Geological Survey (USGS) National Water Information System (NWIS)
  • Meteorological forcings (precipitation, air temperature, snow water equivalent, shortwave radiation, day length, water vapor pressure): Daymet Version 4
  • Actual and potential evapotranspiration: GLEAM4.2a
  • Watershed boundaries and attributes (climate, geology, soil, hydrology, hydrologic modifications, landscapes, topography): NHDPlusV2 (National Hydrography Dataset Plus, Version 2.1)
  • Additional soil attributes: NRCS Soil Survey Geographic Database (SSURGO)
  • Annual land cover: Annual National Land Cover Database (ANLCD) from the Multi-Resolution Land Characteristics (MRLC) Consortium
  • Vegetation index (NDVI): GIMMS-3G+ (Global Inventory Modeling and Mapping Studies-3rd Generation) from NASA Earthdata
  • Anthropogenic influences (dam locations and attributes): National Inventory of Dams (NID)
  • Reference watershed classification: GAGES II dataset
  • Hydrological signatures: Calculated from USGS daily streamflow data using the R package EflowStats
  • Climate indices: Calculated from daily precipitation and temperature data using R, based on CLIVAR ETCCDI definitions

Main Results

• The MACH dataset was created, comprising daily hydrometeorological forcings and a rich set of static and dynamic watershed attributes for 1,014 watersheds in the contiguous United States.
• It unifies 395 MOPEX basins and 671 CAMELS basins, providing consistent data from 1980-2023 for all, and extending the record back to 1948 for the MOPEX subset.
• The dataset incorporates updated and consistent climate forcings from Daymet V4 and GLEAM4.2a, along with comprehensive attributes including topography, soils, geology, annual land cover (1985-2023), long-term NDVI (1982-2022), anthropogenic influences (dams, population/road density), and various hydroclimatic and hydrological indices.
• Validation demonstrated high correlation between MOPEX and MACH for precipitation (0.74-0.95) and temperature, and overall reliability of the merged dataset, with no significant artificial change points introduced at the merger date (January 1, 1980).
• Rigorous validation of basin delineations revealed and addressed discrepancies with original datasets, ensuring consistency with NHDPlusV2 and providing transparent documentation of area differences.

Contributions

• Harmonizes and significantly extends two foundational large-sample hydrology datasets (MOPEX and CAMELS) into a single, consistent, and updated resource (MACH).
• Provides a unified framework with consistent USGS stream gage station identifiers, ensuring continuity and comparability with previous research.
• Integrates state-of-the-art climate forcings (Daymet V4, GLEAM4.2a) with improved spatial resolution and quality compared to previous versions and other global datasets.
• Offers an extended temporal coverage for a substantial portion of the basins (395 MOPEX basins) back to 1948, enabling longer-term hydrologic analyses.
• Includes a comprehensive and diverse suite of static and dynamic watershed attributes, such as annual land cover and long-term NDVI, which are crucial for advanced hydrological modeling and research.
• Addresses and documents known data quality issues and inconsistencies in basin delineations and forcing data present in prior large-sample datasets, enhancing data reliability and transparency.

Funding

No specific funding projects or programs are listed in the provided text.

Citation

@article{Sink2026MACH,
  author = {Sink, Katharine and Brikowski, Tom},
  title = {MACH: A Multi-Attribute Catchment Hydrometeorological dataset},
  journal = {Scientific Data},
  year = {2026},
  doi = {10.1038/s41597-026-07162-x},
  url = {https://doi.org/10.1038/s41597-026-07162-x}
}