CONUS404: Four-kilometer long-term regional hydroclimate reanalysis over the conterminous United States (ver. 3.0, June 2026)

CONUS404 is a unique, high-resolution hydro-climate dataset appropriate for forcing hydrological models and conducting meteorological analysis over the contiguous United States. This metadata record serves as documentation for the authoritative version of the CONUS404 atmospheric forcing dataset. CONUS404 is an abbreviated description for the original 40-year dataset: CONtiguous United States for 40 years at 4-kilometer grid spacing; however, the dataset has been revised to now include 45 years of data. This is a dataset of historical conditions (water years 1980-2024, October 1, 1979-September 30, 2024) and has sufficient temporal and spatial detail to resolve mesoscale atmospheric processes, making it appropriate for forcing hydrological models and conducting meteorological analyses. The spatial domain extends beyond the CONUS into Canada and Mexico, thereby capturing transboundary river basins and covering all contributing areas for the CONUS surface waters.
The CONUS404 simulation was run by the National Center for Atmospheric Research (NCAR) as part of a collaboration with the U.S. Geological Survey (USGS) Water Mission Area. It is the output of the Weather Research and Forecasting (WRF) v3.9.1.1 model (Skamarock and others, 2008), forced with ERA5 reanalysis data (Hersbach and others, 2020), and consists of time series of nearly 200 2-dimensional variables and a wide range of 3-dimensional variables. CONUS404 is the successor to the CONUS1 dataset (Liu and others, 2017) with improved representation of weather and climate conditions in the central United States due to the addition of a shallow groundwater module in the Noah-Multiparameterization (Noah-MP) Land Surface Model, among other improvements.
Access to Original Dataset
The full CONUS404 simulation output files are approximately one petabyte (1 PB) in volume and are being archived on the U.S. Geological Survey's Black Pearl* tape drive system. This is raw model output in a format that does not support efficient subsetting, so it is not optimal for many users. We have two additional pathways that are likely to provide more convenient access to a subset of the data, each of which is described below. The data available through each of these access points will be in a different format than the original model output archived on Black Pearl; however, the data values are scientifically equivalent. Please reach out to the data release contact if you think you need access to the original model outputs archived on Black Pearl.
The metadata in this data release was organized to align with the structure of the original model outputs, but it can be used as a reference for data accessed at any of the access points. The original model outputs include three sets of files, each of which has a different time step.
1. wrfout files are hourly files with all model outputs. There are a total of 394,488 files stored as 1 netCDF4 file per hour.
2. auxhist24 files are 15-minute files for precipitation and 2-meter temperature. There are a total of 16,437 files stored as 1 netCDF4 file per day (96 time steps in each); these daily files have been bundled into monthly tar files.
3. wrfxtrm files are daily minimum, maximum, and mean values of a selection of surface variables. There are a total of 16,437 files stored as 1 netCDF4 file per day; these daily files have been bundled into monthly tar files.
The Entity and Attribute element of the metadata record documents data dictionaries for all variables in each of the three types of output files. These data dictionaries are attached to this data release.
NCAR Research Data Archive
The first alternate access point is NCAR’s Research Data Archive (RDA). The values in this dataset are the same as those in the raw files, with an exception for some variables having a slight reduction in numerical precision, but no loss of scientific information. In addition, duplicate and empty data variables have been removed. In addition, 14 additional variables derived from the raw output have been added to this version of the CONUS404 dataset (details are provided in the data dictionary found on the “Documentation” tab on the RDA page for this dataset). RDA data access can be found at https://rda.ucar.edu/datasets/ds559.0/. There is no cost associated with data download.
Analysis-Ready Zarr Store
Access to a subset of the CONUS404 data is also provided in zarr format, which is a chunked data format (https://www.unidata.ucar.edu/blogs/developer/entry/chunking_data_why_it_matters) that is optimized for workflows that only require a subset of the dataset or workflows that utilize distributed computing. This data includes a subset of the variables available in the wrfout and wrfxtrm files from the original model output. Access to these data files is described here: https://hytest-org.github.io/hytest/dataset_access/CONUS404_ACCESS.html.
Please refer to the Supplemental Information element of this metadata record for further information on CONUS404.
*Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.