WRF Meteorological Drivers

Purpose

This guide describes how to convert hourly WRF output from the Cal-Adapt Analytics Engine into CF-1.8 NetCDF meteorological driver files. The workflow is intended for site-based ecosystem models and other tools that consume CF-standard meteorological forcing.

caladaptaer provides two driver-generation paths:

• cae_write_cf_netcdf() writes a CF NetCDF file from a named list of WRF stars grids.
• cae_build_met_drivers() extracts WRF data for multiple sites and writes one NetCDF file per site per year.

Prerequisites

Install caladaptaer and verify that GDAL is available as described in Getting Started. This guide assumes the package is already installed.

library(caladaptaer)

The driver workflow requires WRF hourly data. LOCA2-Hybrid daily products do not provide the full hourly meteorological variable set required by these functions.

Required Inputs

Multi-site driver generation requires a site table with three columns:

Column	Description
site_id	Site identifier used in output paths
lon	Longitude in WGS84 decimal degrees
lat	Latitude in WGS84 decimal degrees

Additional columns can be present; cae_build_met_drivers() uses only site_id, lon, and lat.

sites <- data.frame(
    site_id = c("fresno", "los_angeles", "sacramento"),
    lon = c(-119.77, -118.24, -121.49),
    lat = c(36.75, 34.05, 38.58)
)

Driver generation also requires a WRF model, scenario, year range, spatial grid, and output directory. Use cae_models(), cae_scenarios(), and cae_search() to confirm availability before starting a large job.

Required Variables

The CF driver files are built from eight WRF surface meteorology variables. wind_speed is derived from u10 and v10.

CF variable	WRF variable	WRF units	CF units	Conversion
air_temperature	t2	K	K	none
air_pressure	psfc	Pa	Pa	none
precipitation_flux	prec	mm per timestep	kg m-2 s-1	divide by timestep seconds
specific_humidity	q2	kg kg-1 mixing ratio	1	q2 / (1 + q2)
surface_downwelling_shortwave_flux_in_air	swdnb	W m-2	W m-2	none
surface_downwelling_longwave_flux_in_air	lwdnb	W m-2	W m-2	none
eastward_wind	u10	m s-1	m s-1	none
northward_wind	v10	m s-1	m s-1	none
wind_speed	u10, v10	m s-1	m s-1	sqrt(u10^2 + v10^2)

Use cae_check_variables() to check whether the required variables are present or derivable for a model, scenario, timescale, and resolution.

cae_check_variables(
    model = "CESM2",
    scenario = "ssp370",
    timescale = "1hr",
    resolution = "d01"
)

Some WRF models store precipitation as accumulated convective and non-convective components rather than the precomputed prec field. caladaptaer derives prec from those components when possible.

Single-Grid Workflow

Use cae_write_cf_netcdf() when the intended output is a CF NetCDF file for a single WRF grid subset. The input is a named list of stars objects, one per required WRF variable.

wrf_vars <- c("t2", "psfc", "prec", "q2", "swdnb", "lwdnb", "u10", "v10")

var_data <- lapply(wrf_vars, function(variable) {
    cae_fetch(
        variable = variable,
        model = "CESM2",
        scenario = "ssp370",
        timescale = "1hr",
        resolution = "d01",
        start_time = "2050-07-01T00:00:00",
        end_time = "2050-07-01T23:00:00"
    )
})
names(var_data) <- wrf_vars

cae_write_cf_netcdf(
    var_list = var_data,
    outfile = "CESM2.ssp370.2050-07-01.cf.nc",
    dt_seconds = 3600,
    overwrite = TRUE
)

The output file contains CF-1.8 global metadata, CF variable names, converted units, and derived wind speed.

Multi-Site Workflow

Use cae_build_met_drivers() when the intended output is one NetCDF file per site per year. The function reads each WRF variable once per year, extracts all sites from that grid, and writes site-year files.

result <- cae_build_met_drivers(
    sites = sites,
    model = "CESM2",
    scenario = "ssp370",
    start_year = 2050,
    end_year = 2051,
    outdir = "met_drivers",
    resolution = "d01"
)

head(result)

By default, existing files are skipped. Set overwrite = TRUE only when files should be regenerated.

result <- cae_build_met_drivers(
    sites = sites,
    model = "CESM2",
    scenario = "ssp370",
    start_year = 2050,
    end_year = 2051,
    outdir = "met_drivers",
    resolution = "d01",
    overwrite = TRUE
)

Output Structure

cae_build_met_drivers() writes files using this path pattern:

{outdir}/{site_id}/{model}.{scenario}.{year}.nc

For example:

met_drivers/fresno/CESM2.ssp370.2050.nc
met_drivers/los_angeles/CESM2.ssp370.2050.nc
met_drivers/sacramento/CESM2.ssp370.2050.nc

Each file contains one site and one year of hourly data. Dimensions are latitude, longitude, and time. The time coordinate is stored as seconds since the start of the file year.