I’ve run into similar compatibility issues between custom NetCDFs from R and ArcMap, especially with Daymet data. Let’s break down why your current setup isn’t working and walk through a fix that aligns with CF Conventions (which ArcMap relies on for projection parsing):

Key Issues in Your Current Implementation #

• Non-compliant standard parallel formatting: The original Daymet file uses duplicate standard_parallel attributes (per CF rules for Lambert Conformal Conic), but you split them into standard_parallel and standard_parallel_2—ArcMap doesn’t recognize this non-standard naming.
• Missing projection-data link: Your raster variables (like temperature or precipitation) aren’t associated with the lambert_conformal_conic projection variable. ArcMap needs this explicit link to apply the projection correctly.
• Incorrect attribute value types: You’re passing numeric parameters (like -100 for central meridian) as strings, which can break ArcMap’s ability to parse projection settings.
• Mismatched time_bnds dimension: The original Daymet file uses time_bnds[nv, time] (where nv=2 for time interval start/end), but your file uses a 1D time_bnds[time]—this disrupts time dimension parsing.

Step-by-Step Fixes #

1. Define Projection Variable with CF-Compliant Attributes #

Use numeric values for all parameters, and set both standard parallels as a vector to create duplicate attributes matching Daymet’s structure:

library(ncdf4)

# Assuming `ncout` is your existing NetCDF output connection
# Define the projection variable (matches Daymet's short data type)
corddef <- ncvar_def("lambert_conformal_conic", "", list(), prec = "short")
ncvar_add(ncout, corddef)

# Add CF-compliant projection attributes (use numeric values, not strings)
ncatt_put(ncout, "lambert_conformal_conic", "grid_mapping_name", "lambert_conformal_conic")
ncatt_put(ncout, "lambert_conformal_conic", "longitude_of_central_meridian", -100)
ncatt_put(ncout, "lambert_conformal_conic", "latitude_of_projection_origin", 42.5)
ncatt_put(ncout, "lambert_conformal_conic", "false_easting", 0)
ncatt_put(ncout, "lambert_conformal_conic", "false_northing", 0)
# Set both standard parallels as a vector to create duplicate attributes
ncatt_put(ncout, "lambert_conformal_conic", "standard_parallel", c(25, 60))
ncatt_put(ncout, "lambert_conformal_conic", "semi_major_axis", 6378137)
ncatt_put(ncout, "lambert_conformal_conic", "inverse_flattening", 298.257232666016)

2. Link Raster Variables to the Projection #

For every data variable in your NetCDF (e.g., tmax, prcp), add a grid_mapping attribute pointing to your projection variable. This tells ArcMap which projection applies to each raster:

# Example: Link temperature maximum variable to the projection
ncatt_put(ncout, "tmax", "grid_mapping", "lambert_conformal_conic")
# Repeat for all your data variables
ncatt_put(ncout, "tmin", "grid_mapping", "lambert_conformal_conic")
ncatt_put(ncout, "prcp", "grid_mapping", "lambert_conformal_conic")

3. Fix the time_bnds Dimension #

Recreate time_bnds with the correct 2D structure (nv for interval bounds, time for time steps) matching Daymet:

# Define the nv dimension (length 2 for start/end of time intervals)
nv_dim <- ncdim_def("nv", "", 1:2, unlim = FALSE)
# Get your existing time dimension (assuming it's named "time")
time_dim <- ncout$dim$time
# Define the time_bnds variable
time_bnds_def <- ncvar_def("time_bnds", "days since 1980-01-01 00:00:00 UTC", list(nv_dim, time_dim), prec = "float")
ncvar_add(ncout, time_bnds_def)
# Populate with your time interval data (e.g., start/end dates for each time step)
ncvar_put(ncout, time_bnds_def, your_time_bnds_data)

4. Verify Metadata #

After writing and closing the file, check that the metadata matches Daymet’s structure:

nc_close(ncout)
# Open the new file to validate
nc_check <- nc_open("your_custom_daymet_file.nc")
print(nc_check)
nc_close(nc_check)

Confirm:

• Duplicate standard_parallel entries (25 and 60)
• Each data variable has grid_mapping = "lambert_conformal_conic"
• time_bnds has dimensions nv, time

Testing in ArcMap #

Once you’ve generated the corrected file:

1. Use the Create NetCDF Raster Layer tool
2. Select your target data variable (e.g., tmax)
3. Set X Dimension to x, Y Dimension to y
   ArcMap should now automatically recognize the Lambert projection and render the raster without distortion or unresponsiveness.

内容的提问来源于stack exchange，提问作者Cooper Springer