How to preserve NetCDF dimensions with gdal.warp

Question

I have a netcdf file that is structured with precipitation grid values with dimensions (time, x, y)

When I perform gdal.warp (in python if that matters) I get a new file that has multiple bands that have dimensions (x, y), with the number of bands = len(time) of the previous file and time is not part of the new file.

Is there a way to preserve the dimensions, precipitation name, and time in the new file?

import gdal

warped = gdal.Warp(
                destNameOrDestDS=destNameOrDestDS,
                srcDSOrSrcDSTab=srcDSOrSrcDSTab,
                srcSRS=srcSRS,
                dstSRS=dstSRS,
                dstNodata=float(srcNodata),
                format="NETCDF",
                xRes=xRes,
                yRes=yRes,
                targetAlignedPixels=targetAlignedPixels,
            )

So my source netcdf file looks like:

• crs
• QPE
• time
• x
• y
• z

where crs is the coordinate system and QPE is precipitation and then my output looks like:

• Albers
• Band1
• Band2
• Band3
• x
• y

I would like to preserve the original structure so combine the Bands into a single 3 dimensional array with time as the 3rd dimension.

Answer 1

NetCDF driver is documented in https://gdal.org/drivers/raster/netcdf.html

Creation Issues

This driver supports creation of NetCDF file following the CF-1 convention. You may create set of 2D datasets. Each variable array is named Band1, Band2, … BandN. Each band will have metadata tied to it giving a short description of the data it contains.

Is seems that you must use some other tool than GDAL for creating a 3D dataset.

Answer 2

You need to transform your data in .tiff with gdal_translate, make the resize with gdalwarp, and then back to .nc with gdal_translate again.

Tips for increased speed: https://trac.osgeo.org/gdal/wiki/UserDocs/GdalWarp