Basic Examples

Save a dataset into a compressed file

Once enstools-compression has been installed, enstools.io.write accepts the compression parameter that allows us to compress our files apply lossless or lossy methods.

An example with lossless compression:

write(dataset, "output.nc", compression="lossless")

An example with lossy compression:

write(dataset, "output.nc", compression="lossy,zfp,rate,3.2")

Compress an existing file

Using the Command Line Interface it is very easy to compress existing files as well:

An example with lossless compression:

>>> enstools-compression compress "input.nc" -o "output.nc" --compression "lossless"

An example with lossy compression:

>>> enstools-compression compress "input.nc" -o "output.nc" --compression "lossy,sz,rel,1e-5"

Analyze a dataset to find compression specifications

To find compression specifications, an automatic search function enstools.compression.analyze_dataset can be used to find which compression parameters yield the highest compression ratio while keeping certain quality constrains:

specifications, metrics = analyze_dataset(dataset)

By default the constrains used are

>>> constrains="correlation_I:5,ssim_I:2"

This stands for 5 9s of correlation (0.99999) and 2 9s of structural similarity (0.99).

specifications, metrics = analyze_dataset(dataset, constrains="correlation:0.99999,ssim:0.99")

Analyze a file to find compression specifications using the command line interface

The same function enstools.compression.analyze_dataset can be used from the Command Line Interface.

>>> enstools-compression analyze "input.nc"

By default the constrains used are

>>> constrains="correlation_I:5,ssim_I:2"

This stands for 5 9s of correlation (0.99999) and 2 9s of structural similarity (0.99).

>>> enstools-compression analyze "input.nc" --constrains "correlation_I:5,ssim_I:2"

More examples:

Explore the Examples section to see more examples on all the other features!