Skip to content

Latest commit

 

History

History
45 lines (34 loc) · 2.25 KB

README.md

File metadata and controls

45 lines (34 loc) · 2.25 KB
Raw

ioda-plots

Tools to perform spatial and temporal binning of observations in JEDI IODA files, manipulate those binned statistics, and then plot those diagnostics.

The goal of this package is to be able to very quickly produce useful diagnostics with minimal or no configuration required, but at the same time allow for optional customization of plotting through yaml configuration files.

Installation

Assuming you have python3 already installed, ioda-plots can be installed by running

pip install --user .

Alternatively, if you don't feel like doing that, you can simply point to the source directory by setting the following (and noting that you'll have to run iodaplots.py instead of iodaplots):

PATH=<ioda-plots-path>/src/iodaplots/bin:$PATH
PYTHONPATH=<ioda-plots-path>/src:$PYTHONPATH

Use

(NOTE: Running iodaplots --help will provide documentation of the commands available, and running iodaplots <command> --help will provide further documentation about those commands.)

Using ioda-plots consists of three steps:

  1. Spatial binning - use iodaplots bin to take input IODA files and spatially bin them. Unless specific configuration is given in a yaml file, this will by default be done on a regular latlon grid, and globally, for observation value, O-B, O-A, and A-B statistics. This step should usually be done once separately for each cycle of a DA experiment. (this is the slow part, the rest of the steps are a lot faster.)

  2. Temporal merging/concatenation (optional) - use iodaplots merge to combine output from the previous step, as if you had run iodaplots bin on the entire length of the experiment. Or, use iodaplots cat to combine the output from the previous step, but instead by adding a new datetime dimension to the binning. cat will then allow for generation of timeseries plots.

  3. Plotting - use iodaplots plot on the output of the previous step to generate a whole bunch of plots. By default every metric that could be plotted is done, which is probably way more plots than you want. A yaml configuration can select which plots are generate, and allows for fine tuning of those plots.

(More detailed documentation and examples will be added later!)