climetlab-maelstrom-nogwd 0.2.0

maelstrom-nogwd

A CliMetLab (https://climetlab.readthedocs.io) dataset plugin for the dataset maelstrom-nogwd.

Features

In this README is a description of how to get the maelstrom-nogwd.

Dataset description

Contains the input/output dataset for learning non-orographic gravity wave drag, as described in https://arxiv.org/abs/2101.08195

Data is grouped by forecast start date.

Data has been preprocessed into inputs "x" and outputs "y". "x" contains vertical profiles of winds & temperature plus surface values of pressure and geopotential. "y" contains the the wind increments due to parametrised non-orographic gravity wave drag. The machine learning task is to predict y given x. Unlike many ML tasks within the field of weather and climate, this task can be predicted independently for each column of atmosphere.

Using climetlab to access the data

Data can be accessed either by forecast start-date or dataset type. With neither argument provided, the first file is loaded, corresponding to 2015-01-01 (the tier-1 dataset). Incorrect dates will be flagged. Other dataset types are "training", "validation" & "testing" corresponding to the date groups outlined in https://arxiv.org/abs/2101.08195

The climetlab python package allows easy access to the data with a few lines of code such as:

!pip install climetlab climetlab_maelstrom_nogwd
import climetlab as cml
cmlds = cml.load_dataset("maelstrom-nogwd", date='2015-01-01')
ds = cmlds.to_xarray()
#or
cmlds = cml.load_dataset("maelstrom-nogwd", dataset='training')
ds = cmlds.to_xarray()

See https://git.ecmwf.int/projects/MLFET/repos/maelstrom-nogwd/browse/notebooks/demo_nogwd.ipynb for a short tutorial.