"Variances estimation using ECMWF forecasts and NMC method"

In this repo we include two Jupyter notebooks for now:

• get_mars is an example of retrieving data from the ECMWF's MARS archive.
• error_cov is an example on how to estimate the variance of the error in the forecast model using the NMC method.

Purpose of the repo

Produce usable mean and covariances for wind retrievals using Jupyter Notebooks and modern tools such as:

• xarray
• Dask
• Pandas
• Cartopy
• Etc.

We want to propose a methodology of work were development and distribution is based on Jupyter Notebooks; this is a great environment to develope, prototype and deploy solutions for geoscience at scale. It is also very well suited to "connect" with Machine Learning tooling, such as Sci-kit Learn, TensorFlow, etc.

Documentation

Every Notebook in this repo is documented via Markdown.