Jetstream - A replication library for the paper [insert paper name here]

This library contain the replication code for the paper [paper name here]. Data adquisition, processing, and methods are included here. You can download either model or reanalysis data and calculate: t_ref and t_prime, and also generate some derivative products, like anomalies and Hovmuller plots to explore the movement of wind masees in the Northern Hemisphere

Data adquisition

The reanalysis_getter is a wrapper of the CDS API. The package is still in a bare bones stage, but it is able to translate multiple requests to the CDS API and retrieve the desired data, either waiting in the CDS user queue, or downloading the data directly (after waiting for the data to be processed).

API Credentials

Following the API configuration, a ~/.cdsapirc file with API credentials must be created before doing any request. See the API documentation for more details on how to get your user credentials.

Data processing

We run our tests on two types of products: reanalysis (ERA-5), and several CIMP6 global climate models model. To process both, the user can use jetstream.model.Analysis or jetstream.model.Model to either process reanalysis data or GCM data. Both classes are abstract classes inherited from jetstream.model.template and adapted to capture all the particularities from each data set.

This project is heavily reliant on both Dask and xarray, and uses the parallel powers from the latter to take big datasets and process the outlined methods in our paper. Hence, there are some hardware requirements that are needed to fully replicate our methods on a complete product, especially with daily data.

Both classes are able to take a dask.distributed.Client object from the environment and start calculation using the powers of embarassing distributed computing, onn both local and remote environments.

Configuration and How-to Install

You can use: python setup.py install --user to install the modules of this library. We recoment you to use a virtualenv to avoid conflicts with your local libraries, or use any of the Dask Docker images.

Use case

Download sub-daily (each 3 hours )surface temperature (2-meter temperature) between December 2007 and March 2008:

from datetime import datetime
from src.requester import request_wrapper
request_wrapper(file_name = None, 
                start_date = datetime(2007, 1 ,1),
                end_date = datetime(2008, 3, 1),
                variables_of_interest = [167.128] # See ERA-5 documentation for more on this
                subday_frequency = 3,
                pressure_levels = 'sfc'
                )

Data will be requested and a queue will start. Once data is processed remotely, the download process will start. By default, data will be stored in the cdsapi_requested_files directory.