Apache Beam pipelines to make weather data accessible and useful.
This project contributes a series of command-line tools to make common data engineering tasks easier for researchers in climate and weather. These solutions were born out of the need to improve repeated work performed by research teams across Alphabet.
The first tool created was the weather downloader (
weather-dl). This makes it easier to ingest data from the European
Center for Medium Range Forecasts (ECMWF).
weather-dl enables users to describe very specifically what data they'd
like to ingest from ECMWF's catalogs. It also offers them control over how to parallelize requests, empowering users to
retrieve data efficiently. Downloads are driven from a
configuration file, which can be reviewed (and version-controlled) independently of pipeline or
We also provide two additional tools to aid climate and weather researchers: the weather mover (
weather-mv) and the
weather splitter (
weather-sp). These CLIs are still in their alpha stages of development. Yet, they have been used for
production workflows for several partner teams.
We created the weather mover (
weather-mv) to load geospatial data from cloud buckets
into Google BigQuery. This enables rapid exploratory analysis and visualization of
weather data: From BigQuery, scientists can load arbitrary climate data fields into a Pandas or XArray dataframe via a
simple SQL query.
The weather splitter (
weather-sp) helps normalize how archival weather data is stored in cloud buckets:
Whether you're trying to merge two datasets with overlapping variables — or, you simply need
to open Grib data from XArray, it's really useful to split datasets into
their component variables.
It's recommended that you create a local python environment (with Anaconda). Otherwise, these tools can be installed with pip:
pip install google-weather-tools
From here, you can use the
weather-* tools from your python environment. Currently, the following tools are available:
weather-dl(beta) – Download weather data (namely, from ECMWF's API).
weather-mv(alpha) – Load weather data into BigQuery.
weather-sp(alpha) – Split weather data by variable.
Together, let's download Era 5 pressure level data and ingest it into Google BigQuery.
- Acquire and install a license from ECMWF's Copernicus (CDS) API.
- Create an empty BigQuery Dataset. This can be done in
or via the
bqCLI. For example:
bq mk --project_id=$PROJECT $DATASET_ID
weather-dlto acquire the Era 5 pressure level data.
For simplicity, let's run everything on your local machine. For the downloader, this means we'll use the
weather-dl configs/era5_example_config_local_run.cfg --local-run
Recommendation: Pass the
-d, --dry-runflag to any of these commands to preview effects.
(optional) Split your downloaded dataset up by variable with
weather-sp --input-pattern "./local_run/era5-*.nc" --output-dir "split_data"
weather-spdocs for more.
weather-mvto upload this data to Google BigQuery.
weather-mv --uris "./local_run/**.nc" \ # or --uris "./split_data/**.nc" \ --output_table "$PROJECT.$DATASET_ID.$TABLE_ID" \ --temp_location "gs://$BUCKET/tmp" \ # Needed for batch writes to BigQuery --direct_num_workers 2
See these docs for more about this tool.
That's it! Soon, you'll have your weather data ready for analysis in BigQuery.
Note: The exact interfaces for these CLIs are subject to change. For example, we plan to make the CLIs have more uniform arguments (#21).
The weather tools are under active development, and contributions are welcome! Please check out our guide to get started.
This is not an official Google product.
Copyright 2021 Google LLC Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.