SC2.25 | Find it, fetch it, plot it: a hands-on introduction of ECMWF open forecast data
Find it, fetch it, plot it: a hands-on introduction of ECMWF open forecast data
Convener: Alba Gomez SeguraECSECS | Co-conveners: Milana Vuckovic, Valters ZeizisECSECS, M. Crepulja

ECMWF makes a substantial part of its real-time forecast production freely available as open data under a CC-BY-4.0 licence, with no account, licence agreement or payment required. This includes the operational medium-range IFS ensemble forecast, as well as the AIFS data-driven forecasts, covering the whole globe and updated several times a day. The data are distributed in GRIB and BUFR format, which can be unfamiliar to newcomers. The aim of this course is to remove that barrier and get participants from "I have heard ECMWF data is open" to "I have a forecast field on my screen" within the session.

The first half of the course deals with discovery and access. We start in the browser on data.ecmwf.int, looking at what is actually published: which streams, resolutions, parameters and lead times exist, how the directory structure and file naming work, and how to read the index files that let you fetch only the fields you need rather than whole files. We then move to programmatic access with the ecmwf-opendata Python client, building requests step by step: selecting stream, type, date and time, restricting to specific parameters and levels, and handling the practical issues that might arise. Participants will run these retrievals themselves during the session.

The second half is about doing something with the data. Using earthkit and ecCodes software, we will open, explore, process and visualise the data. From there we work through a set of typical tasks: subsetting a region, extracting a time series at a point, computing simple ensemble statistics such as the mean and spread, and producing maps with the earthkit-plots library.

The course is fully hands-on and runs from Jupyter notebooks that participants can take away and adapt, all materials will be publicly available. Basic familiarity with Python is assumed, but no prior experience with GRIB or BUFR data or ECMWF services is needed. It is open to anyone curious about weather data: no background in meteorology or forecasting is needed. Participants should bring a laptop with them if they want to follow the practicals.